About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
i
Nanoscience: Underlying Physical Concepts and Phenomena
Arthur M.Sackler COLLOQUIA
OF THE NATIONAL ACADEMY OF SCIENCES
National Academy of Sciences Washington, D.C.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
ii
Arthur M.Sackler, M.D. 1913–1987 Born in Brooklyn, New York, Arthur M.Sackler was educated in the arts, sciences, and humanities at New York University. These interests remained the focus of his life, as he became widely known as a scientist, art collector, and philanthropist, endowing institutions of learning and culture throughout the world.
He felt that his fundamental role was as a doctor, a vocation he decided upon at the age of four. After completing his internship and service as house physician at Lincoln Hospital in New York City, he became a resident in psychiatry at Creedmoor State Hospital. There, in the 1940s, he started research that resulted in more than 150 papers in neuroendocrinology, psychiatry, and experimental medicine. He considered his scientific research in the metabolic basis of schizophrenia his most significant contribution to science and served as editor of the Journal of Clinical and Experimental Psychobiology from 1950 to 1962. In 1960 he started publication of Medical Tribune, a weekly medical newspaper that reached over one million readers in 20 countries. He established the Laboratories for Therapeutic Research in 1938, a facility in New York for basic research that he directed until 1983. As a generous benefactor to the causes of medicine and basic science, Arthur Sackler built and contributed to a wide range of scientific institutions: the Sackler School of Medicine established in 1972 at Tel Aviv University, Tel Aviv, Israel; the Sackler Institute of Graduate Biomedical Science at New York University, founded in 1980; the Arthur M.Sackler Science Center dedicated in 1985 at Clark University, Worcester, Massachusetts; and the Sackler School of Graduate Biomedical Sciences, established in 1980, and the Arthur M.Sackler Center for Health Communications, established in 1986, both at Tufts University, Boston, Massachusetts. His pre-eminence in the art world is already legendary. According to his wife Jillian, one of his favorite relaxations was to visit museums and art galleries and pick out great pieces others had overlooked. His interest in art is reflected in his philanthropy; he endowed galleries at the Metropolitan Museum of Art and Princeton University, a museum at Harvard University, and the Arthur M.Sackler Gallery of Asian Art in Washington, DC. True to his oft-stated determination to create bridges between peoples, he offered to build a teaching museum in China, which Jillian made possible after his death, and in 1993 opened the Arthur M.Sackler Museum of Art and Archaeology at Peking University in Beijing. In a world that often sees science and art as two separate cultures, Arthur Sackler saw them as inextricably related. In a speech given at the State University of New York at Stony Brook, Some reflections on the arts, sciences and humanities, a year before his death, he observed: “Communication is, for me, the primum movens of all culture. In the arts…I find the emotional component most moving. In science, it is the intellectual content. Both are deeply interlinked in the humanities.” The Arthur M.Sackler Colloquia at the National Academy of Sciences pay tribute to this faith in communication as the prime mover of knowledge and culture.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
CONTENTS
iii
PNAS Proceedings of the National Academy of Sciences of the United States of America
Contents
Papers from the Arthur M.Sackler Colloquium of the National Academy of Sciences PERSPECTIVES Emulating biology: Building nanostructures from the bottom up Nadrian C.Seeman and Angela M.Belcher
6451
Quantum dot artificial solids: Understanding the static and dynamic role of size and packing disorder K.C.Beverly, J.L.Sample, J.F.Sampaio, F.Remacle, J.R.Heath, and R.D.Levine
6456
COLLOQUIUM PAPERS Segmented nanofibers of spider dragline silk: Atomic force microscopy and single-molecule force spectroscopy E.Oroudjev, J.Soares, S.Arcdiacono, J.B.Thompson, S.A.Fossey, and H.G.Hansma
6460
Molecular dynamics analysis of a buckyball-antibody complex William H.Noon, Yifei Kong, and Jianpeng Ma
6466
H3PW12O40-functionalized tip for scanning tunneling microscopy In K.Song, John R.Kitchin, and Mark A.Barteau
6471
Energetics of nanocrystalline TiO2 M.R.Ranade, A.Navrotsky, H.Z.Zhang, J.F.Banfield, S.H.Elder, A.Zaban, P.H.Borse, S.K.Kulkarni, G.S.Doran, and H.J.Whitfield
6476
Study of Nd3+, Pd2+, Pt4+, and Fe3+ dopant effect on photoreactivity of TiO2 nanoparticles S.I.Shah, W.Li, C.-P.Huang, O.Jung, and C.Ni
6482
Entropically driven self-assembly of multichannel rosette nanotubes Hicham Fenniri, Bo-Liang Deng, Alexander E.Ribbe, Klaas Hallenga, Jaby Jacob, and Pappannan Thiyagarajan
6487
Combining constitutive materials modeling with atomic force microscopy to understand the mechanical properties of living cells Mike McElfresh, Eveline Baesu, Rod Balhorn, James Belak, Michael J.Allen, and Robert E.Rudd
6493
Designing supramolecular porphyrin arrays that self-organize into nanoscale optical and magnetic materials Charles Michael Drain, James D.Batteas, George W.Flynn, Tatjana Milic, Ning Chi, Dalia G.Yablon, and Heather Sommers
6498
Nanoscale surface chemistry Theodore E.Madey, Kalman Pelhos, Qifei Wu, Robin Barnes, Ivan Ermanoski, Wenhua Chen, Jacek J.Kolodziej, and John E.Rowe
6503
Magnetic nanodots from atomic Fe: Can it be done? E.te Sligte, R.C.M.Bosch, B.Smeets, P.van der Straten, H.C.W.Beijerinck, and K.A.H.van Leeuwen
6509
Distributed response analysis of conductive behavior in single molecules Marc in het Panhuis, Robert W.Munn, Paul L.A.Popelier, Jonathan N.Coleman, Brian Foley, and Werner J.Blau
6514
Design of protein struts for self-assembling nanoconstructs Paul Hyman, Regina Valluzzi, and Edward Goldberg
8488
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
CONTENTS iv
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
EMULATING BIOLOGY: BUILDING NANOSTRUCTURES FROM THE BOTTOM UP
6451
Perspective Emulating biology: Building nanostructures from the bottom up
Nadrian C.Seeman†‡ and Angela M.Belcher§ of Chemistry, New York University, New York, NY 10003; and §Department of Chemistry, University of Texas, Austin, TX 78712 The biological approach to nanotechnology has produced self-assembled objects, arrays and devices; likewise, it has achieved the recognition of inorganic systems and the control of their growth. Can these approaches now be integrated to produce useful systems? We hear continually that nanoscience and nanotechnology are frontier areas. Everyone is aware that nanotechnology and nanoscience involve the construction and analysis of objects and devices that are very small on the macroscopic scale. Nevertheless, if the ultimate feature sizes of nanoscale objects are about a nanometer or so, we are talking about dimensions an order of magnitude larger than the scale exploited by chemists for over a century. Synthetic chemists have manipulated the constituents, bonding, and stereochemistry of vast numbers of molecules on the angstrom scale, and physical and analytical chemists have examined the properties of these molecules. So what is so special about the nanoscale? There are many answers to this question, possibly as many as there are people who call themselves nanoscientists or nanotechnologists. A particularly intriguing feature of the nanoscale is that this is the scale on which biological systems build their structural components, such as microtubules, microfilaments, and chromatin. The associations maintaining these and the associations of other cellular components seem relatively simple when examined by high-resolution structural methods, such as crystallography or NMR—shape complementarity, charge neutralization, hydrogen bonding, and hydrophobic interactions. A key property of biological nanostructures is molecular recognition, leading to self-assembly and the templating of atomic and molecular structures. For example, it is well known that two complementary strands of DNA will pair to form a double helix. DNA illustrates two features of self-assembly. The molecules have a strong affinity for each other and they form a predictable structure when they associate. Those who wish to create defined nanostructures would like to develop systems that emulate this behavior. Thus, rather than milling down from the macroscopic level, using tools of greater and greater precision (and probably cost), they would like to build nanoconstructs from the bottom up, starting with chemical systems. What are the advantages of building from the bottom up? Dense chemical variety is one advantage. Just as the surfaces of cellular components contain many features per unit area, a complex chemical surface can be used as a building block and, in principle, its orientation and position can be controlled. By contrast, top-down methods work on materials with little chemical diversity. A second advantage is the vastness of the chemical scale. Even a picomole of material is nearly 1012 copies. Thus, one can imagine producing complex components that form well defined structural motifs organized over large areas in two dimensions or volumes in three dimensions. †Department
Fig. 1. Formation of a 2D lattice from a junction with sticky ends. X and Y are sticky ends and X′ and Y′ are their complements. Four of the monomers on the left are complexed to yield the structure on the right. DNA ligase can close the gaps left in the complex, which can be extended by the addition of more monomers.
DNA NANOTECHNOLOGY To date, the most successful biomimetic component used for self-assembly has been DNA itself (1). Linear DNA double helices seem to be of limited utility, but one can design synthetic molecules that form stable branched structures, leading to greater structural complexity. Branched DNA molecules can be combined by “sticky-ended” cohesion (2), as shown in Fig. 1. In synthetic systems, sticky ends may be programmed with a large diversity; N-nucleotide sticky ends lead to 4N possible different sequences. Sticky ends of sufficient length cohere by base pairing alone but they can be ligated to covalency. Sticky ends form classic B-DNA when they cohere (3); thus, in addition to the affinity inherent in complementarity, sticky ends also lead to structural predictability. If the positions of the atoms on one component are known near the sticky end, the atoms of the other component are also known. This situation is usefully contrasted with, say, an antibody and its antigen. Although the antigen-combining site may be known, the orientation of the antigen within it cannot be predicted; it must be determined experimentally in each case. The key static aims of DNA nanotechnology are to use DNA as scaffolding to crystallize biological macromolecules artificially for crystallography (2) and to organize the components of nanoelectronics (4). The first, and likely the second, of these applications entail the assembly of DNA into periodic networks. Thus, the quadrilateral of Fig. 1 would be
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC. Abbreviations: DX, double crossover; 2D, two-dimensional. ‡To whom reprint requests should be addressed. E-mail:
[email protected].
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
EMULATING BIOLOGY: BUILDING NANOSTRUCTURES FROM THE BOTTOM UP
6452
most useful if extended to form a two-dimensional (2D) or three-dimensional lattice. The branched junctions shown in Fig. 1 are not rigid enough to use as building blocks for a lattice (5). This problem has been solved by combining two branched junctions to produce DNA doublecrossover (DX) molecules (6, 7), which consist of two double helices fused by strands that cross between them to tie them together.
Fig. 2. Arrays assembled from DX molecules, (a) A two-component array. Two DX molecules (A and B*) are illustrated schematically (a Top). The two helices are drawn as rectangles, and the complementary sticky ends are represented by geometrical shapes. A is a conventional DX molecule, but B* contains a DNA hairpin protruding from the plane. Below these molecules is an array that shows the two components fitting together to tile a plane, (b) A four-component array. The same conventions apply as in a. This array uses four tiles, A, B, C, and D*, where A, B, and C are conventional DX molecules and D* contains a hairpin. The stripes are separated by twice the distance seen in a.
Fig. 2 illustrates two different 2D arrays that have been produced by DX molecules (8). In Fig. 2a, the repeating unit is 2 DX units, and in Fig. 2b, it is 4 DX units. The B* unit in Fig. 2a and the D* unit in Fig. 2b have circles in their centers, representing another helix that is directed out of the plane. This extra helix can serve as a topographic marker for atomic force microscopy. The dimensions of each component are about 4×16 nm. Thus, the extra helices produce stripe-like features every 32 nm in the AB* array, and every 64 nm in the ABCD* array (8). Other DNA motifs have been used to produce nanoscale patterns in 2D. Although the individual branched junction is flexible, well structured DNA parallelograms can be prepared from four of them. DNA parallelograms have been used to produce mesh-works with tunable cavities (9). Likewise, one can prepare triple-crossover (TX) motifs, containing three fused double helices with coplanar axes. These molecules form 2D arrays that can include TX components rotated out of the plane (10). There are several ways to incorporate complexity in DNA arrays. One way is shown in Fig. 2b, where four different units comprise the asymmetric unit of the repeating structure; extension of this approach entails the expense of producing as many components as needed to produce a particular complex pattern in two or three dimensions. Winfree (11) suggested that it is possible to program sticky ends to produce algorithmic assemblies, much like the colored edges of Wang tiles; these tiles form a mosaic in which each tile edge abuts another with the same color. Appropriate sets of Wang tiles can assemble to perform computational operations and to define patterns with much greater complexity than their total number, thereby saving the expense of producing a large number of different tiles. Recently, the feasibility of this approach has been demonstrated in one dimension with triple-crossover molecules, in which a prototype cumulative exclusive OR calculation was executed (12), as shown in Fig. 3. The input and output Boolean values of each step of this calculation are represented by the sticky ends, thus this is a more stringent test of self-assembly than formation of a periodic lattice. The same sticky end on one side of the red answer tiles represents 0, regardless of whether it is on a correct or incorrect tile for a particular position. Despite overall good fidelity, some errors were detected in this experiment. In addition to static structures, DNA can be used to produce nanomechanical devices. An early device (Fig. 4) was based on the transition between right-handed B-DNA and left-handed Z-DNA (13). The system consists of two rigid DX molecules joined by a DNA helix containing a stretch that can undergo the B-Z transition. Yurke et al. (14) have reported a sequence-
Fig. 3. A cumulative XOR calculation. The XOR operation takes two Boolean inputs and produces a 0 if they are the same and a 1 if they are different. Shown in a are blue input tiles, Xi which represent 0 or 1, according to the presence of a particular restriction enzyme site. These have been assembled in a particular order in b. The red tiles in a contain the four Boolean possibilities as sticky ends on their lower helices. The input is connected to the output through the green C1 and C2 tiles. At the end of the self-assembly, one strand that runs through the entire system is ligated together, thereby connecting the input to the output. It is read by partial restriction, followed by a denaturing gel, much like a sequencing reaction.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
EMULATING BIOLOGY: BUILDING NANOSTRUCTURES FROM THE BOTTOM UP
6453
dependent DNA device that uses branch migration to remove strands from a tweezer-like construct, allowing it to undergo an opening transition. This approach has been used to produce a robust sequence-dependent rotary device (H.Yan, X.Zhang, Z. Shen, and N.C.S., unpublished data). The development of sequence-dependent DNA nanomechanical devices suggests a future for DNA as a controlling element in nanorobotics: N different 2-state devices incorporated in a nanorobotic superstructure could lead to 2N distinct structural states that could be programmed serially to execute mechanical tasks.
Fig. 4. A DNA nanomechanical device based on the B-Z transition. The device consists of two DX molecules connected by a helix (yellow section) that can undergo the B-Z transition. When this occurs, the bottom domain of the right DX molecule swings from the bottom to the top through a rotary motion.
The control over DNA systems seems to be relatively robust. However, to what ends can it be used? The physical properties of DNA are important for understanding biological systems, but their utility in nanoelectronic devices is unproved. To get the maximum use from the organizational capabilities of DNA, it will be necessary to combine DNA with other nanoscale systems, particularly inorganic systems, whose physical properties lend themselves to direct applications. These materials include inorganic nanocrystals and carbon nanotubes, which represent the most exciting potential species to organize in two or three dimensions. The large sizes and abundant functional groups on DNA tiles suggest that multiple functionalities could be attached to a single tile. THE BIOLOGICAL-INORGANIC INTERFACE As noted previously, the use of biological materials offers many advantages over traditional processing methods to construct the next generation of miniaturized electronics devices, particularly including spatial control on the nanometer scale, parallel self-assembly of multiple electronic components on a single device, and correctability. The critical factors in developing a bio-directed self-assembly approach are identifying the appropriate compatibilities and combinations of biological-inorganic materials, synthesis of the appropriate building blocks, and understanding and controlling building block self-assembly processes. In natural biological systems, macromolecules exert exceptional control over inorganic nucleation, phase stabilization, assembly, and pattern formation (15, 16). Biological systems assemble nanoscale building blocks into complex and functionally sophisticated structures with high perfection, controlled size, and compositional uniformity. These materials are typically soft and consist of a surprisingly simple collection of molecular building blocks (i.e., lipids, peptides, and nucleic acids) arranged in complex architectures. For example, proteins from bones, shells, diatoms, and magnetic bacteria can spatially and temporally nucleate inorganic structures from the nanoscopic to the macroscopic scale. In addition, selectivity and recognition at the molecular scale is a critical feature of living systems. Among the best known examples are antibody-antigen interactions. Unlike the semiconductor industry, which relies on serial lithographic processing to construct the smallest features on an integrated circuit, organisms execute their architectural blueprints by using mostly noncovalent forces acting simultaneously and selectively on many molecular components. The exquisite selectivity of complementary biological molecules offers a possible avenue to control the formation of complex structures based on inorganic building blocks such as metal or semiconductor nanoparticles. DNA oligomer-nanocrystal complexes, for example, have been examined as building blocks for more complex two- and three-dimensional structures (17, 18). Nanocrystal-labeled proteins have also been used to label biomolecular substrates with increased sensitivity (19, 20). Self-assembled monolayers have been used to template nanocrystal organization and in some cases, covalently bind semiconductor nanocrystals to metal surfaces (21). “Nanonetworks” have been formed with gold nanoclusters by, using dithiol connectors (22), and with iron oxide, using biotin-streptavidin connectors (23). Specific hydrogen-bonding-directed aggregation between nanocrystals, using alkanethiol-modified DNA base pairs, uracil, and 2,6 diaminopyridine, has also been demonstrated (24). Only very modest binding specificity between the biological molecule and
Fig. 5. Peptide selection for electronic materials. The 1.9×109 random peptide sequences are exposed to the different crystal substrates; nonspecific peptide interactions are removed with extensive washes. The phage that bind are eluted by lowering the pH and disrupting the surface interaction. The eluted phage are amplified by infecting the E. coli ER2537 host, producing enriched populations of phage, displaying peptides that interact with the specific crystal substrate. The amplified phage are isolated, titered, and reexposed to a freshly prepared substrate surface, thereby enriching the phage population with substrate-specific binding phage. This procedure is repeated three to five times to select the phage with the tightest and most specific binding. The DNA of phage that show specificity is sequenced to determine the peptidebinding sequence.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
EMULATING BIOLOGY: BUILDING NANOSTRUCTURES FROM THE BOTTOM UP
6454
the inorganic substrate can be achieved through these grafting chemistries. These approaches show potential for the control and placement of nanoparticles but they have not exploited the atomic composition and plane-specific recognition that a biomolecule can exhibit for an inorganic phase or the nanostructural control and regularity that biomolecules typically impose on crystal phases and crystallographic orientations.
Fig. 6. Peptide selectivity for patterned GaAs. Fluorescently labeled phage displaying a peptide specifically selected to bind to GaAs bind only to the patterned GaAs nested square pattern on a wafer. The red lines (1 µm across) correspond to GaAs and the black spaces (4 µm across) are SiO2. This peptide-specific binding could also be used to deliver nanocrystals to specific locations.
In principle, biological molecules can be used to control the assembly of inorganic nanostructures and hybrid inorganic/ organic structures while directing them to self-assemble in the desired manner. Thus, the biological molecules and an unlimited number of different types of nanocrystal building blocks can be mixed in the “pot” and then triggered to self-assemble into their superstructures. To develop the interactions of biological molecules for inorganic materials, biological selection has been used with the goal of synthesizing technologically important inorganic materials to serve as building blocks for new materials. Because nature has not had the opportunity to produce biomolecular interactions with some of the desired materials, A.M.B. and coworkers (25) used phage display to select peptide sequences, and Brown (26) used repeating polypeptides displayed on the surface of the bacterium Escherichia coli to bind selectively to metal particles. To identify the appropriate compatibilities and combinations of biological-inorganic materials, a combinatorial library of genetically engineered M13 bacteriophage was used to select peptides rapidly that could not only recognize but also control the growth of specific inorganic materials (Fig. 5). The phage display library is based on a combinatorial library of random peptides of a given length (e.g., 7- or 12-mers) that are fused to the pIII minor coat protein of the filamentous coliphage M13 (New England Biolabs, www.neb.com). Five copies of the fused random pIII coat protein are located on one end of the phage particle and account for 10–16 nm of the 1 µm viral particle. The library used for the selection consists of 1.9×109 random sequences. This peptide combinatorial approach was used to identify proteins that specifically bind to inorganic nanoparticles such as semiconductor nanocrystals. Hence, this is a promising approach for selecting peptides that can recognize specific inorganic crystals and crystallographic orientations of nonbiological origin, including InP, GaAs, and Si (ref. 25; Fig. 6), and can control II-VI (ZnS, CdS, CdSe, ZnSe, and PbS; A.M.B., unpublished data) semiconductor nanocrystal size, crystal structure, shape, and optical properties. ISSUES AND PROSPECTS Biomimetic nanotechnology holds great promise as a vehicle to achieve progress in the areas of macromolecular crystallography, nanoelectronics, and nanorobotics. The key issues in the area currently include the following: (i) Can control be obtained over the sizes of crystalline or aperiodic arrays? Can 2D crystals be made larger than their current dimensions on the order of a micrometer or two? (ii) Can crystals be produced that are nearly error-free? Applications in nanoelectronics will require high degrees of perfection, although error-tolerant techniques (27) may alleviate this problem. (iii) Can nanomechanical devices transmit forces in the same way that larger devices do in the macroscopic world? (iv) Can the systems described here and in the related experiments pioneered by Alivisatos et al. (17), Mirkin et al. (18), and Mallouk and coworkers (28) be used to interface the architectural prowess of “wet” biomimetic nanotechnology with the functional potency of the “dry” nanotechnology of nanotubes (29) and nanoelectronic components (30, 31)? The successes enjoyed by Whitesides and his coworkers (32) working on a slightly larger scale may be taken as an indication that bottomup assembly and organization can be extended conveniently down through the nanoscale. Current efforts to extend DNA nanotechnology from two- to three-dimensional (3D) arrays can be expected to produce true 3D integration of nanoelectronic components, although this success, when it comes, will lead to other problems of addressing and heat dissipation. Biomolecular recognition and peptide evolution will be used to develop molecular tool kits for the design and synthesis of inorganic nanocrystals with the potential to offer even greater flexibility in materials synthesis and assembly than with current synthetic routes. Biomimetic nanotechnology is just beginning to bloom—the full inflorescence promises to be spectacular. This work has been supported by the National Institute of General Medical Sciences Grant GM-29554 (to N.C.S.), the Office of Naval Research Grant N00014–98–1–0093 (to N.C.S.), the Defense Advanced Research Planning Agency/National Science Foundation Grant CCR-97–25021 (to N.C.S.), National Science Foundation Grants CTS-9986512, EIA-0086015, and CTS-0103002 (to N.C.S.), the Army Research Office Grant DADD19–99– 0155 (to A.M.B.), and by the Welch Foundation (A.M.B.). 1. Seeman, N.C. (1999) Trends Biotechnol. 17, 437–443. 2. Seeman, N.C. (1982) J. Theor. Biol. 99, 237–247. 3. Qiu, H., Dewan, J.C. & Seeman, N.C. (1997) J. Mol. Biol. 267, 881–898. 4. Robinson, B.H. & Seeman, N.C. (1987) Protein Eng. 1, 295–300. 5. Petrillo, M.L., Newton, C.J., Cunningham, R.P., Ma, R.-I., Kallenbach, N.R. & Seeman, N.C. (1988) Biopolymers 27, 1337–1352. 6. Fu, T.-J. & Seeman, N.C. (1993) Biochemistry 32, 3211–3220. 7. Li., X., Yang, X., Qi, J. & Seeman, N.C. (1996) J. Am. Chem. Soc. 118, 6131–6140. 8. Winfree, E., Liu, F., Wenzler, L.A. & Seeman, N.C. (1998) Nature (London) 394, 539–544. 9. Mao, C., Sun, W. & Seeman, N.C. (1999) J. Am. Chem. Soc. 121, 5437–5443. 10. LaBean, T.H., Yan, H., Kopatsch, J., Liu, F., Winfree, E., Reif, J.H. & Seeman, N.C. (2000) J. Am. Chem. Soc. 122, 1848–1860. 11. Winfree E. (1996) in DNA Based Computing, eds. Lipton, E.J. & Baum, E.B. (Am. Math. Soc., Providence, RI), pp. 199–219. 12. Mao, C., LaBean, T.H., Reif, J.H. & Seeman, N.C. (2000) Nature (London) 407, 493–496. 13. Mao, C., Sun, W., Shen, Z. & Seeman, N.C. (1999) Nature (London) 397, 144–146. 14. Yurke, B., Turberfield, A.J., Mills, A.P., Jr., Simmel, F.C. & Neumann, J.L. (2000) Nature (London) 406, 605–608. 15. Belcher, A.M., Wu, X.H., Christensen, R.J., Hansma, P.K., Stucky, G.D. & Morse, D.E. (1996) Nature (London) 381, 56–58. 16. Falini, G., Albeck, S., Weiner, S. & Addadi, L. (1996) Science 271, 67–69.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
EMULATING BIOLOGY: BUILDING NANOSTRUCTURES FROM THE BOTTOM UP
6455
17. Alivisatos, A.P., Johnsson, K.P., Peng, X., Wilson, T.E., Loweth, C.J., Bruchez, M.P. & Schultz, P.G. (1996) Nature (London) 382, 609–611. 18. Mirkin, C.A., Letsinger, R.L., Mucic, R.C. & Stofoff, J.J. (1996) Nature (London) 382, 607. 19. Chan, W.C.W. & Nie, S. (1998) Science 281, 2016–2018. 20. Bruchez, M., Moronne, M., Gin, P., Weiss, S. & Alivisatos A.P. (1998) Science 281, 2013–2016. 21. Colvin, V.L., Goldstein, A.N. & Alivisatos, A.P. (1992) J. Am. Chem. Soc. 114, 5221–5230. 22. Baum, T., Bethell, D., Brust, M. & Schiffrin, D.J. (1999) Langmuir 15, 866–871. 23. Li, M., Wong, K.K.W. & Mann, S. (1999) Chem. Mater. 11, 23–26. 24. Cusack, L., Rao, S.N., Wenger, J. & Fitzmaurice, J.D. (1997) Chem. Mat. 9, 624–631. 25. Whaley, S.R., English, D.S., Hu, E.L., Barbara, P.R. & Belcher, A.M. (2000) Nature (London) 405, 665–668. 26. Brown, S. (1992) Proc. Nat. Acad. Sci. USA 89, 8651–8655. 27. Heath, J.R., Keukes, P.J., Snider, G. & Silliams, R. (1998) Science 280, 1716–1721. 28. Mbindyo, J.K.N., Reiss, B.D., Martin, B.R., Keating, C.D., Natan, M.J. & Mallouk, T.E. (2001) Adv. Mater. 13, 249–254. 29. Colbert, D.T., Zhang, J., McClure, S.M., Nikolaev, P., Chen, Z., Hafner, J.H., Owens, D.W., Kotula, P.G., Carter, C.B., Weaver, J.H., et al. (1994) Science 266, 1218–1222. 30. Reed, M.A., Zhou, C., Muller, C.J., Burgin, T.P. & Tour, J.M. (1997) Science 278, 252–254. 31. Collier, C.P., Wong, E.W., Belohradsky, M., Raymo, F.M., Stoddart, J.F., Keukes, P.J., Williams, R.S. & Heath, J.R. (1999) Science 285, 391–394. 32. Trau, M., Yao, N., Kim E., Xia, Y., Whitesides, G.M. & Aksay, I.A. (1997) Nature (London) 390, 674–676.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
QUANTUM DOT ARTIFICIAL SOLIDS: UNDERSTANDING THE STATIC AND DYNAMIC ROLE OF SIZE AND PACKING DISORDER
6456
Perspective Quantum dot artificial solids: Understanding the static and dynamic role of size and packing disorder K.C.Beverly*, J.L.Sample*, J.F.Sampaio†, F.Remacle‡, J.R.Heath*, and R.D.Levine§¶ *Department of Chemistry and Biochemistry, University of California, Los Angeles, CA 90095; †Depto de Fisica, Universidade Federal de Minas Gerais, CEP. 30123–970, Belo Horizonte, M.G., Brazil; ‡Département de Chimie, B6c, Université de Liège, B 4000 Liège, Belgium; and §The Fritz Haber Research Center for Molecular Dynamics, The Hebrew University, Jerusalem 91904, Israel This perspective examines quantum dot (QD) superlattices as model systems for achieving a general understanding of the electronic structure of solids and devices built from nanoscale components. QD arrays are artificial two-dimensional solids, with novel optical and electric properties, which can be experimentally tuned. The control of the properties is primarily by means of the selection of the composition and size of the individual QDs and secondly, through their packing. The freedom of the architectural design is constrained by nature insisting on diversity. Even the best synthesis and separation methods do not yield dots of exactly the same size nor is the packing in the self-assembled array perfectly regular. A series of experiments, using both spectroscopic and electrical probes, has characterized the effects of disorder for arrays of metallic dots. We review these results and the corresponding theory. In particular, we discuss temperature-dependent transport experiments as the next step in the characterization of these arrays. Over the past few years a number of new chemical (1) and physical (2, 3) techniques have emerged for manipulating and tailoring the electrical properties of solids. Certain of the chemical techniques have been enabled by the development of synthetic methods for the preparation of narrow-size distributions of organically passivated metal (4) and semiconductor quantum dots (QDs) (5). For our purposes, we use metal QDs as artificial atoms: We use them as nanoscale building blocks for constructing extended two-dimensional solids, or QD superlattices. The beauty of QD solids is 2-fold: First, many physical parameters such as particle size and size distribution, disorder, interparticle coupling, etc. represent knobs that can be individually controlled. Variation of any of these parameters translates directly into either subtle or dramatic changes in the collective electronic response of the superlattice. Second, a one-electron model of the coupled QDs, augmented by the concept of the Coulomb charging energy (6), turns out to capture much of the physics of the system and thus enables experiment and theory to progress hand-in-hand. It is this versatility in both experiment and theory that can potentially turn these QD superlattices into model systems for achieving a general understanding of the electronic structure of solids. We have not completely established that model system yet, but it is our goal to get there, and the point of this article is to assess where we want to go and to serve as a progress report toward that goal. For a chemist, it is convenient to think of the highest-lying electrons on each dot as the valence electrons of an atom. Two interacting identical dots are thus similar to two equivalent, covalently bonded atoms. Within a QD superlattice, electron transfer between neighboring QDs is a low-energy process. Technically, this means that QDs have low charging energies (7, 8). Metallic QDs of 2–10 nm diameter (102–104 atoms) have size-dependent charging energies (I) in the range of 0.5 to 0.1 eV. By contrast, actual atoms have charging energies of at least several eVs. The idyllic picture of identical “atoms” within a perfect lattice fails to account for what are effectively intrinsic properties of chemically synthesized QDs. For example, any two adjacent dots are unlikely to be truly identical. Even the best chemical preparations (9, 10) for producing size-tunable and narrow particle size distributions still produce particles with finite distribution widths, and those widths translate directly into a distribution of site energies. Thus, in general, any two-particle interaction will have some ionic character, as one dot is always more electronegative than the other. Finally, the superlattice itself is characterized by some amount of packing disorder, which contributes to local variations in interparticle coupling strengths, among other things. When the parameters that govern the electronic structure of QD solids are varied, those solids can undergo macroscopic, collective changes that can be observed by eye (Fig. 1). The phenomena highlighted in Fig. 1 is a transition to an electronically delocalized state that is triggered by an acoustic wave, and this is a type of quantum phase transition. We speak of a quantum phase transition when the nature of the quantum mechanical state of the system changes. The change can be discontinuous, analogous to a first-order phase transition, or it can be a continuous change in the energy in which case one might consider the change to be an isomerization. Fig. 2 is a phase diagram showing a subset of such possible transitions, as represented by a plot of interparticle (exchange) coupling vs. disorder. Other representations, such as disorder vs. temperature, or disorder vs. electric field strength, or any other combination of size-distribution disorder, packing disorder, exchange coupling, particle size, temperature, external field strength, etc. can generate equally rich phase diagrams. All of these parameters are under experimental control and may be quantum mechanically modeled. It is exactly this versatility that makes these systems such interesting models for study. Disorder has become a parameter that we have been increasingly able to quantify, and so we emphasize the role of disorder by choosing it as the abscissa in the phase diagram. Qualitatively, disorder can arise from size distribution widths and/or packing defects. It is effectively a measurement of the dissimilarity of adjacent dots, and so it acts against a collective behavior. Size fluctuations are obviously coupled to packing disorder but, for a given distribution, one can independently vary the packing disorder via compression of an initially self-assembled lattice. A sufficiently narrow size distribution of QDs will spontaneously crystallize into a well-ordered hexagonal phase (11). The ordinate in the phase diagram, β, represents the strength of the exchange coupling between adjacent dots. β is large or
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC. Abbreviations: QD, quantum dot; SP, surface potential. ¶To whom reprint requests should be addressed. E-mail:
[email protected].
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
QUANTUM DOT ARTIFICIAL SOLIDS: UNDERSTANDING THE STATIC AND DYNAMIC ROLE OF SIZE AND PACKING DISORDER
6457
small in comparison to two other energies. One is the charging energy I, which represents the energetic cost of transferring an electron from one dot to another one that already has its complement of electrons. The exchange coupling β that moves an electron from one dot to another needs to overcome this cost if it is to be effective. The other relevant energy, ∆α, is the range of fluctuations in the energy, α, of the exchanged electron. Because the valence electron is confined to the dot, ∆α/α is a measure of disorder that arises primarily from the size distribution. β is large if it can bridge the local size fluctuations, β>∆α. By compressing a QD monolayer one can increase β. Because the exchange coupling relies on the overlap of the tails of wave functions that are centered on adjacent dots, β scales exponentially with the distance D between the dots. Thus, through compression, β can be tuned over a wide range. The transition to a collective behavior, when β is sufficiently large, is first order. Our initial observation of an optical signature of this transition (12) could be well reproduced by quantum mechanical computations (13) that allowed for size fluctuations. Since that time we have observed other facets of the phase diagram but much more remains to be done.
Fig. 1. Acoustic wave induced transition to a delocalized state. The pressure wave compresses the array and thereby increases the exchange coupling of the QDs, as discussed in connection with Eq. 1. See ref. 12 for the first experimental report and refs. 13 and 28 for more on the theory. Shown is a videocaptured image of a Langmuir monolayer of 4-nm diameter, pentanethiol passivated Ag QDs that has been compressed, using the mobile barriers of the Langmuir trough, to a point that is just short of the transition from an insulator to a metal. A second Teflon barrier, oriented perpendicular to the principal barriers, was connected mechanically to a speaker, and the speaker was electronically connected to a function generator. A 1-V amplitude, ≈100-Hz square wave function was applied to the speaker and transduced as a mechanical vibration in the second Teflon barrier. The mechanical oscillations corresponded a moving wave of high- and low-pressure oscillations through the monolayer. The silvery components of the monolayer are metallic, whereas the darker, reddish regions are insulating. Other more quantitative measurements have indicated that the metallic regions are, in fact, characterized by a free-electron response in the complex dielectric function—i.e., by a negative valued component of the real part of the dielectric function. This ability to dynamically pattern a macroscopic film as an insulator or a metal is, to our knowledge, unique to these materials.
The quantitative results on the role of compression for an array of Ag QD could be fitted by using an exchange coupling of the functional form [1] R is the radius of the dot, D is the lattice constant, and L is the dimensionless range parameter. For small (2R=35 Å) Ag dots, 1/2L=5.5 and α β0=0.5 eV. The range of strong coupling is determined by the value D0/2R=1.2 and beyond this point β decreases exponentially. This very same coupling (14) was able to reproduce the experimentally measured effect of lattice compression on the frequency-dependent complex dielectric function (15). In the study of the conductivity (16) reported (see Fig. 5), the dots have twice the radius. Because the exchange coupling depends on the overlap of the wave functions of adjacent dots and so declines exponentially as exp(−κD) where the length scale 1/κ should remain the same, we use 1/2L=11. Also, to keep the strength of the coupling at contact, D/2R=1, the same, we use D0/ 2R=1.1. Some disorder is always present, and, even at room temperature, disorder may mask the Mott insulator to metal transition (6) that occurs when the exchange coupling β exceeds the charging energy I. It is easy to assess when this will be the case. Size fluctuations with the resulting variations in the energies of the dots will affect the energy of the ionic bands. When disorder is sufficiently high such that ∆α>I, the lowest ionic band can overlap the covalent band. This happens when it is energetically advantageous to move an electron from a dot of higher energy to one with a lower α. The cost is I and the gain is ∆α. Indeed, it is even possible for the ionic band to be lowest in energy (17, 18). For highly disordered lattices, the effects of the charging energy therefore will be masked and the key reduced variable is β/∆α.
Fig. 2. A quantum mechanical phase diagram for an array of QDs with size and packing fluctuations. The full phase diagram is actually much richer than what is shown as there are various other ways to tune the relative energies of the system, as discussed in the text.
Even within the limited context of the phase diagram of Fig. 2, the scope of possible transitions is not exhausted by the first-order transition to a collective, delocalized state or the second-order transition between the covalent and ionic states. In the weak coupling regime where β/∆α≤1, the theory suggests that there is another second-order transition, from a localized to a domain localized state (19). In the localized state each electron is defacto confined to a single lattice site; the individual dots are not effectively coupled. It is a strictly insulating state. The domain-localized state has the electrons delocalized over finite QD domains that are smaller than the overall dimensions of the film. This state has recently (19) been imaged by introducing yet another variable, an electric field in the plane of the lattice, and measuring the surface potential (SP) in the x-y plan of the film (20). The domain localized state is not a packing configuration. Fig. 3 shows not only the contours of the SP but also a co-collected topographic image of the QD array, and the images are uncorrelated. Thus, the crystallographic structure of the array can be quite different from the electronic structure. Room temperature SP images, collected as the interdot separation distance (D/2R) is decreased, exhibit a transition to a collective behavior. This was seen as an Ohmic-like monotonic drop in the SP across the array. However, consistent with the theory, this delocalization transition was observed only (20) in arrays characterized by a limited amount of disorder. Highly disordered arrays remained domain localized, even at high compression. Theoretically, we expect that the application of a voltage gradient in the plane of the array will counteract the effects of size disorder. In the presence of an external potential the site energy of the ith dot has the form
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
QUANTUM DOT ARTIFICIAL SOLIDS: UNDERSTANDING THE STATIC AND DYNAMIC ROLE OF SIZE AND PACKING DISORDER
6458
Fig. 3. SP (Right) and corresponding atomic force microscopy images (Left) of the same silver nanoparticle film, 25 microns in width between the electrodes. The SP images scale from 0.0 V to 0.1 V from dark red to pink, and the topography images scale from 0 nm to 350 nm. Image pairs A-C correspond to a voltage gradient of 0.1, 0.2, and 0.3 V respectively. At 0.4 V, the behavior of the SP image is ohmiclike, showing a monotonic gradation. The structures in the SP images are not correlated to any topographical features. For the film shown D/2R= 1.20 for dots of size=70±7 Å. The film resistance is 80 MΩ. There is a bright spot near the right electrode, which corresponds to a hole in the conductive film as observed in the topographic image. [2]
δαi is the fractional range of variation in the site energies, δα, multiplied by a random number between −1 and 1 so that the site energy of the ith dot in the array is α0(1+ δαi). α0Vi is the external potential at the position of the ith dot, computed from the geometry of the array, and e is the charge on the electron. Without an external potential the energies of different dots are uncorrelated. Once a potential is applied, there is a systematic contribution to the site energy.
Fig. 4. The temperature dependence of the experimentally determined resistance for 70-Å Ag nanoparticle monolayer films is shown for three different size distributions of particles. All resistance measurements were taken at low bias (−0.3 V to 0.3 V) within the ohmic regime of the I/V curve. Three different temperature regimes are identified. At high temperatures (above ≈200 K) a metallic behavior is seen. Below TMA, the resistance changes over to a simple activated process where ln(Resistance) ∝ Ea/T. TMA is linearly correlated with Ea. At temperatures lower still, below TAV, a ln(Resistance) ∝ (T0/T)1/2 dependence is seen and is interpreted in terms of a variable range hopping mechanism. TAV moves to lower temperatures with decreasing size distribution.
Adjacent dots are correlated if they are oriented in the direction of the potential gradient while they remain uncorrelated otherwise. When the applied gradient is large enough, the systematic monotonic variation in the site energies can compete with the fluctuations. We confirmed this theoretical prediction by increasing the voltage drop per particle. Thereby one can observe (20) a transition from a domain localized state where there are “islands” of higher SP to stripe-like regions where the stripes are directed along the gradient. Such quantum mechanical computational results are shown in ref. 20. Upon further compression the array exhibits a transition to collective behavior, meaning a monotonic drop of the SP in the direction of the applied bias was observed, and also is clearly evident in the computations. DC transport measurements provide a particularly appealing way to quantify the electronic phase diagram of two-dimensional QD solids in various dimensions, including exchange coupling, temperature, and disorder. Fig. 4 is a representation of the measured resistivity vs. temperature for three different arrays with a small, but finite, size distribution for the dots. Imaging shows these fairly compressed arrays to be regularly ordered in a hexagonal packing. As temperatures drop to about 200 K, such a film will exhibit a decreasing resistance with decreasing temperature, just as any normal metal. At some temperature TMA (180–230 K), the conductivity properties change over to an activated mechanism such that ln(Resistance) ∝ Ea/T. TMA and Ea are linearly correlated, and Ea decreases with increasing lattice compression, implying that these two quantities represent a measurement of the strength of exchange coupling, β. At a lower temperature TAV (20–80 K), the transport characteristics again change, this time to a ln(Resistance) ∝ (T0/T)1/2 dependence. Such a T−1/2 dependence is known in the solid-state theory of amorphous conductors (21) as the regime of variable range hopping (VRH). The transition temperature to VRH, TAV is not correlated with Ea, but it does exhibit a strong, linear dependence on the width of the particle size-distribution. Thus, TAV is a measure of disorder. One interesting prediction that we have extracted from our experiments is that, at size distributions widths <3%, the transition from simple activated to VRH conductivity should disappear. The implication is that the states near the Fermi energy are no longer localized states, but instead are delocalized —such as those that exist at the mobility edge. This observation may have implications for generating true metallic behavior in a twodimensional lattice down to very low temperatures. Another relevant observation is that the scale parameter T0 for VRH appears to be inversely proportional to the width of the particle size distribution, but there appear to be some scatter in its value for different films. Interestingly, the same is true for the computations discussed below. We understand why that is the case for the computations but we need to further firm up this correspondence, if any, between experiment and theory. The quantum mechanical computations of the surface potential also can be used to determine the electrical conductivity of the array (16). Such computations assume that the charge transport is coherent. This is
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
QUANTUM DOT ARTIFICIAL SOLIDS: UNDERSTANDING THE STATIC AND DYNAMIC ROLE OF SIZE AND PACKING DISORDER
6459
analogous to what is done in solid-state physics for so-called mesoscopic systems (22, 23) and for conductivity of molecular wires (24–27). The current is then given by [3]
Here fl(E) and fr(e) are the Fermi-Dirac occupation probabilities of the left and right electrodes and T(E) is the transmission of the array and is obtained from the scattering computations as the squared modulus of the transition amplitude. In the linear regime one uses a first-order Taylor expansion to bring the voltage V explicitly out. The factor 2 is from the two spin directions. When the voltage drop per site, α0eVi of Eq. 2, is low the computed transmission is independent of the voltage. At higher field gradients the computed transmission function is, by itself, a function of the voltage as discussed in connection with Eq. 2. The trends in the computed transmission function are those that can be expected from the computations of the SP, as discussed above. In the domain localized regime the transmission is negligible unless the external voltage is high enough that stripes can be formed. Upon compression, the transmission increases exponentially up to where the dot-dot exchange coupling β saturates, as in Eq. 1. This saturation is as we have previously seen for both experiment and computations of the second harmonic response. On the other hand, the transmission function depends on the energy and this dependence is nontrivial even in the strong coupling regime (16). Low temperature measurements of the conductivity probe the energy dependence of the transmission on a very fine scale. The simple Hamiltonian that we use for computing and interpreting the other response functions of QD arrays does predict the observed trends in the conductivity (Fig. 5). We turn to a brief interpretation of the computational results. The key point is that, in the mesoscopic regime, the transmission is a weighted density of states. The implication is that the transmission has a cusp-like gap at the Fermi surface. (The detailed shape of the gap depends on the role of the charging energy I and its magnitude as compared to the range ∆α of the fluctuation in site energies.) This gap is being filled in as the disorder is increasing so that, at a high enough disorder the gap disappears. Similarly, the gap disappears as the lattice is expanded so that β decreases. We attribute the lower temperature behavior to the contribution of states in the gap in the transmission that is made possible by thermal excitation. Once the temperature is somewhat higher, one can access energies above the gap and the behavior changes over to a regular activated regime. Another factor that is relevant to the crossover is the exact position of the gap. At the highest possible values of β, that is, at a tightly compressed hexagonal lattice, the highest occupied level in the ground state (the HOMO in chemical terminology or the Fermi energy of solid state) is somewhat above the gap. Then, as the lattice is expanded, the energy of the highest orbital shifts down with respect to the gap and the role of the gap is more evident in the temperature dependence of the conductivity. It is an open question whether this effect, as seen in the computations, has an experimental counterpart. A final caveat about the computation is the sampling of disorder. We have computed by using hexagonal arrays of 55 dots per side, 8,911 dots in all. Even for such large arrays, there are few enough disorder-induced states in the gap that the scale parameter T0 is not strictly independent of the specific distribution of dot energies. Invariably, T0 scales as β2/∆α where ∆α is a measure of the variance of the site energies. The magnitude of T0 (below 0.5 eV) is in the same range as was observed, but it is somewhat sensitive to higher moments of the size distribution. This is not unexpected because the theory of domain localized regime (19) tells us that variable range coupling should scale with the variance of β2/α (which is not the same as β2/variance of α). Finally we mention that starting from the transmission function vs. energy, the resistivity as shown in Fig. 5 is not computed numerically but analytically and this makes it a rather sensitive probe for the role of temperature. In particular, the logarithmic derivative of the resistivity with respect temperature can be analytically computed.
Fig. 5. The temperature dependence of the computed resistivity in the low temperature regime. T in °K for a hexagonal array of 55 dots per side. The plot is for several values of disorder for a compressed lattice, D/2R=1.21. For lower compressions the resistivity is far higher as it is is constant at very low temperatures that shows that lnR scales as for more disordered lattices. The logarithmic derivative, This means that the charge transfer is not necessarily to the neighboring dot but possibly to a dot that is further away but for which the site energies are more nearly equal. At higher temperatures there is a crossover to an activated transport, ln R∝Ea/kT for which case the logarithmic derivative is linear in The value of Ea is not sensitive to the extent of disorder. The resistivity is computed from the transmission that is the thermal average of a weighted density of states for energies above the HOMO (16).
In conclusion, temperature provided us with a finer probe of transport in QD arrays and highlighted effects of disorder not hitherto observed. This work was funded by the Department of Energy and a Collaborative University of California/Los Alamos Research grant. F.R. is a “Maître de Recherches,” Fonds National de la Recherche Scientifique, Belgium and thanks Liège University for a “Crédit d'impulsion” grant. Computational facilities were provided by Sonderforschunbereich 377 (Hebrew University) and Numerically Intensive Computing (Liège University). 1. Markovich, G., Collier, C.P., Henrichs, S.E., Remacle, F., Levine, R.D. & Heath, J.R. (1999) Acc. Chem. Res. 32, 415–423. 2. Schon, J.H., Kloc, C., Hwang, H.Y. & Batlogg, B. (2001) Science 292, 252–254. 3. Mannhart, J. (1995) Philos. Trans. R. Soc. A 353, 377–385. 4. Lin, X.M., Jaeger, H.M., Sorensen, C.M. & Klabunde, K.J. (2001) J. Phys. Chem. B 105, 3353–3357. 5. Peng, Z.A. & Peng, X.G. (2001) J. Am. Chem. Soc. 123, 183–184. 6. Mott, N.F. (1990) Metal-Insulator Transitions (Taylor & Francis , London). 7. Kubo, R. (1962) J. Phys. Soc. Japan 17, 975–986. 8. Lambe, J. & Jaklevic, R.C. (1969) Phys. Rev. Lett. 22, 1371–1375. 9. Murray, C.B., Kagan, C.R. & Bawendi, M.G. (2000) Annu. Rev. Mat. Sci. 30, 545–610. 10. Puntes, V.F., Krishnan, K.M. & Alivisatos, A.P. (2001) Science 291, 2115–2117. 11. Heath, J.R., Knobler, C.M. & Leff, D.V. (1997) J. Phys. Chem. 101, 189–197. 12. Collier, C.P., Saykally, R.J., Shiang, J.J., Henrichs, S.E. & Heath, J.R. (1997) Science 277, 1978–1981. 13. Remacle, F., Collier, C.P., Heath, J.R. & Levine, R.D. (1998) Chem. Phys. Lett. 291, 453–458. 14. Remacle, F. & Levine, R.D. (2000) J. Am. Chem. Soc. 122, 4084–4091. 15. Henrichs, S., Collier, C.P., Saykally, R.J., Shen, Y.R. & Heath, J.R. (2000) J. Am. Chem. Soc. 122, 4077–4083. 16. Remacle, F., Beverly, K.C., Heath, J.R. & Levine, R.D. (2002) J. Phys. Chem., in press. 17. Kim, S.H., Medeiros-Ribeiro, G., Ohlberg, D.A.A., Williams, R.S. & Heath, J.R. (1999) J. Phys. Chem B 103, 10341–10347. 18. Remacle, F. & Levine, R.D. (2000) J. Phys. Chem. A 104, 10435–10441. 19. Remacle, F. & Levine, R.D. (2001) J. Phys. Chem. B 105, 2153–2162. 20. Sample, J.L., Beverly, K.C., Chaudhari, P.R., Remacle, F., Heath, J.R. & Levine, R.D. (2001) Adv. Mat. 14, 124–128. 21. Shklovskii, B.I. & Efros, A.L. (1984) Electronic Properties of Doped Semiconductors (Springer, Berlin). 22. Datta, S. (1995) Electronic Transport in Mesoscopic Systems (Cambridge Univ. Press, Cambridge, U.K.). 23. Imry, Y. & Landauer, R. (1999) Rev. Mod. Phys. 71, S306–S312. 24. Mujica, V., Kemp, M. & Ratner, M.A. (1994) J. Chem. Phys 101, 6849–6855. 25. Yaliraki, S.N. & Ratner, M.A. (1998) J. Chem. Phys. 109, 5036–5043. 26. Samanta, M.P., Tian, W., Datta, S., Henderson, J.I. & Kubiak, C.P. (1996) Phys. Rev. B 53, R7626–R7629. 27. Tian, W.D., Datta, S., Hong, S.H., Reifenberger, R., Henderson, J.I. & Kubiak, C.P. (1998) J. Chem. Phys 109, 2874–2882. 28. Remacle, F. & Levine, R.D. (2000) Proc. Natl. Acad. Sci. USA 97, 553–558.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
SEGMENTED NANOFIBERS OF SPIDER DRAGLINE SILK: ATOMIC FORCE MICROSCOPY AND SINGLE-MOLECULE FORCE SPECTROSCOPY
6460
Colloquium Segmented nanofibers of spider dragline silk: Atomic force microscopy and single-molecule force spectroscopy E.Oroudjev*, J.Soares†, S.Arcidiacono†, J.B.Thompson*, S.A.Fossey†, and H.G.Hansma*‡ *Department of Physics, University of California, Santa Barbara, CA 93106; and †U.S. Army Natick R&D Center, Natick, MA 01760 Edited by Ignacio Tinoco, Jr., University of California, Berkeley, CA, and approved February 5, 2002 (received for review October 3, 2001) Despite its remarkable materials properties, the structure of spider dragline silk has remained unsolved. Results from two probe microscopy techniques provide new insights into the structure of spider dragline silk. A soluble synthetic protein from dragline silk spontaneously forms nanofibers, as observed by atomic force microscopy. These nanofibers have a segmented substructure. The segment length and amino acid sequence are consistent with a slab-like shape for individual silk protein molecules. The height and width of nanofiber segments suggest a stacking pattern of slab-like molecules in each nanofiber segment. This stacking pattern produces nano-crystals in an amorphous matrix, as observed previously by NMR and x-ray diffraction of spider dragline silk. The possible importance of nanofiber formation to native silk production is discussed. Force spectra for single molecules of the silk protein demonstrate that this protein unfolds through a number of rupture events, indicating a modular substructure within single silk protein molecules. A minimal unfolding module size is estimated to be around 14 nm, which corresponds to the extended length of a single repeated module, 38 amino acids long. The structure of this spider silk protein is distinctly different from the structures of other proteins that have been analyzed by single-molecule force spectroscopy, and the force spectra show correspondingly novel features. The last decade has seen a significant increase in the scientific literature on spider dragline silk. This interest is due to the impressive mechanical properties of spider dragline silk, at a time when biomaterials and biomimetics are both exciting interest in the rapidly growing field of materials research. The viscoelastic fibers of spider dragline silk combine both a high tensile strength that is comparable to steel and is only slightly inferior to Kevlar (≈2/3 of its tensile strength), and a high elasticity (≈30% elongation before failure) that is comparable to rubber (1– 4). This unique combination makes spider dragline silk mechanically superior to almost any other natural or man-made material. It is apparent that the mechanical properties of the dragline silk protein's intramolecular structure as well as the intermolecular organization of these proteins in the fiber are critical for spider silk performance (2, 5). We report here the partial mechanical and structural characterization of a recombinant dragline silk protein. This recombinant silk protein provides a valuable test system for establishing the relationships between protein structure and mechanical properties in spider silk. Spider dragline silk can be pictured as a composite material consisting of a semiamorphous matrix filled with tiny (<10 nm) nanocrystalline-like particles (6, 7). The amino acid sequence for spider dragline silk proteins is comprised of poly(A) [poly(alanine)], for some silks substituted by poly(GA) [poly(glycyalanine)], and glycine-rich sequences (2, 8, 9). Despite intensive structural studies on spider dragline silk proteins, their exact structural organization remains to be solved. The 4- to 10-residue-long poly(A) and poly(GA) motifs are thought to be involved in the formation of β-sheet nano-crystalline-like particles. Glycine-rich sequences are thought to fold into some non-α-helical helical structure for GGX or into β-turns for GPGGX, thus forming the semiamorphous matrix (2, 10). On the other hand, a few reports suggest that at least part of these glycine-rich motifs can also fold into β-sheets (11) and/or form an interphase between crystalline-like objects and a semiamorphous matrix (12). NMR and x-ray diffraction experiments show that the crystalline-like particles are well oriented along the silk fiber with polypeptide chains parallel and alanine residues perpendicular to the fiber axis (7, 13–17). These findings suggest that protein molecules are overall well oriented in the silk fiber. This high degree of molecular orientation, together with structural organization of dragline silk proteins, is a prerequisite for the unique mechanical properties of the whole silk fiber (18). Recombinant spider silk proteins (19–21) have advantages over natural dragline silk for single-molecule research. Recombinant spider silk proteins have a regular, known sequence and can readily be purified in adequate quantities. Natural spider dragline silk proteins are larger and are difficult to solubilize after they have formed silk fibers. Two consensus sequences, SPI and SPII, represent major repetitive elements from spider dragline silk proteins. A family of recombinant spider silk proteins have been synthesized from these SPI and SPII sequences (22). The SPI sequence consists of 38 aa and includes 16-aa-long poly(A) and poly(GA) stretches, flanked on both sides with a total of 22 aa forming glycine-rich GGX motifs (Fig. 1). The SPII sequence consists of 12 aa representing two GPGGX motifs. All circular dichroism measurements on proteins from this protein family indicate that almost two thirds of the secondary structure of (SPI)n
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC. This paper was submitted directly (Track II) to the PNAS office. Abbreviations: AFM, atomic force microscope/microscopy; (GGX)n, repeating aa sequences glycine-glycine-X, where X is usually leucine, tyrosine, or glutamine, represented as a loose or tight spiral in Figs. 3A and 5A and B; H-bond, hydrogen bond; poly(A/GA), aa sequences of poly(alanine)/poly (glycylalanine), represented as a zig-zag in Figs. 3A and 5 A and B; poly(A/GA+GGX), a poly(A/GA) sequence of SPI plus the (GGX)n sequence of SPI that follows it: GAGAAAAAAAAAAGGAGQGGYGGLGSQGTSGRGGLGGQ; pS(4+1), modular recombinant spider silk protein composed of 16 SPI and 4 SPII modules arranged as follows: (SPI)4-SPII-(SPI)4-SPII-(SPI)4-SPII-(SPI)4-SPII (Fig. 1); SPI, 38-aa-long synthetic spider silk sequence based on sequence from Nephila clavipes: SGRGGLGGQGAGAAAAAAAAAAGGAGQGGYGGLGSQGT; SPII, 12-aa-long synthetic spider silk sequence based on sequence from N. clavipes: SGPGGYGPGQQT; WLC, worm-like chain (polymer model). ‡To whom reprint requests should be addressed. E-mail:
[email protected].
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
SEGMENTED NANOFIBERS OF SPIDER DRAGLINE SILK: ATOMIC FORCE MICROSCOPY AND SINGLE-MOLECULE FORCE SPECTROSCOPY
6461
modular proteins is made of β-sheets and the rest is mostly β-turns or similar structures with no significant amount of α-helical structure observed (22). This observation is in good agreement with data obtained previously on native spider dragline silk fibers (9, 16) and suggests that recombinant synthetic modular SPI/SPII proteins have a structural organization very similar (if not identical) to native dragline silk proteins. We used the SPI/SPII modular protein in Fig. 1 for studying the mechanical properties and structural organization of spider silk proteins. The modular protein in Fig. 1 has the formula [(SPI)4+SPII]4, which we call pS(4+1). With a 52-kDa molecular mass, this is the largest protein available to us from the family of SPI/SPII modular proteins. Its large size is advantageous for both atomic force microscopy (AFM) imaging and single molecule force spectroscopy.
Fig. 1. The sequence of pS(4+1) recombinant silk protein, with selected SPI and SPII modules identified, as well as poly(A/GA) sequences and (GGX)n sequences. N′- and C′- indicate the N-terminal and C-terminal amino acids, respectively.
MATERIALS AND METHODS Sample Preparation. The gene for a pS(4+1) modular recombinant protein was produced as described (22) and inserted in the pET24 expression vector (Novagen). The protein sequence is made of 16 SPI and 4 SPII modules (22) arranged in (SPI)4-SPII-(SPI)4-SPII-(SPI)4-SPII(SPI)4-SPII pattern (Fig. 1). The pS(4+1) recombinant protein was produced in the Escherichia coli expression strain BL21(DE3) pLysS grown to midlog phase in defined salts medium (23) with 30 µg/ml kanamycin and 34 µg/ml chloramphenicol. Expression was induced with 1 mM isopropyl β-D-thiogalactoside (IPTG). The cells were harvested by centrifugation and lyophilized for purification. The collected cell pellet was lysed with organic acid under denaturing conditions and clarified by centrifugation. The solution was dialyzed into 2 M urea/10 mM Tris, pH 9.9, and loaded on a QAE-Sephadex A50 (Amersham Pharmacia Biotech) column. The flow-through was collected, and the column was washed to recover the silk protein. Protein was dialyzed in a “silk buffer” (160 mM urea/10 mM NaH2PO4/1 mM Tris/10 mM glycine, pH 5.0) and kept at 4°C as a 1 mg/ml stock solution. The silk protein prepared in this way was 99.1% pure, based on amino acid composition analysis. For AFM observations, the stock solution was diluted in the silk buffer to a desired concentration (100–300 µg/ml) and deposited as a drop on a freshly cleaved mica surface. Protein molecules were allowed to bind to the mica by incubating for 3–5 min, and excess protein was removed by washing with a flow of silk buffer. For observation in air, this procedure was followed by a wash with milliQ-purified water, and the sample was dried under a stream of compressed air purified by passing it through a 0.22-µm filter. For observation under the liquid, the sample was washed with the corresponding solution and was imaged immediately under water or the corresponding buffer to prevent the sample from drying. For force spectroscopy experiments, samples were prepared essentially as for AFM imaging. No difference was detected between air-dried and non-dried samples. All samples were submerged, just before pulling, under milliQ-purified water with 10 mM CaCl2. AFM Imaging and Force Spectroscopy. AFM imaging, both in air and under aqueous solution, was performed by standard procedures in tapping mode on a MultiMode AFM with E scanner and Nanoscope III electronics (Digital Instruments, Santa Barbara, CA). Cantilevers for AFM imaging were also obtained from Digital Instruments. AFM images were captured, processed, and analyzed with NANOSCOPE III software, versions 4.42r4 and 4.43r8 (Digital Instruments). Force spectroscopy was performed on a Molecular Force Probe MFP-SA (Asylum Research, Santa Barbara, CA). Silicon nitride cantilevers were obtained from Park Scientific (Stanford, CA). The spring constant for each cantilever was determined by measuring the amplitude of its thermal fluctuations (24), by using the Asylum Research software. The MFP was used to obtain force-vs.-piezo-extension curves, which were calibrated and partially analyzed by a manufacturer-supplied software based on IGOR PRO (WaveMetrics, Lake Oswego, OR) software. To perform worm-like-chain (WLC) curve-fit analyses, this software was correspondingly modified by using the built-in macro language. RESULTS AND DISCUSSION AFM of Synthetic Silk. Most of the pS(4+1) silk protein appeared as a fibrous material with a tendency to form nanofiber aggregates on freshly cleaved mica, when imaged in air or in liquid (Fig. 2). The pS(4+1) silk nanofibers were observed not only on bare mica, which is hydrophilic and negatively charged, but also on surfaces that were hydrophobic and either neutral or positively charged (data not shown). The two hydrophobic surfaces were prepared on mica by depositing monolayers of positively charged APTES (3-aminopropyltriethoxysilane), or neutral OcTES (n-octyltriethoxysilane). These findings indicate that nanofiber formation is an intrinsic property of the pS(4+1) silk protein rather than an artifact caused by the interactions of pS(4+1) with the bare mica surface. Each fiber shows a distinct segmented substructure (Fig. 2 B and C), with an average segment length of 35±9 nm, average height of 3±1.5 nm at the center of the segment, and apparent width of 34±5 nm when imaged in air. Mean volumes of these segments are estimated to be in the range of 1000–1300 nm3. Volumes of biomolecules in AFM images tend to correspond to the volumes predicted by their molecular masses, assuming a molecular density of 1–1.3 g/ml (25–27). Based on this relationship, segments have a mean molecular mass of ≈1000–1600 kDa. This mass is of order 20–35 times the molecular mass of the pS(4+1) protein monomer of 52 kDa. In addition to pS(4+1) nanofibers, we observed single blobs that are roughly comparable in size and volume to the segments in the nanofibers (Fig. 2, fat arrows). We tentatively identify these blobs as isolated nanofiber segments. There are only a few blobs; the great majority of the material on the surface is either fibrous or what looks like aggregates of fibrous material. If the
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
SEGMENTED NANOFIBERS OF SPIDER DRAGLINE SILK: ATOMIC FORCE MICROSCOPY AND SINGLE-MOLECULE FORCE SPECTROSCOPY
6462
pS(4+1) blobs bind to the surface as well as the pS(4+1) nanofibers, then nearly 100% of the pS(4+1) molecules were competent for nanofiber formation under imaging conditions.
Fig. 2. Silk nanofibers formed from pS(4+1) silk protein deposited on mica; AFM height images. (A) pS(4+1) silk protein is present primarily as aggregates of nanofibers. (B and C) Close-up AFM images of the pS(4+1) nanofibers show segmented substructure. Fat arrows indicate isolated blobs that are predicted to be segments of nanofibers, based on their sizes. Thin arrows indicate bulges, which often occur at branch points on nanofibers and may be due to nanofibers overlapping.
Bulges were observed on some of the nanofibers (narrow arrows in Fig. 2); their heights were nearly twice as high as the 3-nm heights of typical nanofiber segments. Although observed in different parts of nanofibers, these bulges often coincide with occasions when two nanofibers appear to overlap each other (Fig. 2C Lower). A simple pile-up of the nanofibers can explain such bulges. As seen in Fig. 2, pS(4+1) nanofibers show little branching, and, in fact, we propose that much of the observed nanofiber branching is due to nanofibers overlapping or piling up. Therefore, we propose that the pS(4+1) silk protein possesses a regular and well defined secondary and, possibly, tertiary structure, with well defined fiber-forming structures/ signals localized on two opposite ends of each pS(4+1) nanofiber segment. The nanofibers of pS(4+1) protein display a high degree of flexibility at intersegment contacts as well as a high variability in fiber length. We observed nanofibers as short as 60 nm and as long as 660 nm. The longest nanofibers have more than 20 segments. It is obvious, however, that some factors restrict infinite growth of the nanofibers. We propose that bonds at contact sites are relatively weak and break because of shearing forces on long nanofibers. On the other hand, “shortening” of the nanofibers can be due to misfolding of “fiber-forming” structures/ signals at one of the segment ends. Although the pS(4+1) silk protein is a synthetic protein with a molecular mass (52 kDa)≈1/6 the mass of native spider dragline silk proteins (250–350 kDa; ref. 5), it is reasonable to expect that the secondary and tertiary structures of these proteins are very closely related and that native silk proteins display similar fiber-forming properties. If this assumption is true, it sheds new light on the silk production picture. Dragline silk is produced by a draw down of a liquid silk dope (28). It seems that, at some stages of the silk dope processing, the protein molecules go from an isotropic state to a nematic liquid-crystalline state (5, 28). This transition is important both for reducing the viscosity of the dope and for reducing the energy needed to align the protein molecules along the silk fiber axis. Fiber formation, similar to the one observed for pS(4+1), can facilitate such isotropic-to-nematic transition for native silk proteins. Transition of monomeric pS(4+1) into fibrous form can lower the minimal necessary flow rates needed to convert from the isotropic state to the nematic state in the ducts of silk-producing ampullate glands and will further reduce the viscosity of silk dope. There is evidence from observations with transmitted polarized light microscopy (29) that the nematic phase of spider silk could be the result of just such a supermolecular assembly of silk polymers as we have observed here.
Fig. 3. Single-molecule force spectroscopy for pS(4+1) silk protein molecules. (A) Diagram of experimental setup, in which rupture peaks are believed to stretch one or more poly(A/GA+GGX) repeating units. (B and C) Force spectra (force-vs.-extension curves) for the unfolding of single molecules of pS(4+1) silk protein. The WLC model curves are fitted to each rupture force peak. The persistence length for each fit is 0.4 nm. Arrows indicate high-force rupture events at the beginning of the pulls, most likely because of multiple protein molecules attached to the tip of the AFM probe.
Molecular Force Spectroscopy of Synthetic Silk. Molecular force spectroscopy gives force-vs.-extension plots for the pS(4+1) silk protein molecules (Fig. 3). The numerous high-force rupture events at the beginning of the pulls are, most likely, because of multiple protein molecules attached to the tip of the AFM probe (arrows, Fig. 3 B and C). As the probe is moved away from the surface, it breaks contact with the surface and with most of the protein molecules. In some cases, only a single molecule will survive the initial pull distances and will stretch with sequential breaking of structural elements, as diagrammed in Fig. 3A. Such a molecule will break off the tip only after being pulled for significantly longer distances. In these cases, a single-molecule
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
SEGMENTED NANOFIBERS OF SPIDER DRAGLINE SILK: ATOMIC FORCE MICROSCOPY AND SINGLE-MOLECULE FORCE SPECTROSCOPY
6463
force spectrum is recorded at the latter part of the pull as a series of rupture peaks (Fig. 3 B and C).
Fig. 4. Histogram analysis of length increase (peak-to-peak distances) for rupture events in pS(4+1) silk protein molecules. Note that most of the data coincide with ≈N×14 nm in length increase, where N=1, 2, or 3. (Inset) Length increase vs. corresponding rupture force.
In numerous previous force spectroscopy studies, the WLC model (30) was used to characterize mechanical unfolding of proteins. The WLC describes the relationship between the protein extension and the entropic force generated as a result of such extension. In particular, the WLC gives peak-to-peak distances that agree closely with the distances expected from the lengths of unfolded polypeptide (31). We fitted the WLC model to our experimental data by varying two model parameters: persistence length and contour length for the polypeptide backbone chain. The fit was first done to the last rupture event on each selected plot, as a test for a number of pulled molecules. If the fitted persistence length was near 0.4 nm, a typical value for a single polypeptide chain at high force (32), then all preceding rupture events on the plot were fitted with the WLC model for contour length with the persistence length fixed at 0.4 nm. Experimental data and WLC fits were plotted on the same graph (Fig. 3 B and C) to visually assess the quality of the fit. Results from all pulls with single molecule rupture events were analyzed statistically. The average force for rupture peaks was 176±73 pN (Fig. 4 Inset). The high degree of variability in rupture force can be due to both different pulling speeds (refs. 32 and 33; we used from 200 nm/s to 1500 nm/s) and stochastic variability. These average rupture forces are close to the ones obtained by force spectroscopy for mechanical/structural proteins such as titin (31) and tenascin (31, 34). Both tenascin and titin have Ig-like β-sheet barrel folds made by two antiparallel β-strands, each held together by a number of inter-strand hydrogen bonds (H-bonds) arranged in parallel (35). On the other hand, rupture forces are smaller for α-helix-rich proteins such as spectrin (32), and for proteins that do not normally experience forces, such as the enzyme barnase (36). This difference is attributed to the fact that intrahelix H-bonds in α-helices are arranged in series and that the interhelix hydrophobic or van der Waals interactions between α-helices are significantly weaker than the H-bonds between β-sheets. Thus, the pS(4+1) silk protein seems mechanically to belong to the β-sheet group of proteins. This result correlates well with the predicted β-sheet/β-turn fold for spider dragline silk proteins (2, 10) and with CD data for SPI/ SPII modular proteins (22) and dragline silk fibers (9, 16). The sawtooth pattern of the unfolding force curve for pS(4+1) (Fig. 3 B and C) is similar to other proteins with modular structures (31, 32, 34). This pattern shows that, after the pulling force reaches a certain threshold value, a defined structural unit in the protein breaks in a cooperative way. It is a common observation that the difference in contour length between adjacent rupture peaks corresponds to the length of polypeptide chain released after the preceding rupture event/peak. We calculated the total length of the extended polypeptide chain for a single pS(4+1) protein molecule to be ≈243 nm, based on the assumption that a single amino acid in the polypeptide chain can be extended up to 0.37 nm in length. The maximum measured total extension for the single-molecule rupture events never exceeded 246 nm. This result indicates that our pulls, as in Fig. 3 B and C, are on single pS(4+1) protein molecules, and that single unfolded pS(4+1) molecules lose contact with the other pS(4+1) molecules in the silk nanofiber after they unfold. The peak-to-peak distances in pS(4+1) unfolding force curves were not uniform. As seen in the histogram of Fig. 4, the increases in contour length between successive rupture peaks form three distinct groups clustered around ≈14 nm, 28 nm, and 42 nm. The calculated contour length for a fully extended SPI module is 14.1 nm; SPII modules will add an additional 4.4 nm, respectively, and SPI sequences occur four times as frequently as SPII sequences (Fig. 1). These findings indicate that an individual rupture event unfolds a length comparable to 1–3 SPI modules (sometimes together with an SPII module). We only rarely recorded peak-to-peak rupture distances above 45 nm (≈6% of rupture distances were >45 nm). There was no correlation between the force required for a particular rupture event and the length of polypeptide chain released (Fig. 4 Inset). As in our model of Fig. 3A, we predict that a 14-nm rupture event actually pulls up a poly(A/GA) sequence plus the following (GGX)n sequence. We name these repeating units “poly(A/ GA+GGX).” The peak-to-peak distances are uniform for the other recombinant modular proteins studied by force spectroscopy to date (31, 32, 34). Titin, for example, is composed of tandem Ig domains, and each rupture event corresponds to the unfolding of one independently folded Ig domain. This independent unfolding of titin's Ig domains suggests weak interactions or no interactions between the Ig domains of titin. Titin also differs from pS(4+1) silk protein in that much “hidden” polypeptide length is released when a domain unfolds, because only a few of the amino acids in Ig domains are subjected to the mechanical load during pulling (31). Unlike titin's ruptures, the ruptures of pS(4+1) silk protein often release two or more repeating units of poly(A/ GA+GGX). Furthermore, according to our model in Fig. 3A, poly(A/GA+GGX) silk protein repeats are not independently folded. Instead, these poly (A/GA+GGX) repeats are bonded to distant sequences on the pS(4+1) protein molecule. Specifically, the poly(A/GA) repeats form β-sheets within the protein molecule. We rarely observe a catastrophic unfolding of an entire pS(4+1) protein molecule. Therefore, the intramolecular forces hold the pS(4 +1) molecule together even after one or more poly (A/GA+GGX) repeats have been pulled from it. Model for Silk Structure. The pS(4+1) nanofiber morphology, as well as its mechanical unfolding pattern, can be explained by the model in Fig. 5. In this model, a single pS(4+1) protein molecule folds into a well defined structure in which β-strands of poly(A/ GA) sequences in SPI modules form four H-bond-stabilized β-sheets (Fig. 5A). These β-sheets alternate with the non-β (GGX)n sequences of the SPI modules, which form random coils
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
SEGMENTED NANOFIBERS OF SPIDER DRAGLINE SILK: ATOMIC FORCE MICROSCOPY AND SINGLE-MOLECULE FORCE SPECTROSCOPY
6464
or non-α-helices. These folded molecules have the shape of an elongated slab with a pattern of alternating hydrophobic and hydrophilic regions formed by β-sheets and by non-β (GGX)n regions, correspondingly. Each slab is a single pS(4+1) molecule, as in Figs. 1 and 5A. The calculated dimensions of a slab are ≈0.54 nm high×2 nm wide×40 nm long. The height and width come from the spacing between typical antiparallel βsheets (estimated as 0.53–0.55 nm) and the interchain distance along the H-bond direction (estimated as 0.47 nm; refs. 37 and 38). The length comes from the length of four folded SPI domains, calculated from their amino acid sequences [0.32–0.35 nm/aa in β-strand poly(A/GA) and ≈0.2 nm/aa in non-α-helical (GGX)n sequences]. The calculated length of the slabs—≈40 nm—is close to the measured segment length in AFM images of pS(4+1) silk nanofibers, which is 35±9 nm.
Fig. 5. A model for pS(4+1) silk nanofiber organization. (A) The single pS(4+1) protein molecule's polypeptide chain folds into a flat slablike structure in which four hydrophobic β-sheets of poly(A/GA) (zig-zags) are separated by hydrophilic non-α-helical GGX structures (spirals). Four poly(A/GA) β-strands form each of the β-sheets that compose the crystalline-like structures of spider dragline silk. The (GGX) n sequences are random coils or 310-helices or other non-α-helical helical structures. (B) Approximately 30 of these slab-like molecules form a “stack” or “nanofiber segment” because of hydrophobic interactions between β-sheets in aqueous environment. These nanofiber segments are thought to bind to each other through specific “fiber-forming” signals/structures at their ends. Alternate segment models show either perfect or imperfect (staggered) alignment. (C) The whole pS(4+1) protein nanofiber can be viewed as a chain of segments with each segment representing a single pS(4+1) protein “stack.” Under a stretching force (not shown), as in the draw-down step of silk processing, the secondary structure of semiamorphous GGX “matrix” transitions into a more extended form and locks into a 31-helix or β-strand configuration. Numerous inter- and intramolecular H-bonds are formed at this point between these newly formed structures.
In an aqueous environment, these molecular slabs will tend to aggregate, because of intermolecular hydrophobic interactions between their β-sheets. This aggregation is modeled in Fig. 5B as a stack of pS(4+1) molecular slabs. The silk protein molecules are aligned along the axis of the β-strands, as is known from NMR and x-ray diffraction (7, 15). In our diagram of Fig. 5B, these stacks are six slabs high and five slabs wide, based on the dimensions and estimated molecular weight of nanofiber segments in AFM images (Fig. 2 B and C). Because of strong and uniform hydrophobic interactions and H-bonds, the β-sheets in these slabs will form crystalline three-dimensional structures like ones found in native spider dragline silks (6, 7). The H-bonds between β-strands in a β-sheet (and the hydrophobic interactions between β-sheets to a lesser degree) can be viewed as the major stabilizing forces in such structures. The ends of each stack seem to carry signals/structures that facilitate pS (4+1) nanofiber formation. Thus, we suggest that each nanofiber segment corresponds to a separate pS(4+1) stack (Fig. 5 B and C). The draw-down process, occurring as one of the last steps in a silk fiber production pathway (5), causes major structural rearrangements in silk proteins, resulting in a significant improvement in the fiber mechanical strength (39, 40). During this draw-down step, the glycine-rich part of silk proteins can partially transition from random coil or non-α-helical to the more extended structures, such as 31-helix (10) or β-strand (11). Similar structural rearrangements occur in the silk fibers when they are under a force load (41). We propose that, during single-molecule force spectroscopy, one or more repeating units [poly(A/GA+GGX)] are pulled from the edge of a slab (Fig. 3A). The spacing between rupture peaks occurs in multiples of ≈14 nm (Fig. 4), which is longer than the 10-nm length of a poly(A/ GA+GGX) repeat in the compact “slab” form. This extension from 10 nm to 14 nm is predicted to occur primarily in the non-β GGX sequences, because they have much smaller amino acid spacings in their compact form than in their extended forms. Spider dragline silk, like lustrin in abalone shell and titin in muscle, thus appears to derive much of its combination of strength and toughness from its modular sacrificial bonds (31, 42). Of course, detailed structural and mechanical studies on spider dragline silk proteins should be performed in the future to validate our model, which remains highly speculative at this point. Our results show that pS(4+1)-like proteins can be used as a test model for additional studies on recombinant modular SPI/ SPII proteins with structurally modified modules. Research on such model systems is enhancing our understanding of the relationship between the spider dragline silk proteins' sequences/ structures and their mechanical properties. We thank Nathan Becker for writing valuable macros to analyze pulling curves. This research was supported by National Science Foundation Molecular and Cellular Biosciences (H.G.H., E.O.), National Science Foundation Division of Materials Research-9988640 (J.B.T.), the Materials Research Laboratory (National Science Foundation DMR96– 32716 (to J.B.T.)), and Asylum Research. 1. Hinman, M.B., Jones, J.A. & Lewis, R.V. (2000) Trends Biotechnol. 18, 374–379. 2. Hayashi, C.Y., Shipley, N.H. & Lewis, R.V. (1999) Int. J. Biol. Macromol. 24, 271–275. 3. Tirrell, D.A. (1996) Science 271, 39–40. 4. Cunniff, P.M., Fossey, S.A., Auerbach, M.A., Song, J.W., Kaplan, D.L., Adams, W.W., Eby, R.K., Mahoney, D. & Vezie, D.L. (1994) Polym. Adv. Technol. 5, 401–410. 5. Vollrath, F. & Knight, D.P. (2001) Nature (London) 410, 541–548. 6. Grubb, D.T. & Jelinski, L.W. (1997) Macromolecules 30, 2860–2867. 7. Yang, Z.T., Liivak, O., Seidel, A., LaVerde, G., Zax, D.B. & Jelinski, L.W. (2000) J. Am. Chem. Soc. 122, 9019–9025. 8. Gatesy, J., Hayashi, C., Motriuk, D., Woods, J. & Lewis, R. (2001) Science 291, 2603–2605. 9. Warner, S.B., Polk, M. & Jacob, K. (1999) J. Macromol. Sci. Rev. Macromol. Chem. Phys. C39, 643–653.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
SEGMENTED NANOFIBERS OF SPIDER DRAGLINE SILK: ATOMIC FORCE MICROSCOPY AND SINGLE-MOLECULE FORCE SPECTROSCOPY
6465
10. Kummerlen, J., Vanbeek, J.D., Vollrath, F. & Meier, B.H. (1996) Macromolecules 29, 2920–2928. 11. Fukushima, Y. (2000) Polym. Bull. 45, 237–244. 12. Fossey, S.A. & Tripathy, S. (1999) Int. J. Biol. Macromol. 24, 119–125. 13. Seidel, A., Liivak, O., Calve, S., Adaska, J., Ji, G.D., Yang, Z.T., Grubb, D., Zax, D.B. & Jelinski, L.W. (2000) Macromolecules 33, 775–780. 14. Jelinski, L.W., Blye, A., Liivak, O., Michal, C., LaVerde, G., Seidel, A., Shah, N. & Yang, Z. (1999) Int. J. Biol. Macromol. 24, 197–201. 15. Grubb, D.T. & Ji, G. (1999) Int. J. Biol. Macromol. 24, 203–210. 16. Shao, Z., Vollrath, F., Sirichaisit, J. & Young, R.J. (1999) Polymer 40, 2493–2500. 17. Shao, Z.Z. & Vollrath, F. (1999) Polymer 40, 1799–1806. 18. Termonia, Y. (1994) Macromolecules 27, 7378–7381. 19. Lewis, R.V., Hinman, M., Kothakota, S. & Fournier, M.J. (1996) Protein Expression Purif. 7, 400–406. 20. Fahnestock, S.R. & Bedzyk, L.A. (1997) Appl. Microbiol. Biotechnol. 47, 33–39. 21. Arcidiacono, S., Mello, C., Kaplan, D., Cheley, S. & Bayley, H. (1998) Appl. Microbiol. Biotechnol. 49, 31–38. 22. Prince, J.T., McGrath, K.P., Digiolamo, C.M. & Kaplan, D.L. (1995) Biochemistry 34, 10879–10885. 23. Riesenberg, D., Schulz, V., Knorre, W.A., Pohl, H.D., Korz, D., Sanders, E.A., Ross, A. & Deckwer, W.D. (1991) J. Biotechnol. 20, 17–27. 24. Hutter, J.L. & Bechhoefer, J. (1993) Rev. Sci. Instrum. 64, 3342–3342. 25. Golan, R., Pietrasanta, L.I., Hsieh, W. & Hansma, H.G. (1999) Biochemistry 38, 14069–14076. 26. Pietrasanta, L.I., Thrower, D., Hsieh, W., Rao, S., Stemmann, O., Lechner, J., Carbon, J. & Hansma, H. (1999) Proc. Natl. Acad. Sci USA 96, 3757– 3762. 27. Schneider, S.W., Larmer, J., Henderson, R.M. & Oberleithner, H. (1998) Pflügers Arch. Eur. J. Physiol. 435, 362–367. 28. Knight, D.P. & Vollrath, F. (1999) Proc. R. Soc. London Ser. B 266, 519–523. 29. Viney, C., Huber, A.E., Dunaway, D.L., Kerkam, K. & Case, S.T. (1994) in Silk Polymers: Materials Science and Biotechnology, ACS Symposium Series, ed. Kaplan, D. (Am. Chem. Soc., Washington, DC), Vol. 544, pp. 120–136. 30. Bustamante, C., Marko, J.F., Siggia, E.D. & Smith, S. (1994) Science 265, 1599–1600. 31. Fisher, T.E., Oberhauser, A.F., Carrion-Vazquez, M., Marszalek, P.E. & Fernandez, J.M. (1999) Trends Biochem. Sci. 24, 379–384. 32. Rief, M., Pascual, J., Saraste, M. & Gaub, H.E. (1999) J. Mol. Biol. 286, 553–561. 33. Evans, E. (2001) Annu. Rev. Biophys. Biomol. Struct. 30, 105–128. 34. Oberhauser, A.F., Marszalek, P.E., Erickson, H.P. & Fernandez, J.M. (1998) Nature (London) 393, 181–185. 35. Brändén, C.-I. & Tooze, J. (1999) Introduction to Protein Structure (Garland, New York). 36. Best, R.B., Li, B., Steward, A., Daggett, V. & Clarke, J. (2001) Biophys. J. 81, 2344–2356. 37. Obrien, J.P., Fahnestock, S.R., Termonia, Y. & Gardner, K.C.H. (1998) Adv. Mat. 10, 1185–1195. 38. Arnott, S., Dover, D.S. & Elliott, A. (1967) J. Mol. Biol. 30, 201–208. 39. Guess, K.B. & Viney, C. (1998) Thermochim. Acta 315, 61–66. 40. Knight, D.P., Knight, M.M. & Vollrath, F. (2000) Int. J. Biol. Macromol. 27, 205–210. 41. Sirichaisit, J., Young, R.J. & Vollrath, F. (2000) Polymer 41, 1223–1227. 42. Smith, B.L., Schaffer, T.E., Viani, M., Thompson, J.B., Frederick, N.A., Kindt, J., Belcher, A., Stucky, G.D., Morse, D.E. & Hansma, P.K. (1999) Nature (London) 399, 761–763.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
MOLECULAR DYNAMICS ANALYSIS OF A BUCKYBALL-ANTIBODY COMPLEX
6466
Colloquium Molecular dynamics analysis of a buckyball-antibody complex
William H.Noon*, Yifei Kong†, and Jianpeng Ma*†‡§ *Department of Bioengineering, Rice University, 6100 Main, MS-142, Houston, TX 77005; and †Graduate Program of Structural and Computational Biology and Molecular Biophysics, and ‡Verna and Marrs McLean Department of Biochemistry and Molecular Biology, Baylor College of Medicine, One Baylor Plaza, BCM-125, Houston, TX 77030 Edited by William N.Lipscomb, Harvard University, Cambridge, MA, and approved November 28, 2001 (received for review October 5, 2001) This is a multinanosecond molecular dynamics study of a bio-nano complex formed by a carbon nanoparticle, a buckyball C60, and a biological molecule, an antibody, with high binding affinity and specificity. In the simulation, the ball is completely desolvated by the binding site of the antibody by means of a nearly perfect shape complementarity and extensive side-chain interactions, with the exception that about 17% of the surface is persistently exposed to solvent and could be used for functional derivatization. The interactions are predominantly hydrophobic, but significant polar interactions occur as well. There exists a rich body of various πstacking interactions. Aromatic side chains are involved in both double and triple stackings with the ball. Some ionic side chains, such as the guanidinium group of arginine, also form π-stackings with the ball. The results suggest that π-stackings are very efficient and common modes of biological recognition of π-electron-rich carbon nanoparticles. Most importantly, the results demonstrate that, in general, an ordinary protein binding site, such as that of an antibody, can readily bind to a carbon nanoparticle with high affinity and specificity through recognition modes that are common in protein-ligand recognition. molecular dynamics simulation | molecular recognition | bio-nano conjugate | π-stackings In 1985, a third allotropic form of carbon was discovered (1). The molecule was named Buckministerfullerene, commonly known as the buckyball, because of its geodesic structure (2). Six years later, the fullerene family was expanded with the discovery of nanotubes (3). Because of the unique structural properties associated with these molecules (4), there is great interest in using them in real-world applications (5–9), including integrating nanoparticles into biological systems, a fast-emerging field known as bio-nanotechnology. Examples of potential applications in bio-nanotechnology are transporting devices for drug delivery (10, 11), carriers of radioactive agents for biomedical imaging (12, 13), and templates for designing pharmaceutical agents, such as HIV type 1 protease inhibitors (8, 9), antioxidant (14–16), chemotactic agents (17), and neuroprotectants (18). However, to introduce artificial nanomaterials into living cells, one must deal with issues such as water solubility, biocompatibility, and biodegradability. This requires a comprehensive understanding of the interactions of nanomaterials with biological molecules such as proteins, nucleic acids, membrane lipids, and even water molecules. As in the studies of protein interactions (19, 20), computer simulation techniques are very useful to investigate the interfacial properties of bio-nano systems, especially the dynamic, thermodynamic, and mechanical properties, at different spatial and temporal resolutions (21, 22). One particularly interesting subject is the study of the interactions of nanoparticles within the binding sites of proteins, and optimizing the interactions for improved bio-nano recognition. Recent biochemical and structural studies reveal the existence of certain natural proteins that can recognize specific nanoparticles (23, 24). One such example is an antibody that was selected from the mouse immune repertoire to specifically recognize derivatized C60 fullerenes (23, 24) and had a binding affinity of ≈25 nM (23). The crystal structure of the Fab fragment of this antibody has been determined (23) (Fig. 1). Although the fullerene-antibody complex structure is not available, it was speculated that the fullerene-binding site is formed at the interface of the antibody light and heavy chains lined with a cluster of shape-complementary hydrophobic amino acid residues (23). The covalent modifications of the functionalized buckyball used in the experiments for solubility purpose occupy only a small fraction of the ball surface (see figure 1. in ref. 24); therefore, the unoccupied surface area would be large enough to interact with the antibody. To understand the detailed interactions of the buckyball in the binding site of the antibody, we study the buckyball-antibody complex by using molecular dynamics simulation (19). The purpose of our computational modeling/simulation is to identify the energetically favorable binding modes between the antibody and the buckyball. These results will be useful in developing new biologically compatible fullerene molecules. METHODS We performed molecular dynamics simulations of a buckyball-antibody complex. The initial coordinates of the antibody were available from the Protein Data Bank (PDB ID code 1EMT) (23). The coordinates of the buckyball (C60) were provided by Richard E.Smalley at Rice University (Houston). Although original biochemical experiments were done on a derivatized buckyball for solubility reasons (24), we omitted the derivatizations in our simulation and focused on the ball-protein interactions. In this particular case, it seems reasonable to assume, as a first approximation, that the hydrophilic derivatizations on the ball do not play a critical role in the predominantly hydrophobic interactions between the ball and the antibody. Because all of the derivatizations were attached to the balls by two neighboring carbon atoms, we argue that the electronic structures of the derivatized balls, at the least the aromaticity, may not be disturbed dramatically at the opposite face, where the ball contacts the antibody. Because the original x-ray structure of the antibody does not contain the buckyball substrate, we docked the ball into the
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC. This paper was submitted directly (Track II) to the PNAS office. Abbreviations: SBMD, stochastic boundary molecular dynamics; VL, variable region light chain; VH, variable region heavy chain. §To whom reprint requests should be addressed. E-mail:
[email protected].
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
MOLECULAR DYNAMICS ANALYSIS OF A BUCKYBALL-ANTIBODY COMPLEX
6467
suggested binding site (23). We then performed 200 steps of minimization, using the steepest descent method and 300 steps of minimization using the Adapted Basis Newton Raphson method. To reduce the necessary simulation time, the stochastic boundary molecular dynamics (SBMD) method was used (see ref. 25 for details). This is a highly efficient method for simulating the localized interactions in the active site of a protein as exemplified in a recent study of enzyme catalytic mechanism (26). The CHARMM program (27) was used for the simulation. Polarhydrogen potential function (PARAM19) (28) was used for the protein and a modified TIP3P water model (29) was used for the solvent. Atomic partial charges for the buckyball were set to neutral (30) and the van der Waals parameters of its atoms were the same as an aromatic carbon atom (31) carried in the CHARMM force field (28). The system was separated into a reaction zone and a reservoir region, and the reaction zone was further divided into a reaction region and a buffer region (25). The reference point for partitioning the system in SBMD was chosen to be near the center of the buckyball. The reaction region around the active site was a sphere of radius r of 14 Å, the buffer region of 14 Å
16 Å; all atoms in the reservoir region were deleted. The simulation system, shown in Fig. 2, consisted of 106 protein residues, a buckyball C60, and 166 water molecules. Atoms inside the reaction region were propagated by molecular dynamics, whereas atoms in the buffer region were propagated by the Langevin dynamics. Atoms inside the buffer region were retained by harmonic restoring forces with constants derived from the temperature factors in the crystal structure. Water molecules were confined to the active-site region by a deformable boundary potential (32). The friction constant in the Langevin dynamics was 250 ps−1 for protein atoms and 62 ps−1 for water molecules. During the simulation, all bonds with hydrogen atoms were fixed by using the SHAKE algorithm (33). A 1-fs time step was used for integrating the equations of motion during the molecular dynamics simulation, whereas initial random velocities were sampled from the Boltzmann distribution (34). The system was equilibrated for 50 ps at 300 K, and was followed by a 5-ns production run.
Fig. 1. Ribbon diagram of the crystal structure of the Fab fragment of the fullerene-specific antibody (ref. 23; PDB ID code, 1EMT). The two polypeptide chains, variable region heavy chain (VH) and light chain (VL), are marked. The suggested binding site of the buckyball substrate is indicated by the circle. The figure is made by using software MOLSCRIPT (39) and rendered by RASTER3D (40).
Fig. 2. The molecular dynamics simulation system with the stochastic boundary condition. It contains 106 protein residues (ribbon), the buckyball (space-filling model in yellow), and 166 water molecules (ball-and-stick).
As an approximation, the simulated buckyball was treated as a nonpolarizable hydrophobic entity. To a first approximation, this treatment is reasonable based on the experimental observation that an unmodified buckyball is insoluble in water. The overwhelmingly large number of hydrophobic interactions in the binding site also justifies such a treatment. We also simulated the systems containing the whole antibody molecule submerged in a large periodic water box with and without the presence of the buckyball in the binding site for a shorter period (200 ps). The results were compared with those from the SBMD simulation. RESULTS We first observed during the 5-ns simulation that a single buckyball C60 molecule can be readily accommodated in the suggested binding site of the antibody. The ball inside the binding site undergoes a small relative translational motion, but a significant rotational motion. Further analysis of the angular momentum reveals no favored axis of rotation. The ball is nearly rigid, therefore the deformational motion of the ball is negligible. About 17% of the surface of the ball is exposed to solvent throughout the simulation, with the antibody covering the remaining surface. Fig. 3 shows the exposed surface area as a function of time in a 5-ns simulation window. The persistently
Fig. 3. Solvent-exposed surface area as a function of time in a 5-ns simulation window. The average value is about 17%.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
MOLECULAR DYNAMICS ANALYSIS OF A BUCKYBALL-ANTIBODY COMPLEX
Fig. 4. (a) Stereo pair snapshot of the buckyball inside the binding site of the antibody. Some key protein side chains are explicitly drawn and the rest of the protein matrix is given in a dotted surface representation. The view is from the top in Fig. 1. (b) Stereo pair of a triple πstacking. A piece of the buckyball and the side chains of Trp-33 (VH) and Tyr-52 (VH) are shown. (c) Stereo pair of stacking interactions made by Trp-47 (VH) and Phe-96 (VL). The Hε atom of Trp-47 (VH) points directly toward the ball, which can induce a weak hydrogen bond with the π-electron of the ball. Phe-96 (VL) is in stacking with the ball as well. (d) Stereo pair of a different configuration of the interactions in c. The side chain motions bring the guanidinium group of Arg-50 (VH) to a triple π-stacking with Trp-47 (VH) and the ball. The side chain of Phe-96 (VL) moves aside, but remains in stacking with the ball.
6468
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
MOLECULAR DYNAMICS ANALYSIS OF A BUCKYBALL-ANTIBODY COMPLEX
6469
solvent-exposed ball surface could be used for additional functional derivatization (24). Although some of the ball-antibody interactions had been suggested from an earlier docking study (23), our results from molecular dynamics simulation are more thorough and reliable. Fig. 4a is a snapshot of the ball in the binding site, surrounded predominantly by hydrophobic amino acid side chains. Some of the important side chains of the antibody are explicitly shown, and the rest of the protein matrix is represented by a dotted surface. Of particular interest is the presence of rich π-interactions between the ball and the aromatic side chains of the antibody. Phe-96 (VL), Tyr-49 (VL), and Tyr-91 (VL) residues all form π-stacking arrangements with the ball. A three-tiered π-stacking interaction is observed between the ball, Tyr-52 (VH), and Trp-33 (VH) (Fig. 4b). Another interesting interaction arises from the motion of the side chains of Trp-47 (VH) and Arg-50 (VH). Two distinct interaction modes made by these side chains were observed. Fig. 4c shows that Phe-96 (VL) is π-stacking with the ball while the Hε atom of Trp-47 (VH) points toward the ball, which induces a weak hydrogen bond with the rich π-electrons of the ball. In Fig. 4d, however, a rotation of the side chain of Trp-47 (VH) results in a triple π-stacking between the ball, the guanidinium group of Arg-50 (VH), and the side chain of Trp-47 (VH). In this case, the side chain of Phe-96 (VL), though moved aside, remains in π-stacking with the ball. Similar π-stacking interactions have been observed in different antibody-antigen complexes (35) and other proteins (36). The interaction modes of aromatic side chains with the buckyball are also remarkably similar to those observed in the x-ray structure of a buckyball cocrystallized with benzene molecules (37). Moreover, in a recent experimental study (38), π-stacking was found to be very effective in noncovalently immobilizing functional groups on nanotubes. These results indicate that π-stacking is indeed a very efficient and common mode for biological recognition of π-electron-rich carbon nanoparticles. In addition to the π-stacking interactions, the complementary shape of the antibody pocket also plays an important role in recognition. The interface between the ball and the antibody-binding pocket is nearly seamless and completely desolvated. Several hydrogen-bonded side chains and van der Waals contacts contribute to the formation of the complementary pocket. Trp-103 (VH) lies at the ball-antibody interface, but is not oriented in a manner expected for π-stacking; it is hydrogen bonded to another interfacial residue, Tyr-36 (VL). This hydrogen-bonding network extends to residues Asn-34 (VL) and Gln-89 (VL), both of which make contacts with the ball. Other residues that are in van der Waals contacts with the ball are Leu-46 (VL), Ala-97 (VH), and Ala-101 (VH). For comparison, we also ran simulations of the systems containing the whole antibody molecule submerged in a large periodic water box with and without the buckyball in the binding site for a shorter period (200 ps). In the absence of the buckyball, we observed a big vacuum bubble around the hydrophobic binding site of the antibody. When the buckyball is present inside the binding site, the observed interactions are qualitatively similar to those from the SBMD simulation. It is worth pointing out that no electronic polarizability effect of the ball was included in our simulations. The observed rich π-stacking interactions arise primarily from the shape complementarity between the stacking aromatic side chains and the buckyball, which is a prerequisite for the stacking interactions. CONCLUDING DISCUSSION We have conducted a molecular dynamics study of a bio-nano complex formed by a carbon nanoparticle, a buckyball C60, and a biological molecule, an antibody with a high binding affinity and specificity. The results agree well with known biochemical (24) and structural (23) data of the system. The simulation shows that the high binding affinity and specificity are achieved through complementary shape and extensive side-chain interactions, including a set of rich π-stacking interactions. This finding also suggests that π-stacking is a very efficient and common mode for biological recognition of π-electron-rich carbon nanoparticles. It is notable that, in addition to tight binding, there is about 17% of the surface of the ball persistently exposed to the solvent. This amount of exposure may leave enough room for further manipulation of biocompatible buckyballs. Finally, the simulation results demonstrate that, in general, an ordinary protein binding site, such as that of an antibody, can readily bind to a carbon nanoparticle with high affinity and specificity through recognition modes that are common in proteinligand recognition. A dynamic animation of the motion of the ball inside the binding site of the antibody can be found as Movie 1, which is published as supporting information on the PNAS web site, www.pnas.org. We thank Prof. Richard E.Smalley and Dr. Kevin Ausman for their encouragement on the project and for providing us the atomic coordinates of the buckyball C60. We also thank Prof. Robert Hauge for helpful discussions. This work is supported by grants from National Science Foundation, American Heart Association, and The Robert A.Welch Foundation (to J.M.). 1. Kroto, H.W., Heath, J.R., O'Brien, S.C., Curl, R.F. & Smalley, R.E. (1985) Nature (London) 318, 162–163. 2. Yakobson, B.I. & Smalley, R.E. (1997) Am. Sci. 85, 324–337. 3. Iijima, S. (1991) Nature (London) 354, 56–58. 4. Dresselhaus, M.S., Dresselhaus, G. & Eklund, P.C. (1996) Science of Fullerenes and Carbon Nanotubes (Academic, San Diego). 5. Collins, P.G., Arnold, M.S. & Avouris, P. (2001) Science 292, 706–709. 6. Balavoine, F., Schultz, C., Richard, C., Mallouh, V., Ebbesen, T.W. & Mioskowski, C. (1999) Angew. Chem. Int. Ed. 38, 1912–1915. 7. Da Ros, T. & Prato, M. (1999) J. Chem. Soc. Chem. Commun., 663–669. 8. Friedman, S.H., DeCamp, D.L., Sijbesma, R.P., Srdanov, G., Wudl, F. & Kenyon, G.L. (1993) J. Am. Chem. Soc. 115, 6506–6509. 9. Sijbesma, R., Srdanov, G., Wudl, F., Castoro, J.A., Wilkins, C., Friedman, S.H., DeCamp, D.L. & Kenyon, G.L. (1993) J. Am. Chem. Soc. 115, 6510– 6512. 10. Wilson, L.J., Cagle, D.W., Thrash, T.P., Kennel, S.J., Mirzadeh, S., Alford, J.M. & Ehrhardt, G.J. (1999) Coord. Chem. Rev. 190, 199–207. 11. Wilson, L.J. (1999) Interface 8, 24–28. 12. Cagle, D.W., Kennel, S.J., Mirzadeh, S., Alford, J.M. & Wilson, L.J. (1999) Proc. Natl. Acad. Sci. USA 96, 5182–5187. 13. Wharton, T., Alford, J.M., Husebo, L.O. & Wilson, L.J. (2000) in Recent Advances in the Chemistry and Physics of Fullerenes and Related Materials (Electrochem. Soc., Pennington, NJ), Vol. 9, pp. 258–266. 14. Straface, E., Natalini, B., Monti, D., Franceschi, C., Schettini, G., Bisaglia, M., Fumelli, C., Pincelli, C., Pellicciari, R. & Malorni, W. (1999) FEBS Lett. 454, 335–340. 15. Lin, A.M., Chyi, B.Y., Wang, S.D., Yu, H.H., Kanakamma, P.P., Luh, T.Y., Chou, C.K. & Ho, L.T. (1999) J. Neurochem. 72, 1634–1640. 16. Chueh, S.C., Lai, M.K., Lee, M.S., Chiang, L.Y., Ho, T.I. & Chen, S.C. (1999) Transplant Proc. 31, 1976–1977. 17. Toniolo, C., Bianco, A., Maggini, M., Scorrano, G., Prato, M., Marastoni, M., Tomatis, R., Spisani, S., Palu, G. & Blair, E.D. (1994) J. Med. Chem. 37, 4558–4562. 18. Dugan, L.L., Turetsky, D.M., Du, C., Lobner, D., Wheeler, M., Almli, C.R., Shen, C.K., Luh, T.Y., Choi, D.W. & Lin, T.S. (1997) Proc. Natl. Acad. Sci. USA 94, 9434–9439. 19. Brooks, C.L., III, Karplus, M. & Pettitt, B.M. (1988) Adv. Chem. Phys. 71, 1–249. 20. McCammon, J.A. & Harvey, S. (1987) Dynamics of Proteins and Nucleic Acids (Cambridge Univ. Press, Cambridge, U.K.). 21. Xu, C. & Scuseria, G.E. (1994) Phys. Rev. Lett. 72, 669–672. 22. Yakobson, B.I., Brabec, C.J. & Bernholc, J. (1996) J. Comp. Aided Mater. Des. 3, 173–182. 23. Braden, B.C., Goldbaum, F.A., Chen, B.X., Kirschner, A.N., Wilson, S.R. & Erlanger, B.F. (2000) Proc. Natl. Acad. Sci. USA 97, 12193–12197.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
MOLECULAR DYNAMICS ANALYSIS OF A BUCKYBALL-ANTIBODY COMPLEX
24. Chen, B.X., Wilson, S.R., Das, M., Coughlin, D.J. & Erlanger, B.F. (1998) Proc. Natl. Acad. Sci. USA 95, 10809–10813. 25. Brooks, C.L., III, & Karplus, M. (1989) J. Mol. Biol. 208, 159–181. 26. Ma, J., Zheng, X., Schnappauf, G., Braus, G., Karplus, M. & Lipscomb, W.N. (1998) Proc. Natl. Acad. Sci. USA 95, 14640–14645. 27. Brooks, B.R., Bruccoleri, R.E., Olafson, B.D., States, D.J., Swaminathan, S. & Karplus, M. (1983) J. Comput. Chem. 4, 187–217. 28. Neria, E., Fischer, S. & Karplus, M. (1996) J. Chem. Phys. 105, 1902– 1921. 29. Jorgensen, W.L. (1981) J. Am. Chem. Soc. 103, 335–340. 30. Andreoni, W. (1998) Annu. Rev. Phys. Chem. 49, 405–439. 31. Klein, D.J., Schmalz, T.G., Hite, G.E. & Seitz, W.A. (1986) J. Am. Chem. Soc. 108, 1301–1302. 32. Brooks, C.L., III, & Karplus, M. (1983) J. Chem. Phys. 79, 6312–6325. 33. Ryckaert, J.P., Ciccotti, G. & Berendsen, H.J.C. (1977) J. Comput. Phys. 23, 327–341. 34. Allen, M.P. & Tildesley, D.J. (1980) Computer Simulation of Liquids (Clarendon, Oxford). 35. Braden, B.C., Souchon, H., Eisele, J.L., Bentley, G.A., Bhat, T.N., Navaza, J. & Poljak, R.J. (1994) J. Mol. Biol. 243, 767–781. 36. Hu, G., Gershon, P.D., Hodel, A.E. & Quiocho, F.A. (1999) Proc. Natl. Acad. Sci. USA 96, 7149–7154. 37. Meidine, M.F., Hitchcock, P.B., Kroto, H.W., Taylor, R. & Walton, D.R.M. (1992) J. Chem. Soc. Chem. Commun., 1534–1537. 38. Chen, R.J., Zhang, Y., Wang, D. & Dai, H. (2001) J. Am. Chem. Soc. 102, 3838–3839. 39. Kraulis, P.J. (1991) J. Appl Crystallogr. 24, 946–950. 40. Bacon, D.J. & Anderson, W.F. (1988) J. Mol. Graphics 6, 219–220.
6470
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
H3PW12O40-FUNCTIONALIZED TIP FOR SCANNING TUNNELING MICROSCOPY
6471
Colloquium H3PW12O40-functionalized tip for scanning tunneling microscopy
In K.Song*†, John R.Kitchin*, and Mark A.Barteau*‡ *Center for Catalytic Science and Technology, Department of Chemical Engineering, University of Delaware, Newark, DE 19716 Edited by Jack Halpern, University of Chicago, Chicago, IL, and approved February 7, 2002 (received for review September 28, 2001) Recent reports of C60-functionalized metal tips [Kelly, K.F., Sarkar, D., Hale, G.D., Oldenburg, S.J. & Halas, N.J. (1996) Science 273, 1371–1373] and carbon nanotube tips [Dai, H., Hafner, J.H., Rinzler, A.G., Colbert, D.T. & Smalley, R.E. (1996) Nature (London) 384, 147–151] demonstrate the potential of controlling the chemical identity and geometric structure of tip atoms in scanning tunneling microscopy (STM). This work reports the performance of a heteropolyacid (HPA)-functionalized Pt/Ir tip, which was formulated by contacting a mechanically formed tip with a solution of H3PW12O40 molecules. Attachment of an H3PW12O40 molecule on the metal tip was confirmed by observing the characteristic negative differential resistance (NDR) behavior of H3PW12O40 in tunneling spectroscopy. Atomic resolution images of bare graphite as well as of H6P2W18O62 HPA monolayers on graphite were successfully obtained with a Pt/ Ir-HPA tip. In the H3PW12O40 molecule on a metal tip, it is likely that a terminal oxygen of W=O (an oxygen species projecting outward from the pseudospherical H3PW12O40 molecule) serves as an atomically sharp and stable tip. Additionally, superimposed superperiodic structures commensurate with the underlying graphite lattice were regularly observed with the modified tips. This result suggests that tip functionalization with these metal oxide molecules may enhance resolution in a fashion analogous to functionalization with C60. STM tip | functionalization | heteropolyacid | terminal oxo group | atomically sharp probe Atomically sharp and stable tips are of great importance in scanning tunneling microscopy (STM). Many ways to produce sharp tips of various materials for STM have been developed, including electrolytic (electrochemical) polishing/etching, chemical polishing/etching, ion milling, cathode sputtering, whisker growth, electron-beam deposition, flame polishing, mechanical sharpening, cutting, machining, fragmenting, and so on (1). However, none of these production methods is universally applicable to all materials of interest. It is emphasized that the critical components of high-resolution tunneling tips are actually minitips with radii less than 100 Å, the structure of which is much more difficult to control (2). Minitips occur frequently as a result of the variability of tip preparation techniques, but the origin of these structures is not clear and there are no truly reproducible means of generating them. The shape and arrangement of the minitips are important factors that can affect the stability of the image (3). If several atoms or small clusters of atoms at the end of the tip act independently, the image will actually be a superposition of images. By positioning graphite-covered W tips consisting of four or fewer independent atoms in different arrangements and different orientations with respect to the graphite, for example, it has been shown that a variety of different periodic images of graphite can be produced (4). The chemical identity of the tip atom is also important in STM imaging. A simulation study on the imaging of graphite with various single-atom tips, such as Na and Ca, has shown that the graphite image is very sensitive to the identity of the tip atom, and that the corrugation amplitude is much smaller than expected (5). Heteropolyacids (HPAs) are early-transition metal oxygen anion clusters that exhibit a wide range of sizes, compositions, and molecular architectures (6). One of the great advantages of HPAs is their well-defined molecular structure and tunable redox properties (7). A recent STM study of HPAs has shown that these inorganic molecules form two-dimensional ordered arrays on graphite surfaces and exhibit distinctive current-voltage (I-V) behavior, referred to as negative differential resistance (NDR) in their tunneling spectra (8). NDR peak positions observed for pure HPA arrays not only serve as fingerprints of the HPAs, but also correlate well with the reduction potentials of HPAs (9). In this work we attempted to chemically modify a mechanically formed Pt/Ir tip by attaching HPA molecules to functionalize the Pt/Ir tip as an STM probe. Among various HPA structural classes, Keggin-type (10) H3PW12O40 was chosen for its structural simplicity. Fig. 1A shows the molecular structure of the soccer ball-like [PW12O40]3− heteropolyanion, constructed from the x-ray diffraction (XRD) data (11). In this representation, oxygen atoms are represented as spheres. The structure of Keggin-type heteropolyanion, [PW12O40]3−, consists of a heteroatom, P, at the center of the anion cluster, tetrahedrally coordinated to four oxygen atoms. This tetrahedron is surrounded by four groups of three edgesharing WO6 octahedra, and the groups are interconnected by corner-sharing. A sphere at the outermost vertex of each WO6 octahedron represents a terminal oxo species. The W=O terminal group has a bond length of 1.7 Å. A single H3PW12O40 molecule has twelve W=O terminal bonds at its outermost surface, depicted schematically in Fig. 1B. The molecular size of H3PW12O40 is 11≈12 Å as determined by XRD (11) and STM (12). The performance of Pt/Ir-H3PW12O40 tips was demonstrated by probing bare graphite as well as Wells-Dawson-type (13) H6P2W18O62 HPA monolayers on graphite. MATERIALS AND METHODS Preparation of Pt/Ir-H3PW12O40 Tip. H3PW12O40 was purchased from Aldrich Chem (Metuchen, NJ), and H6P2W18O62 was synthesized according to published methods (14). About 0.01 M
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC. This paper was submitted directly (Track II) to the PNAS office. Abbreviations: STM, scanning tunneling microscopy; HPA, heteropolyacid; NDR, negative differential resistance. †Present address: Department of Industrial Chemistry, Kangnung National University, Kangnung 210–702, Korea. ‡To whom correspondence should be addressed. E-mail: [email protected].
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
H3PW12O40-FUNCTIONALIZED TIP FOR SCANNING TUNNELING MICROSCOPY
6472
aqueous solutions of each sample were prepared. A mechanically formed Pt/Ir (90/10) tip was contacted with a drop of the H3PW12O40 solution, and then the tip was allowed to dry in air for 1 h at room temperature in order for H3PW12O40 molecules to be attached on the tip. The tip was maintained in a downward orientation during the entire processes.
Fig. 1. (A) Molecular structure of the pseudospherical Keggin-type [PW12O40]3− heteropolyanion. Oxygen atoms are represented as spheres. A sphere projecting outward in each WO6 octahedron represents terminal oxygen species. (B) A simplified representation of the W=O projection.
STM Imaging and Tunneling Spectroscopy. STM images of highly oriented pyrolytic graphite were obtained with Pt/Ir-H3PW12O40 tips in air by using a Topometrix TMX 2010 Discoverer scanning tunneling microscope. Scanning was done in the constant current mode at a positive sample bias of 100 mV and a tunneling current of 1≈1.5 nA. All STM images presented in this report are unfiltered, and the reported periodicities (lattice constants) represent average values determined by performing two-dimensional fast Fourier transform analyses on at least three images for each sample. Tunneling spectra were also obtained with Pt/Ir-H3PW12O40 tips in air on a graphite surface. Both Topometrix TMX 2010 and LK Technologies LK-1000 scanning tunneling microscopes were used to confirm consistency and reproducibility of tunneling spectra. To measure a tunneling spectrum, the sample bias was ramped from −2 to +2 V with respect to the tip and the tunneling current was monitored. Attachment of an H3PW12O40 molecule on the Pt/Ir tip was confirmed by observation of its characteristic NDR behavior in tunneling Spectroscopy measurements before STM imaging. Several tunneling spectra were measured on the graphite surface with the Pt/IrH3PW12O40 tip to ensure the stability of the tip and the reproducibility of the tunneling spectra. The Pt/Ir-H3PW12O40 tip was also used as an STM probe in imaging H6P2W18O62 molecules on graphite to test its ability to probe inorganic monolayers. For this purpose, a drop of 0.01 M aqueous solution of H6P2W18O62 sample was deposited on a graphite surface and allowed to dry for 1 h at room temperature before imaging. RESULTS AND DISCUSSION Confirmation of H3PW12O40 Attachment. Fig. 2A shows the typical I-V spectrum obtained on bare graphite using a Pt/Ir-H3PW12O40 tip. The voltage resolution of our STM instruments is 0.05 V. The spectrum did not represent a typical I-V response of graphite, but instead showed NDR behavior, which was also observed consistently for H3PW12O40 deposited on graphite and probed by a bare metal tip (12). This result indicates that H3PW12O40 molecules were attached to the Pt/Ir tip and that electrons tunneled through these molecules. The NDR peak position measured on bare graphite with the Pt/Ir-H3PW12O40 tip is also comparable to that of a H3PW12O40 monolayer on graphite probed by a normal Pt/Ir tip. Fig. 2 B and C show the distribution of NDR peak voltages of H3PW12O40 measured for each case (NDR peak voltage was defined as the voltage at which the local maximum current was observed in the tunneling spectrum). In the normal case (H3PW12O40 monolayer probed with a bare Pt/Ir tip), the most frequent NDR peak voltage measured was found to be −1.20 V, and the statistical average of NDR peak voltages was −1.14±0.09 V. In the case of Pt/Ir-H3PW12O40 system, the modified tip exhibited the most frequent NDR peak at −1.15 V with a statistical average of −1.15±0.07 V. The values observed in both cases are nearly indistinguishable, indicating the presence of H3PW12O40 on the modified tip in the latter case.
Fig. 2. (A) A typical I-V spectrum obtained on bare graphite by using a Pt/Ir-H3PW12O40 tip. (B) Distribution of NDR peak voltages of H3PW12O40 on graphite obtained with a normal Pt/Ir tip (total number of tunneling spectra=24). (C) Distribution of NDR peak voltages measured on bare graphite with a Pt/Ir-H3PW12O40 tip (total number of tunneling spectra=62).
Atomically Sharp Pt/Ir-H3PW12O40 Probe. Controlling the number and orientation of H3PW12O40 molecules attached on a Pt/Ir tip is still challenging. Despite this difficulty, atomic resolution images of graphite were frequently obtained with a Pt/Ir-H3PW12O40 tip. The atomic resolution suggests that electron tunneling occurs through a sharp feature of the H3PW12O40 molecule on the Pt/Ir tip, presumably a W=O terminal bond. Considering the molecular size of H3PW12O40 and its geometric structure shown in Fig. 1, a terminal oxo group projecting outward from the pseudospherical polyanion can possibly serve as an atomically sharp probe. It has been demonstrated that a single electronegative adatom, such as sulfur or oxygen, on a W tip contributes more to the filled density of states on the tip than to the empty states (15). Atomic resolution images are obtained when electrons flow primarily from the single adatom on top of the metal tip (the occupied density of states) to the sample (the empty density of states). However, no atomic resolution image is obtained for the opposite bias, when electrons tunnel from a sample (the occupied density of states) to a large number of tip-metal surface atoms (the empty density of states), because the electronegative ada
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
H3PW12O40-FUNCTIONALIZED TIP FOR SCANNING TUNNELING MICROSCOPY
6473
tom contributes only very weakly to the unoccupied density of states (15). In imaging single adsorbed species on metal surfaces, it was also demonstrated that spatial resolution was increased in topographical scans by attaching a C=O molecule on the tip (oxygen atom is projecting outward on attachment) (16). Our STM imaging condition and H3PW12O40-functionalized tip meet the above requirements for obtaining atomic resolution images; electrons flow from tip to sample in the imaging mode, and the Pt/Ir-H3PW12O40 tip has an electronegative oxygen end atom chemically bound to tungsten. Moreover, the rigid and well-defined molecular structure of H3PW12O40 provides no need to control the position of W=O, which makes Pt/Ir-H3PW12O40 a highly stable and atomically sharp probe for STM imaging.
Fig. 3. (A) Atomic resolution STM image of graphite obtained with a Pt/Ir-H3PW12O40 tip. (B) STM image of H6P2W18O62 monolayer on graphite obtained with a Pt/Ir-H3PW12O40 tip. (C) Schematic representation of unit cell of the H6P2W18O62 array. (D) Molecular structure of the ellipsoidal Wells-Dawson-type [P2W18O62]6− heteropolyanion. The polyanion consists of two defect-Keggin-type fragments, [PW9O34]9−. Each fragment consists of a central PO4 tetrahedron sharing corners with nine WO6 octahedra—the octahedra are somewhat distorted from an ideal octahedron. Three WO6 octahedra form a compact group by sharing edges, whereas the remaining six octahedra in each of the [PW9O34]9 − fragments form a zigzag ring by alternately sharing edges and corners. The two fragments are linked by six nearly linear W—O—W bonds.
The performance of Pt/Ir-H3PW12O40 as an atomically sharp and stable tip was demonstrated by probing bare graphite as well as WellsDawson-type (13) H6P2W18O62 HPA monolayers on graphite. Fig. 3A shows the atomic resolution image of graphite obtained with a Pt/IrH3PW12O40 tip. This image clearly shows the hexagonal symmetry of graphite with a periodicity of 2.46 Å. Fig. 3B shows a molecularresolution STM image of a H6P2W18O62 monolayer on graphite probed by a Pt/Ir-H3PW12O40 tip. This image clearly shows a two-dimensional well-ordered array with rugby-ball-like (ellipsoidal) features. The unit cell of H6P2W18O62 arrays constructed on the basis of lattice constants determined from two-dimensional fast Fourier transform shows that the arrays have the primitive unit cell (rhombus) and the conventional unit cell (centered oblique rectangle), shown in Fig. 3C. The primitive cell has sides of 14.6 Å with an included angle of 35.2°, and the conventional unit cell has a minor axis of 10.4 Å. The molecular structure of the [P2W18O62]6− heteropolyanion constructed from the x-ray diffraction data (14) is shown in Fig. 3D. This molecule consists of two defect-Keggin-type [PW9O34]9− fragments, and has a rugby-ball-like (ellipsoidal) shape with dimensions of 11 Å×14.5 Å. This level of agreement between molecular dimensions and the periodicity of heteropolyanion monolayer is comparable to that obtained with metal tips (12). Thus, a well-formulated Pt/Ir-H3PW12O40 tip can provide high-resolution images; the tunneling tip is of atomic scale rather than the molecular scale (1 nm) of the HPA.
Fig. 4. (A–C) A set of unusual STM images of graphite obtained with a Pt/Ir-H3PW12O40 tip with varying scan size. (D) A schematic represents an azimuthal angle representation of unit cells of the superperiodic structure and underlying graphite arrays for = 27°. between lattice vectors, a1 and b1. Superperiodic lattice vectors can be expressed in terms of the graphite lattice vectors by b1=18a1+15a2 and b2= −15a1+33a2 for =27°, and b1=15a1 +18a2 and b2=−18a1+33a2 for =33°.
Superimposed Images of Superperiodic Structures and Underlying Graphite. Fig. 4 shows a set of unusual STM images of graphite obtained with a Pt/Ir-H3PW12O40 tip in air with varying scan sizes. Fig. 4A clearly shows well-ordered hexagonal superperiodic features with a periodicity of 70.4 Å. The periodicity and orientation of the superperiodic structures were consistent, regardless of the variation of scanning parameters such as set point current, scan angle (rotation), and scan rate. More importantly, the superperiodic and real-size graphite structures were probed simultaneously by decreasing the scan size (Fig. 4B). Even in a very small scan area (Fig. 4C), we could observe bright domains representing the superperiodic structure. The simultaneous observation of superimposed images of superperiodic structures and real-size graphite obtained with a Pt/Ir-H3PW12O40 tip in air is quite interesting. One more important feature of the superimposed images
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
H3PW12O40-FUNCTIONALIZED TIP FOR SCANNING TUNNELING MICROSCOPY
6474
shown in Fig. 4 is that both the superperiodic structure and real-size graphite have hexagonal symmetry, but the orientations of the two are rotated with respect to each other. Fig. 4D shows the schematic representation of unit cells of the superperiodic structure and underlying graphite. The superperiodic structure has hexagonal symmetry (β=60°) with lattice constants of b1= b2=70.4 Å. The underlying graphite also has hexagonal symmetry (α=60°) with lattice constants of a1=a2=2.46 Å. Because both lattices have hexagonal symmetry, two reciprocal array representations can be built within the angle of 0° to 60°. The measured azimuthal angle ( ), between two lattice vectors, a1 and b1, can be either 27° or 33°. When applying the experimentally determined lattice parameters to the EPICALC software developed at the University of Minnesota (17), we obtained a dimensionless potential (V/V0) value of 0.5, indicating that the superperiodic structure is commensurate with the underlying hexagonal graphite.
Fig. 5. Another set of unusual STM images of graphite obtained with a Pt/Ir-H3PW12O40 tip with varying scan size, showing a superimposed hexagonal superperiodic structure (β=60°, b1=b2=14.97 Å) and underlying real-size graphite (α=60°, a1=a2=2.46 Å). is either 25.3° or 34.7°. Superperiodic lattice vectors can be expressed in terms of the graphite lattice vectors by b1=4a1+3a2 and b2=−3a1+7a2 for =25.3°, and b1=3a1+ 4a2 and b2=−4a1+7a2 for =34.7°.
Fig. 5 also shows another set of unusual STM images of graphite obtained with a Pt/Ir-H3PW12O40 tip, showing superimposed images of a well-ordered hexagonal superperiodic structure (β=60°, b1=b2=14.97 Å) and underlying real-size graphite (α=60°, a1=a2=2.46 Å), with array rotation with respect to each lattice ( =25.3° or 34.7°). These lattice constants also produced the dimensionless potential value of 0.5 by the EPICALC simulation, indicating that the superperiodic structure is again commensurate with the underlying graphite. As demonstrated in Figs. 4 and 5, hexagonal superperiodic structures commensurate with underlying graphite were observed in imaging graphite with Pt/Ir-H3PW12O40 tips, although the periodicity and rotation angle of the superperiodic structures with reference to underlying graphite varied. These images are similar to some of the reported Moiré-pattern images of graphite (with superperiodic structures ranging from 17 to 148 Å) obtained with bare metal tips (18–24). One of the most accepted explanations for the observation of superperiodic structures along with the underlying graphite lattice is the Moiré-rotation hypothesis, which assumes a simple rotation (misorientation) of surface graphite layer with respect to the underlying layer(s) (18–20). Moiré patterns are interference patterns arising from rotation between two layers of any repeating lattice, which can cause a superperiodic structure having the same symmetry as the original lattice. The images presented in Figs. 4 and 5 are particularly reminiscent of those reported by Bernhardt et al. (23) and Beyer et al. (24). In the both cases, the appearance of superperiodic structures was produced by manipulation of graphite sheets with the tip to generate rotation relative to the underlying graphite. As noted by Bernhardt et al. (23), the orientation angle of the superperiodic structure relative to the atomic lattice of the graphite layer is given by =30°±(θ/2), where D=d/2sin(θ/2), with d and D representing the lattice constants of the graphite and the superperiodic structure, respectively. The values of and D for the images in Fig. 5 are consistent with these equations. The larger superperiodicity observed in Fig. 4 should be associated with a value of nearer 30°, as observed, although the expected values of based on the superperiodicity in Fig. 4 (70.4 Å) are 29° and 31°. It has been reported that three-dimensional electron tunneling (electron scattering) in lattice-mismatched systems can also produce the superperiodic and underlying lattice images at the same time (25). Surface layers often reconstruct or relax to minimize the surface energy, so it is not unreasonable to think the outermost layer might have a slightly different lattice constant than the layers beneath it, giving rise to a latticemismatched system. However, this mechanism may not account for the rotation of the superperiodic structure with respect to the graphite lattice. As can be inferred from the molecular structure of H3PW12O40 shown in Fig. 1, another possibility for the origin of superimposed images is that the terminal W=O groups on a single H3PW12O40 molecule can serve as multiple tips depending on the molecular orientation, providing multiple pathways for electron tunneling. Simulation studies were carried out to test this hypothesis. The simulation allowed for up to four tips in any spatial arrangement desired, as well as arbitrary weights and heights for each tip. A hexagonal surface was created with a function reported in the literature for the simulation (21). Tips were examined that had the same symmetry as the surface but were scaled with integer and noninteger factors. Simulations were performed for both constant current topography and constant height topography STM modes. However, no superperiodic structures resulted from any of the simulations, indicating that independent multiple tips are not sufficient to explain the observed superperiodicity. Indeed, we conclude that tunneling with multiple tips can produce variations in the STM images at the length scale of the surface lattice constant or smaller—no long-range periodicity is produced in the STM images in such cases. We conclude, therefore, that the images in Figs. 4 and 5 are Moiré-rotation patterns caused by misalignment of the top few graphite layers of the substrate. We observe superperiodic structures approximately 20% of the time when imaging graphite with Pt/Ir-H3PW12O40 tips. We have never observed such structures when imaging graphite with bare Pt/Ir tips in air with our instruments. The question then arises as to why the HPA-functionalized tips are more sensitive to these structures than are the corresponding bare metal tips. Some insight may be offered by the unusual sensitivity of C60-functionalized tips to electronic perturbations of the surface reported by Halas and coworkers (26–28), who demonstrated that C60-functionalized tips were much more sensitive to electron scattering from point defects on graphite than were bare metal tips. Although C60-terminated tips were able to image anisotropic threefold scattering from point defects at room temperature, cryogenic temperatures are required to produce similar images with a bare metal tip (29). This difference has been explained in terms of the differences in the density of states of clean versus C60-functionalized tips. With the bare metal tip, thermal broadening of the density of states about the Fermi level of the tip reduces its ability to resolve the electronic perturbations at point defects; low temperatures are required to sharpen the occupancy of the tip density of states at the Fermi level (29). In contrast, the tip functionalized with molecular C60 is characterized by electronic states of the molecule that are narrow in energy compared with the bulk bands of
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
H3PW12O40-FUNCTIONALIZED TIP FOR SCANNING TUNNELING MICROSCOPY
6475
the metallic tip (26), and is able to resolve electron scattering from surface structures at ambient conditions. It is clear from the present work that this enhanced sensitivity to perturbations of surface electronic structure is not unique to C60 functionalization of STM tips. HPAs, pseudospherical metal oxide molecules of comparable dimensions to C60, may afford similar enhanced sensitivity to surface electronic structure in STM. Thus, this work extends the range of molecular tip-functionalization agents from conductive C60 to semiconducting metal oxides (HPAs). The observation of negative differential resistance behavior for HPAs [and the ability to vary the potential at which it occurs by choice of HPA structure and composition (8, 9)] suggests further opportunities for tuning the electronic properties of STM tips to probe a range of surface and subsurface electronic properties of imaged samples. CONCLUSIONS H3PW12O40 molecules were attached to a mechanically formed Pt/Ir tip to functionalize the metal tip as an STM probe. Attachment of H3PW12O40 on the Pt/Ir tip was confirmed by observing NDR behavior of the H3PW12O40 in tunneling spectra. Atomic resolution images of graphite as well as of H6P2W18O62 monolayers on graphite could be obtained with a Pt/Ir-H3PW12O40 tip. STM images of graphite probed by a Pt/Ir-H3PW12O40 tip showed the superimposed images of superperiodic structures and the real-size graphite lattice. The superperiodic structures have hexagonal symmetry commensurate with underlying graphite. These images represent Moiré patterns produced by misorientation of the top-most layers of the graphite surface. The HPA-functionalized tips appear to be particularly sensitive to these electronic perturbations of the surface owing to the discrete electronic levels provided by the semiconducting oxide (HPA) molecules. The Topometrix TMX 2010 was acquired by means of an equipment grant from the United States Department of Energy. I.K.S. acknowledges fellowship support from the Seoam Scholarship Foundation. 1. Melmed, A.J. (1991) J. Vac. Sci. Technol. B 9, 601–608. 2. Rohrer, G. (1993) in Scanning Tunneling Microscopy and Spectroscopy, ed. Bonnell, D. (VCH, New York), pp. 155–187. 3. Park, S., Nogami, J. & Quate, C.F. (1987) Phys. Rev. B 36, 2863–2866. 4. Colton, R.J., Baker, S.M., Driscoll, R.J., Youngquist, M.G., Baldeschwieler, J.D. & Kaiser, W.J. (1988) J. Vac. Sci. Technol. A 6, 349–353. 5. Tersoff, J. & Lang, N.D. (1990) Phys. Rev. Lett. 65, 1132–1135. 6. Pope, M.T. (1983) in Heteropoly and Isopoly Oxometalates (Springer, New York), pp. 58–100. 7. Okuhara, T., Mizuno, N. & Misono, M. (1996) Adv. Catal. 41, 113–252. 8. Kaba, M.S., Song, I.K. & Barteau, M.A. (1996) J. Phys. Chem. 100, 19577–19581. 9. Song, I.K., Kaba, M.S., Barteau, M.A. & Lee, W.Y. (1998) Catal. Today 44, 285–291. 10. Keggin, J.F. (1933) Nature (London) 131, 908–909. 11. Brown, G.M., Noe-Spirlet, M.R., Busing, W.R. & Levy, H.A. (1977) Acta. Crystallogr. B 33, 1038–1046. 12. Kaba, M.S., Song, I.K., Duncan, D.C., Hill, C.L. & Barteau, M.A. (1998) Inorg. Chem. 37, 398–406. 13. Dawson, B. (1953) Acta Crystallogr. 6, 113–126. 14. Strandberg, R. (1975) Acta Chem. Scand. A 29, 350–358. 15. Tromp, R.M., van Loenen, E.J., Demuth, J.E. & Lang, N.D. (1988) Phys. Rev. B 37, 9042–9046. 16. Lee, H.J. & Ho, W. (1999) Science 286, 1719–1722. 17. Hillier, A.C. & Ward, M.D. (1996) Phys. Rev. B 54, 14037–14051. 18. Kuwabara, M., Clarke, D.R. & Smith, D.A. (1990) Appl. Phys. Lett. 56, 2396–2398. 19. Liu, C., Chang, H. & Bard, A.L. (1991) Langmuir 7, 1138–1142. 20. Xhie, J., Sattler, K., Ge, M. & Venkateswaran, N. (1993) Phys. Rev. B 47, 15835–15841. 21. Cee, V.J., Patrick, D.L. & Beebe, T.P., Jr. (1995) Surf. Sci. 329, 141–148. 22. Osing, J. & Shvets, I.V. (1998) Surf. Sci. 417, 145–150. 23. Bernhardt, T.M., Kaiser, B. & Rademann, K. (1998) Surf. Sci. 408, 86–94. 24. Beyer, H., Müller, M. & Schimmel, Th. (1999) Appl. Phys. A 68, 163–166. 25. Kobayashi, K. (1996) Phys. Rev. B 53, 11091–11099. 26. Kelly, K.F., Sarkar, D., Hale, G.D., Oldenburg, S.J. & Halas, N.J. (1996) Science 273, 1371–1373. 27. Kelly, K.F. & Halas, N.J. (1998) Surf. Sci. 416, L1085–L1089. 28. Kelly, K.F., Mickelson, E.T., Hauge, R.H., Margrave, J.L. & Halas, N.J. (2000) Proc. Natl. Acad. Sci. USA 97, 10318–10321. 29. Kushmerick, J.G., Kelly, K.F., Rust, H.-P., Halas, N.J. & Weiss, P.S. (1999) J. Phys. Chem. B 103, 1619–1622.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
ENERGETICS OF NANOCRYSTALLINE TIO2
6476
Colloquium Energetics of nanocrystalline TiO2 M.R.Ranade*, A.Navrotsky*†, H.Z.Zhang‡, J.F.Banfield‡, S.H.Elder§, A.Zaban¶, P.H.Borse∥, S.K.Kulkarni∥, G.S.Doran**, and H.J.Whitfield†† *University of California, Department of Chemical Engineering and Materials Science, Thermochemistry Facility, Davis, CA 95616; ‡University of Wisconsin, Department of Geology and Geophysics, Madison, WI 53706; §Intel Corporation, Hillsboro, OR 97124; ¶Bar-Ilan University, Department of Chemistry, Ramat-Gan 52900, Israel; ∥University of Pune, Department of Physics, Pune 411007, India; **CSIRO Manufacturing Science and Technology Clayton, Victoria 3169, Australia; and ††Department of Applied Physics, Royal Melbourne Institute of Technology University, GPO Box 2476V, Melbourne, Victoria 3001, Australia Contributed by A.Navrotsky, October 9, 2001 The energetics of the TiO2 polymorphs (rutile, anatase, and brookite) were studied by high temperature oxide melt drop solution calorimetry. Relative to bulk rutile, bulk brookite is 0.71±0.38 kJ/mol (6) and bulk anatase is 2.61±0.41 kJ/mol higher in enthalpy. The surface enthalpies of rutile, brookite, and anatase are 2.2±0.2 J/m2, 1.0±0.2 J/m2, and 0.4±0.1 J/m2, respectively. The closely balanced energetics directly confirm the crossover in stability of nanophase polymorphs inferred by Zhang and Banfield (7). An amorphous sample with surface area of 34,600 m2/mol is 24.25±0.88 kJ/mol higher in enthalpy than bulk rutile. Titania is an important accessory oxide mineral (1) and is used widely in technology (2–4). Rutile is the stable high temperature phase, but anatase and brookite are common in fine grained (nano-scale) natural and synthetic samples (5–7). On heating concomitant with coarsening, the following transformations are all seen, each under somewhat different conditions of particle size, starting material, temperature, and other parameters (refs. 4 and 7–9; Appendix A): anatase to brookite to rutile, brookite to anatase to rutile, anatase to rutile, and brookite to rutile. These variable transformation sequences imply very closely balanced energetics as a function of particle size (10). It has been proposed that the surface enthalpies of the three polymorphs are sufficiently different that crossover in thermodynamic stability can occur under conditions that preclude coarsening, with anatase and/or brookite stable at small particle size (3, 7). The energetics of anatase, brookite, and rutile were measured by high temperature oxide melt solution calorimetry, first by Navrotsky and Kleppa (5) and later by Mitsuhashi and Kleppa (6). Their results, as well as those of other studies (11–14), scatter significantly (Table 1 and Appendix B). Previously, using high temperature oxide melt calorimetry for nanocrystalline Al2O3, it was shown that γ- Al2O3 (the phase observed for nano-sized particles) is more stable in enthalpy than nanophase α-Al2O3 (corundum), the macrocrystalline thermodynamically stable phase (15, 16). The goal of this work is to gather calorimetric evidence concerning the analogous proposed phase stability reversal for nanocrystalline TiO2. Such information is essential for understanding fundamental solid state chemistry, for predicting phase equilibria, for controlling nucleation, grain growth, and phase transformation, and thus is fundamental to technological applications. Energetics of anatase, brookite, and rutile have been measured by high-temperature oxide melt drop solution calorimetry at 975 K with 3Na2O·4MoO3 solvent, using a Calvet twin microcalorimeter. The effects of particle size and adsorbed water are considered. EXPERIMENTAL SECTION Synthesis. Anatase NA1 was synthesized from (NH4)2 Ti(OH)2(C3H5O3)2 and cetyltrimethylammonium chloride (CTAC). A white precipitate was formed when water was added. Vigorous stirring readily dissolved it. Water addition and stirring was continued until irreversible precipitation occurred. The reaction was stirred first at room temperature overnight, then at 343 K for 24 h, and finally at 373 K for 48 h in a sealed Teflon reactor. The precipitate was washed and centrifuged several times with water. The sample was calcined at 623 K for 2 h in air to remove CTAC. Anatase-rutile samples NAR1-NAR6 were generated by heating NA1 for 4.5, 24, 48, 72, 144, and 240 h at 623 K. Table 1. Reported enthalpies of transformation for TiO2 polymorphs ∆H, kJ/mol Method Rutile=anatase −11.67* High P-T studies −0.42* Tabulation 0.42±0.21 Differential thermal analysis Oxide melt calorimetry, surface area considered 2.61±0.41† 2.93±1.26 Differential scanning calorimetry 3.26±0.84 Oxide melt calorimetry 6.57±0.79 Oxide melt calorimetry 8.37±5.92 Fluorine combustion calorimetry Rutile=brookite 0.42±0.31 Differential thermal analysis Oxide melt calorimetry 0.71±0.38† 0.84±0.42 Differential scanning calorimetry Fluorine combustion calorimetry 41.84±9.36*
Reference 11 13 12 This work 6 6 5 14 12 6 6 14
*Unreasonable. †Recommended values.
Amorphous TiO2 (AM) was synthesized by mixing titanium ethoxide with ethanol (17). The mixture was stirred at 273 K for 2 h. After dripping water (1.6 mol) that contained four drops of acetic acid, the mixture was aged for 0.5 h. The product was white and x-ray amorphous. Anatase samples NA2, NA3, and NA4 were obtained by heating AM for 3 h in air at 673 K, 771 K, and 748 K, respectively. NR1 and NR2 nanorutile samples were synthesized by using a sol-gel method (18). Solutions of TiCl3 and NH4OH were
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC. Abbreviation: XRD, x-ray diffraction patterns. †To whom reprint requests should be addressed at: 4440 Chemistry Annex, Thermochemistry Facility, Department of Chemical Engineering and Materials Science, University of California, One Shields Avenue, Davis, CA 95616. E-mail: [email protected].
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
ENERGETICS OF NANOCRYSTALLINE TIO2
6477
mixed in the volume ratio 1:6 and 1:3, respectively. The initial pH was 0.1, which slowly increased to 2. The mixture was stirred for 48 h at 300 K. The precipitate was centrifuged and washed with isopropyl alcohol and dried at room temperature to obtain nanorutile. To make NR3, titanium isopropoxide and dry isopropyl alcohol were added dropwise to a well stirred solution of nitric acid (pH, 0.5; ref. 18). The sample was stirred for 8 h. Isopropyl alcohol was evaporated at 355 K. The sample was transferred to a Teflon container and heated in a titanium autoclave at 523 K for 26 h under continuous stirring. The powder was dried. Pure nanophase brookite NB was synthesized as described (19, 20). Titanium isopropoxide was mixed with distilled water in a beaker and stirred at room temperature for 1 h. The suspension was vacuum filtered and washed with distilled water, ensuring that the precipitate remained wet. This paste was added to an appropriate amount of NaOH in an autoclave and heated at 423 K for 14 days. The autoclave was cooled for a day and the precipitate was collected, filtered, washed with distilled water, and dried. Bulk rutile (99.995% metals basis) dried at 973 K overnight was used to complete the thermochemical cycles. Characterization. X-ray diffraction patterns (XRD) were collected by using a Scintag (Santa Clara, CA) PAD V diffractometer (Cu Kα, 45 kV, 40 mA, θ-2θ goniometer geometry) with a step size of 0.02° and collection time of 10 s per step (without an internal standard). The specific surface area was measured by using the Brunauer-Emmett-Teller (BET) technique with a Gemini 2360 instrument (Micromeritics, Norcross, GA). For electron microprobe analysis, a Cameca (Paris) SX-50 instrument was used. Carbon content was obtained from LECO (St. Joseph, MI). Water contents were determined by using weight loss measurements. TiO2 samples, accurately weighed inside a glovebox, were heated for 12 h at 1,473 K. The products were white, implying negligible oxygen deficiency. The difference between the final weight and the initial weight gives the water content. The molar compositions (formula weights) were normalized to obtain 1 M of sample, x M of water, and y M of other impurities. Calorimetry. All moisture-sensitive nanocrystalline samples were handled in an argon-filled glovebox (O2 and H2O<1 ppm). High temperature drop solution calorimetry in 3Na2O·4MoO3 solvent was performed in a custom-built Calvet twin microcalorimeter described in refs. 21 and 22. Oxygen gas was flushed through the glassware at ≈90 ml/min and bubbled through the solvent at ≈5 ml/min. This procedure maintains oxidizing conditions, helps remove evolved water, and agitates the solvent to aid dissolution. The samples were pressed into pellets (usually ≈15 mg) inside the glovebox, weighed, and stored in a glass vial. When a stable calorimeter baseline was obtained, the pellet from the glass vial was dropped into the calorimeter. The total time the pellet was exposed to the atmosphere was less than 10 s. For all samples, calibration was performed by using the heat content of corundum pellets of similar weight. RESULTS AND DISCUSSION Sample Characterization. XRD verified that the samples were single-phase (anatase NA1–NA4, rutile NR1–NR3, or brookite NB) and quantified the amounts of anatase and rutile in the two-phase mixtures, according to procedures similar to those of Zhang and Banfield (7). In addition to water, nanocrystalline samples prepared by sol-gel methods may retain some organic impurities. In all of the nano TiO2 samples, ≈0.1 wt % carbon was detected. We suspect this is physisorbed CO2 resulting from exposure to air during the commercial chemical analysis and call it “atmospheric” carbon. We believe this was not present on the samples handled in the glovebox and used for calorimetry. A scan combining thermogravimetry, differential thermal analysis, and mass spectroscopy of calorimetric samples did not detect CO or CO2. The “atmospheric” carbon was subtracted from the measured carbon content to determine the actual carbon content of the calorimetric samples. In anatase-rutile mixtures, the major phase (>95%) was anatase. XRD showed that when the particle size of anatase reached ≈13 nm, rutile began to form, as observed (3). Once formed, the rutile particles grew very rapidly. Thus, samples NAR1–NAR6 were mixtures of nanocrystalline anatase and rutile with rapidly increasing particle size. The specific surface area rapidly decreased, suggesting coarsening. The surface area and transformation enthalpy for the anatase-rutile mixtures are normalized to correct for rutile to obtain the variation of enthalpy of pure anatase with surface area (see footnote in Table 2). Because rutile coarsens more rapidly than anatase, we used the enthalpy of bulk rutile for this correction. This effect does not introduce significant error because the amount of rutile is small (<5%). The difference between particle sizes estimated by XRD and BET suggests considerable agglomeration of nanoparticles (Table 2). Thermochemistry. Table 2 lists the enthalpies of drop solution. The enthalpy difference between a nanocrystalline sample and bulk rutile arises from polymorphism, surface energy, and presence of water and any other impurities. Correction for water, present in all nanocrystalline TiO2 samples, is performed through a thermochemical cycle (Appendix C). Because the weight loss observed by thermogravimetry occurs below 673 K, the water is assumed to be physisorbed and to have the enthalpy of bulk liquid water. The carbon impurity in amorphous TiO2 (AM) is assumed to be energetically equivalent to graphite (Appendix C). The brookite sample (NB) also needs to be corrected for a sodium impurity—this is assumed to be Na2CO3 (Appendix C). The difference between the corrected drop solution enthalpy of the nanocrystalline sample and bulk rutile is equal to the sum of the enthalpy due to polymorphism and surface enthalpy. It is called the transformation enthalpy and can be written, for example for anatase, as
[1]
where γ is the surface enthalpy and A is the surface area. Fig. 1 shows the transformation enthalpies of nanocrystalline samples (kJ/mol) versus surface areas (m2/mol). A linear fit for each structure yields both the surface enthalpy (slope) and the bulk phase transformation enthalpy (intercept). Fig. 1a represents the enthalpy of nanorutile. Bulk rutile is taken as the reference point of zero enthalpy for all samples. A linear fit using these three nanorutile data are therefore forced through the origin to derive the surface enthalpy of nanocrystalline rutile as 2.2±0.2 J/m2. The uncertainty comes from the fitting program, but there are too few data points for more detailed statistical analysis. Fig. 1b shows the anatase data (normalized to pure anatase as discussed above). A linear fit yields the surface enthalpy of anatase as 0.4 ±0.1 J/m2 and the enthalpy of phase transformation of bulk rutile to bulk anatase as 2.61±0.41 kJ/mol. The uncertainty comes from the statistical fitting program, and the 95% confidence limits are shown in the figure.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
ENERGETICS OF NANOCRYSTALLINE TIO2
Table 2. Sample characterization and thermochemical data Sample Composition Wt % C Particle size XRD, nm Rutile NA1 NA2 NA3 NA4 NAR1 NAR2 NAR3 NAR4 NAR5 NAR6 NR1 NR2 NR3 NB AM
Alfa Aesar Puratronic TiO2 Anatase, 0.027 H2O Anatase, 0.1606 H2O Anatase, 0.03283 H2O Anatase, 0.07308 H2O 0.018 Rutile, 0.982 anatase, 0.05883 H2O 0.048 Rutile, 0.952 anatase, 0.04446 H2O 0.045 Rutile, 0.955 anatase, 0.04428 H2O 0.040 Rutile, 0.960 anatase, 0.01612 H2O 0.036 Rutile, 0.964 anatase, 0.01707 H2O 0.051 Rutile, 0.949 anatase, 0.01500 H2O Rutile, 0.23473 H2O Rutile, 0.06613 H2O Rutile, 0.08316 H2O Brookite, 0.1219 H2O, 8.325e-3 Na2CO3 Amorphous, 0.03283 H2O, 0.0404 C
6478
Surface area BET, m2/mol and m2/g* >1000
Particle size calculated from BET, nm
∆Hds (measured), kJ/ mol†
∆Hds (corrected), kJ/ mol‡
57.95±0.71 (23)
57.95±0.71 (23)
∆Htrans, kJ/mol§
0.143
10.4
11985 (150)
10.4
50.73±0.65 (8)
50.93±0.65
7.02±0.96
0.086
7.2
9588 (120)
13.0
60.49±0.60 (5)
49.41±0.60
8.54±0.93
0.039
24.2
3036 (38)
41.1
59.05±0.42 (6)
56.79±0.42
1.16±0.82
0.067
13.5
5273 (66)
23.7
59.04±0.39 (5)
53.99±0.39
3.95±0.81
0.049
13.1 (A) 23.8 (R)
5193 (65) 5289 (66)¶
24.0
55.43±0.44 (6)
51.37±0.44
0.042
16.0 (A) 19.0 (R)
4235 (53) 4448 (56)¶
29.4
56.53±0.49 (6)
53.46±0.49
0.039
20.4 (A) 48.4 (R)
3436 (43) 3598 (45)¶
36.3
58.03±0.57 (6)
54.97±0.57
0.042
22.2 (A) 41.1 (R)
479 (6) 499 (6)¶
259.8
55.26±0.39 (6)
54.15±0.39
0.050
24.4 (A) >100.0 (R)
399 (5) 414 (5)¶
312.0
56.26±0.29 (6)
55.08±0.29
0.128
23.9 (A) 63.3 (R)
399 (5) 421 (5)¶
310.9
55.64±0.33 (6)
54.61±0.33
0.087
8.0
5833 (73)
19.3
61.54±0.73 (7)
45.15±0.73
0.052
12.0
4554 (57)
24.7
53.54±0.58 (6)
48.98±0.58
6.58±0.83 6.70 ±0.83¶ 4.48±0.87 4.71 ±0.87¶ 2.98±0.91 3.12 ±0.91¶ 3.80±0.81 3.96 ±0.81¶ 2.87±0.77 2.97 ±0.77¶ 3.34±0.78 3.52 ±0.78¶ 12.76 ±1.01 8.97±0.91
—**
22.0
1758 (22)
64.0
57.77±0.24 (4)
52.03±0.24
5.92±0.75
0.210
23.0
6392 (80)
18.0
60.36±0.63 (10)
50.84±0.88
7.11±1.13
0.422
—
34596 (433)
—
58.34±0.53 (6)
33.69±0.53
24.25 ±0.88
*Values in parentheses are the experimentally obtained BET areas in m2/g. †Value is the mean of the number of experiments indicated in parentheses. Error is two standard deviations of the mean. ‡After correction for H O and other impurities. 2 §For reaction TiO (rutile, bulk)=TiO (polymorph, nano). 2 2 ¶Normalized values of the surface area and transformation enthalpy for anatase-rutile mixtures (NAR1–NAR6). Calculations to obtain 100% anatase from these samples are as follows: x anatase · (1−x) rutile mixture with surface area A and transformation enthalpy H is normalized to 100% anatase with surface area of A/x and transformation enthalpy of H/x. **Not detected because of the limited amount of sample.
Fig. 1c represents the enthalpy of the one nanocrystalline brookite (NB) and the bulk brookite-rutile phase transformation enthalpy (0.71 ±0.38 kJ/mol) of Mitsuhashi and Kleppa (6). Attempts to obtain pure brookite of other surface areas, either by direct synthesis or by coarsening sample NB, were not successful. Therefore, the linear fit is constrained by only two points. It gives the surface enthalpy of brookite as 1.0±0.2 J/ m2. This uncertainty is only an estimate. Fig. 1d summarizes the enthalpy of nanocrystalline titania. The intersections of the linear fits of brookite and rutile and of brookite and anatase place a limit on the stability field of various polymorphs. Rutile is energetically stable for surface area <592 m2/mol (7 m2/g), brookite is energetically stable from 592 to 3,174 m2/mol (7–40 m2/g), and anatase is energetically stable for greater surface areas. The anatase and rutile energetics cross at 1,452 m2/mol (18 m2/g). The dark solid lines represent the phases of lowest enthalpy as a function of surface area. The energetic stability crossovers are confirmed. Assuming spherical particles, the calculated average diameters of rutile and brookite for 7 m2/g surface area are 201 nm and 206 nm, and of brookite and anatase for 40 m2/g surface area are 36 nm and 39 nm. These differences in particle size at the same surface area exist because of the difference in density of the phases. If the phase transformation takes place without further coarsening, the particle size should be smaller after the transformation. In practice, this effect has probably not been observed. Of course phase stability in a thermodynamic sense is governed by the Gibbs free energy (∆G=∆H−T∆S) rather than the enthalpy. Lowtemperature heat capacity and entropy data are available for anatase and rutile (6), but not for brookite. The data for anatase probably refer to a fine-grained (perhaps nanophase) sample, but no characterization is given. Thus the anatase entropy may contain contributions from both bulk and surface terms. The data suggest that rutile and anatase have the same entropy within experimental error [S0 (298 K, rutile)= 50.6±0.6 J/ mol·K and S0 (298 K, anatase)=49.9±0.3 J/mol·K (35)]. Thus the T∆S will not significantly perturb the sequence of stability seen from the enthalpies. Amorphous TiO2 with surface area of 34600 m2/mol was found to be 24.25±0.88 kJ/mol higher in enthalpy than bulk rutile. General Implications of Surface Enthalpies and Enthalpies of Phase Transformation. Zhang and Banfield (7) reported the average surface enthalpies of rutile and anatase as 1.93 and 1.34 J/m2, respectively, based on atomistic simulations. They estimated the surface enthalpy of brookite as 1.66 J/m2. These theoretical values were consistent with the values of 0.5–1.7 J/m2 estimated by differential scanning calorimetry by Terwilliger and Chiang (36). Vittadini et al. (37) used the generalized gradient approximations of density functional theory to calculate the energy of the anatase (101) and (001) surface as 0.52 and 0.81 J/m2 and that of the rutile (110) surface as 0.82 J/m2. This suggests that rutile has the highest surface enthalpy, brookite the intermediate, and anatase the lowest. The calorimetric surface enthalpies for rutile, brookite, and anatase, 2.2±0.2, 1.0±0.2, and 0.4± 0.1 J/m2, respectively, agree with this sequence, although their numerical values are somewhat different. The theoretical cal
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
ENERGETICS OF NANOCRYSTALLINE TIO2
6479
culations calculate energies for specific crystal faces. However, it is not known what factors, both thermodynamic and kinetic, control the development of various crystal faces in the nanocrystalline powders. The relative abundance of different surfaces, even for the same area, may vary for differently prepared samples. The surface enthalpies derived from the calorimetric data thus represent a direct determination of the surface enthalpy averaged for different crystal faces as they exist in real nanocrystalline samples. The scatter in the nanoanatase calorimetric data may reflect such differences rather than purely random statistical factors.
Fig. 1. Enthalpy of nanocrystalline samples with respect to bulk rutile (kJ/mol) versus surface area (m2/mol) for nanorutile samples NR1– NR3 (a), for nanoanatase samples NA1–NA4 (b), and normalized nanoanatase-rutile mixtures NAR1-NAR6 (dashed curves represent 95% confidence limits for the mean) for nanobrookite sample NB and brookite-rutile phase transformation from Mitsuhashi and Kleppa (6) (c), and phase stability crossover of titania. The lines are taken from a–c and the darker line segments indicate the energetically stable phases (d).
For nanocrystalline titania, the variable observed transformation sequence can be explained by using the enthalpy crossover (Fig. 1d). If the initially formed brookite has surface area >40 m2/g, it is metastable with respect to both anatase and rutile, and the sequence brookite to anatase to rutile during coarsening is energetically downhill. If anatase formed initially, it can coarsen and transform first to brookite (at 40 m2/ g) and then to rutile. The energetic driving force for the latter reaction (brookite to rutile) is very small, explaining the natural persistence of coarse brookite. In contrast, the absence of coarse grained anatase is consistent with the much larger driving force for its transformation to rutile. APPENDIX A: BRIEF LITERATURE REVIEW OF ANATASE-BROOKITE-RUTILE PHASE TRANSFORMATION Zhang and Banfield (3) and Gribb and Banfield (4) observed that the synthesis of ultrafine titania resulted in anatase and/or brookite, which on coarsening transformed to rutile after reaching a certain particle size. Once rutile was formed, it grew much faster than anatase. They analyzed the phase stability of anatase and rutile thermodynamically to conclude that anatase became more stable than rutile for particle size <14 nm. Hwu et al. (23) commented that whether TiO2 was rutile or anatase depended on the preparation method. For small particle size (<50 nm) anatase seemed more stable and transformed to rutile at ≈973 K. Zhang and Banfield (7) studied the phase transformation behavior of nanocrystalline aggregates during their growth for isothermal and isochronal reactions by using XRD. They suggested that transformation sequence and thermodynamic phase stability depend on the initial particle sizes of anatase and brookite. They concluded that, for equally sized particles, for particle size <11 nm, anatase was thermodynamically stable, for particle size between 11 nm and 35 nm, brookite was stable, and for particle size >35 nm, rutile was stable. They cautioned that, for real samples, the particle sizes of different phases were not equal, which could alter the direction of initial transformation. They concluded that the energetics of these polymorphs were sufficiently close that they could be reversed by small differences in surface energy. Ye et al. (8) studied the thermal behavior of nanocrystalline brookite by using thermogravimetry, differential thermal analysis, and diffraction. They observed a slow brookite to anatase phase transition below 1,053 K along with grain growth. Between
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
ENERGETICS OF NANOCRYSTALLINE TIO2
6480
1,053 K and 1,123 K, they noticed rapid brookite to anatase and anatase to rutile transformations. Above 1,123 K, they saw rapid grain growth of rutile, which became the dominant phase. They concluded that brookite cannot transform directly to rutile but has to transform to anatase first. Kominami et al. (9) observed that nanocrystals of brookite directly transformed to rutile above 973 K, in contrast to the observations of Ye et al. Zhang et al. (24) measured micro-Raman spectra of nanosized TiO2 powders prepared by vapor hydrolysis. They obtained amorphous TiO2 at 533 K and predominantly anatase at 873– 1,173 K. The anatase-rutile transformation temperature depended on particle size and was complete at ≈1,323 K. They noted that the phase transformation of amorphous TiO2 was a two-step process: amorphous to anatase followed by anatase to rutile. They proposed that the rutile formation started at the surface and migrated into the bulk. They also noted that the brookite impurities, detected by Raman spectroscopy but not by XRD, were present on the anatase surface. Zhang and Banfield (25, 26) proposed that the mechanism of anatase-rutile phase transformation was temperature-dependent. They suggested this transformation was dominated by interface nucleation below 873 K, by both interface and surface nucleation at 893–1,273 K, and by bulk nucleation above 1,273 K. The anatase-rutile transformation depends on impurities, grain size, reaction atmosphere, and synthesis conditions (26– 31). Zhang and Banfield (26) reported that the anatase-rutile phase transformation occurred at higher temperature with the addition of Al2O3. They attributed this to suppression of coarsening caused by surface diffusion. Okada et al. (27) and Yoshinaka et al. (28) found that the anatase-rutile phase transformation occurred at higher temperature with the addition of SiO2. Yang et al. (29) showed that synthesis conditions (chemicals/ peptizing agents) affect the crystallinity and anatase-rutile phase transition temperature. Zaban et al. (30) noted that the surface structure of TiO2 is affected by the preparation conditions. Ahonen et al. (31) studied the effect of gas atmospheres (nitrogen and air) and temperature on the crystal structure and specific surface area. They observed that anatase synthesized in air transformed to rutile at 973 K, whereas anatase synthesized in nitrogen persisted to 1,173 K. Gouma and Mills (32) studied the anatase-rutile phase transformation in commercial TiO2 powders with an average particle size of 100 nm. Using transmission and scanning electron microscopy, they concentrated on the structural evolution (shape and morphology) of the particles. They proposed that rutile plates were formed initially by a shear force and subsequent coarsening involved interactions between the transforming particle and surrounding anatase particles. APPENDIX B: PREVIOUS STUDIES OF TRANSFORMATION ENTHALPIES B1 Rutile-Anatase. Navrotsky and Kleppa (5) studied rutile-anatase phase transformation by using high-temperature oxide melt solution calorimetry. They obtained the enthalpy of phase transformation at 968 K as 6.57±0.79 kJ/mol. They did not report the water contents, impurities, and particle size. They concluded that anatase was metastable with respect to rutile under all conditions of temperature and pressure. Mitsuhashi and Kleppa (6) determined the enthalpy of the rutile-anatase phase transformation by using high-temperature oxide melt solution calorimetry and differential scanning calorimetry (DSC) to be 3.26±0.84 kJ/mol and 2.93±1.26 kJ/mol, respectively. The values from solution calorimetry and DSC agree. However, they did not specify the particle size for their hydrothermally prepared anatase. Zhang and Banfield (7) suggested that this sample is likely to be fairly coarse. Mitsuhashi and Kleppa (6) mentioned that the samples contained <0.7 wt % volatiles and ≈1–3 mol % other TiO2 impurities, but they apparently did not correct the calorimetric results for the impurities. They suggested that the discrepancy between their results and earlier studies (5) might be due to incomplete dissolution in the earlier work. We believe that the difference in particle size may be the governing factor. Rao (12) estimated the enthalpy of the rutile-anatase phase transformation at 1,176 K to be 0.42±0.21 kJ/mol by DTA using spectroscopically pure anatase. The sample preparation was identical to that of Czanderna et al. (33). They reported a surface area of ≈40 m2/g (which corresponds to ≈40-nm-diameter particles). Margrave and Kybett (14) reported a rutile-anatase transformation enthalpy of 8.37±5.92 kJ/mol when using fluorine combustion calorimetry. Based on high-pressure-temperature phase data, Vahldiek (11) estimated the rutile-anatase transformation enthalpy as −11.67 kJ/ mol. Robie and Waldbaum (13) tabulated the rutile-anatase phase transformation enthalpy as −0.42 kJ/mol. B2 Rutile-Brookite. Rao et al. (34) estimated the enthalpy of the rutile-brookite phase transformation as 0.42±0.31 kJ/mol based on DTA experiments on a natural sample that contained ≈4 wt % impurities. Margrave and Kybett (14), using fluorine combustion, reported the rutilebrookite phase transformation enthalpy to be 41.84±9.36 kJ/mol. This value seems unreasonably large. Mitsuhashi and Kleppa (6) synthesized brookite by hydrolysis of tetraisopropyl orthotitanate in an aqueous solution of 1 M NaF. They treated the precipitate hydrothermally for 48 h at 703 K and 1 kbar (1 bar=100 kPa), washed with hot dilute HNO3, and dried at 408 K. Oxide melt solution calorimetry gave the rutile-brookite transformation enthalpy as 0.71±0.38 kJ/mol. In the same study, a DSC experiment on a natural brookite sample (with unspecified impurity content) gave the transformation enthalpy as 0.84±0.42 kJ/mol. The authors noted difficulties in using the natural sample for reliable solution calorimetry because of its impurity content and in using the synthesized sample for DSC experiments because of baseline problems resulting from shrinkage of the sample. However, they concluded that the magnitude of the enthalpy for transformation was well established and suggested that the value derived from solution calorimetry should be used. Both their samples are likely to be relatively coarse. APPENDIX C: THERMOCHEMICAL CYCLES C1 Thermochemical cycle to calculate the enthalpy from bulk rutile to pure anatase, pure rutile, and anatase-rutile mixtures. TiO2·x H2O (nano, 298 K)→TiO2 (soln, 975 K)+ x H2O (g, 975 K) x H2O (g, 975 K)→x H2O (1, 298 K) TiO2 (nano, 298 K)+x H2O (1, 298 K)→ TiO2·x H2O (nano, 298 K) TiO2 (soln, 975 K)→TiO2 (rutile, 298 K) TiO2 (rutile, 298 K)→TiO2 (nano, 298 K) ∆H5=−(∆H1+∆H2+∆H3+∆H4)
∆H1 ∆H2 ∆H3 ∆H4 ∆H5
C2 Thermochemical cycle to calculate the enthalpy of bulk rutile-amorphous TiO2 (with C impurities assuming impurities behave as mechanical additions, see text). TiO2·x H2O·y C (amorphous, 298 K)+y O2 (g, 975 K)→ TiO2 (soln, 975 K) +x H2O (g, 975 K)+y CO2 (g, 975 K) x H2O (g, 975 K)→x H2O (1, 298 K) y O2 (g, 298 K)→y O2 (g, 975 K) y CO2 (g, 975 K)→y CO2 (g, 298 K) y CO2 (g, 298 K)→y C (graphite, 298 K)+ y O2 (g, 298 K) TiO2 (amorphous, 298 K)+x H2O (1, 298 K)→
∆H6 ∆H7 ∆H8 ∆H9 ∆H10
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
ENERGETICS OF NANOCRYSTALLINE TIO2
TiO2·x H2O (nAB, 298 K) TiO2 (soln, 975 K)→TiO2 (rutile, 298 K) TiO2 (rutile, 298 K)→TiO2 (amorphous, 298K) ∆H13=−(H6+H7+∆H8+∆H9+∆H10+∆H11+∆H12)
6481
∆H11 ∆H12 ∆H13
C3 Thermochemical cycle to calculate the enthalpy of bulk rutile-brookite (with Na2CO3 impurities assuming impurities behave as mechanical additions, see text). TiO2·x H2O·y Na2CO3 (nB, 298 K)→TiO2 (soln, 975 K)+ y Na2O (soln, 975 K) +y CO2 (g, 975 K)+x H2O (g, 975 K) x H2O (g, 975 K)→x H2O (1, 298 K) TiO2 (nB, 298 K)+x H2O (1, 298 K)→TiO2·x H2O (nB, 298 K) TiO2 (soln, 975 K)→TiO2 (rutile, 298 K) y Na2O (soln, 975 K)+y CO2 (g, 975 K)→y Na2CO3 (crystalline, 298 K) TiO2 (rutile, 298 K)→TiO2 (nB, 298 K) ∆H19=−(∆H14+∆H15+∆H16+∆H17+∆H18)
∆H14 ∆H15 ∆H16 ∆H17 ∆H18 ∆H19
S.H.E. performed a portion of this work at the William R.Wiley Environmental Molecular Sciences Laboratory at Pacific Northwest National Laboratory. S.H.E. is grateful to the Office of Basic Energy Sciences, Division of Materials Science, Department of Energy, for supporting this work. S.K.K. thanks the Indian Space Research Organization (ISRO), India, for financial support. We thank J.Majzlan and J.M.Neil for assistance and discussion. National Science Foundation Grants EAR9725020 and EAR0123998 supported the work of A.N., and Grant EAR9814333 supported the work of H.Z.Z. and J.F.B. 1. Banfield, J.F., Bischoff, B.L. & Anderson, M.A. (1993) Chem. Geol. 110, 211–230. 2. Elder, S.H., Cot, F.M., Su, Y., Heald, S.M., Tyryshkin, A.M., Bowman, M.K., Gao, Y., Joly, A.G., Balmer, M.L., Kolwaite, A.C., et al. (2000) J. Am. Chem. Soc. 122, 5138–5146. 3. Zhang, H.Z. & Banfield, J.F. (1998) J. Mater. Chem. 8, 2073–2076. 4. Gribb, A.A. & Banfield, J.F. (1997) Am. Mineral. 82, 717–728. 5. Navrotsky, A. & Kleppa, O.J. (1967) J. Am. Ceram. Soc. 50, 626. 6. Mitsuhashi, T. & Kleppa, O.J. (1979) J. Am. Ceram. Soc. 62, 356–357. 7. Zhang, H.Z. & Banfield, J.F. (2000) J. Phys. Chem. B 104, 3481–3487. 8. Ye, X.S., Sha, J., Jiao, Z.K. & Zhang, L.D. (1997) Nanostruct. Mater. 8, 919–927. 9. Kominami, H., Kohno, M. & Kera, Y. (2000) J. Mater. Chem. 10, 1151–1156. 10. Navrotsky, A. (2001) Thermochemistry of Nanomaterials in Nanoparticles and the Environment, Reviews in Mineralogy and Geochemistry, eds. Banfield, J.F. & Navrotsky, A. (Mineralog. Soc. Am.), in press. 11. Vahldiek, F.W. (1966) J. Less Common Met. 11, 99–110. 12. Rao, C.N.R. (1961) Can. J. Chem. 39, 498–500. 13. Robie, R.A. & Waldum, D.R. (1968) U.S. Geol. Surv. Bull. 1259, 146. 14. Margrave, J.L. & Kybett, B.D. (1965) Tech. Rep. No. AFMO-TR-65 (Air Force Materials Laboratory, Research and Technology Division, Air Force Systems Command, Wright-Patterson Air Force Base, Ohio), p. 123. 15. McHale, J.M., Auroux, A., Perrota, A.J. & Navrotsky, A. (1997) Science 277, 788–791. 16. McHale, J.M., Navrotsky, A. & Perrota, A.J. (1997) J. Phys. Chem. 101, 603–613. 17. Zhang, H.Z., Finnegan, M. & Banfield, J.F. (2001) Nanoletters 1, 81–85. 18. Aruna, S.T., Tirosh, S. & Zaban, A. (2000) J. Mater. Chem. 10, 2388–2391. 19. Keesman, I. (1966) Z. Anorg. Chem. 346, 30–43. 20. Bastow, T.J., Doran, G. & Whitfield, H.J. (2000) Chem. Mater. 12, 436–439. 21. Navrotsky, A. (1997) Phys. Chem. Miner. 24, 222–241. 22. Navrotsky, A. (1977) Phys. Chem. Miner. 2, 89–104. 23. Hwu, Y., Yao, Y.D., Cheng. N.F., Tung, C.Y. & Lin, H.M. (1997) Nanostruct. Mater. 9, 355–358. 24. Zhang, Y., Chan, C.K., Porter, J.F. & Guo, W. (1998) J. Mater. Res. 13, 2602–2609. 25. Zhang, H.Z. & Banfield, J.F. (1999) Am. Mineral. 84, 528–535. 26. Zhang, H.Z. & Banfield, J.F. (2000) J. Mater. Res. 15, 437–448. 27. Okada, K., Yamamoto, N., Kameshima, Y. & Yasumori, A. (2001) J. Am. Ceram. Soc. 84, 1591–1596. 28. Yoshinaka, M., Hirota, K. & Yamaguchi, O. (1997) J. Am. Ceram. Soc. 80, 2749–2753. 29. Yang, J., Mei, S. & Ferreira, J.M.F. (2000) J. Am. Ceram. Soc. 83, 1361–1268. 30. Zaban, A., Aruna, S.T., Tirosh, S., Gregg, B.A. & Mastai, Y. (2000) J. Phys. Chem. B 104, 4130–4133. 31. Ahonen, P.P., Kauppinen, E.I., Joubert, J.C., Deschanvres, J.L. & Van Tendeloo, G. (1999) J. Mater. Res. 14, 3938–3948. 32. Gouma, P.I. & Mills, M.J. (2001) J. Am. Ceram. Soc. 84, 619–622. 33. Czanderna, A.W., Rao, C.N.R. & Honig, J.M. (1958) Trans. Faraday Soc. 54, 1069–1073. 34. Rao, C.N.R., Yoganarasimhan, S.R. & Faeth, F.A. (1961) Trans. Faraday Soc. 57, 504–510. 35. Robie, R.A. & Hemingway, B.S. (1995) U.S. Geol. Surv. Bull. 2131. 36. Terwilliger, C.D. & Chiang, Y.M. (1995) J. Am. Ceram. Soc. 78, 2045–2055. 37. Vittadini, A., Selloni, A., Rotzinger, F.P. & Gratzel, M. (1998) Phys. Rev. Lett. 81, 2954–2957.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
STUDY OF ND3+, PD2+, PT4+, AND FE3+ DOPANT EFFECT ON PHOTOREACTIVITY OF TIO2 NANOPARTICLES
6482
Colloquium Study of Nd3+, Pd2+, Pt4+, and Fe3+ dopant effect on photoreactivity of TiO2 nanoparticles S.I.Shah*†‡, W.Li*, C.-P.Huang§, O.Jung¶ and C.Ni* Departments of *Materials Science and Engineering, †Physics and Astronomy, and §Civil and Environmental Engineering, University of Delaware, Newark, DE 19716; and ¶Department of Environmental Engineering, Chosun University, Gwang Ju, 501–759, Republic of Korea Edited by Alexandra Navrotsky, University of California, Davis, CA, and approved January 3, 2002 (received for review October 1, 2001) The metallorganic chemical vapor deposition method was successfully used to synthesize pure TiO2 and Nd3+-, Pd2+-, Pt4+-, and Fe3 +-doped TiO nanoparticles. Polycrystalline TiO structure was verified with x-ray diffraction, which showed typical characteristic 2 2 anatase reflections without any separate dopant-related peaks. Transmission electron microscopy observations confirmed the existence of homogeneously distributed 22±3 nm TiO2 nanoparticles. The particle size remained the same for the doped samples. The doping level of transition metals was kept at ≈1 atomic percent, which was determined by x-ray photoelectron spectra and energy dispersive xray spectroscopy. The effects of different types of dopants on the photocatalytic activity were revealed by the degradation of 2chlorophenols with an UV light source. The photocatalytic efficiency was remarkably enhanced by the introduction of Pd2+ and Nd3+. Nd3+-doped TiO2 showed the largest enhancement. However, Pt4+ changed the 2-chlorophenol degradation rate only slightly, and Fe3+ was detrimental to this process. These effects were related to the position of the dopants in the nanoparticles and the difference in their ionic radii with respect to that of Ti4+. Titanium dioxide occurs in three forms (rutile, anatase, and brookite), among which anatase is believed to be the most efficient photocatalyst during chemical reactions (1, 2). It has been extensively investigated for its catalytic and electrochemical properties based on its wide applications as photocatalyst and gas sensor. Most of the studies were focused on the nanosized TiO2 with the purpose of improving the light absorption. The high surface-to-volume ratio, inherent in nanoparticles, was useful. Additionally, the small size of TiO2 crystals can make indirect band electron transition possible and increase the generation rate of electrons and holes. Increase of the generation rate of charge carriers is one way to enhance the photocatalytic activity. On the other hand, electron and hole trapping during their transportation from the interior of the particle to the surface is also very crucial to preventing the recombination of electron and hole pairs. Doping of TiO2 with transition metal ions offers a way to trap charge carriers and extend the lifetime of one or both of the charge carriers. Consequently, dopants enhance the efficiency of the photocatalyst. The primary driving force in this research was to study the photocatalytic characteristics of TiO2 nanoparticles and the effect of transition metal dopants on TiO2 nanoparticles performance. In photocatalysis, it is the photon-generated electron-hole (e−/h+) pairs that can facilitate redox reactions on particle surface. The total number of free carriers on the surface determines the efficiency of catalysts. The number and the lifetime of free e−/h+ are particle size- and dopant-dependent. For large particles, the volume recombination of electrons and holes dominates. This condition largely reduces the number of free charges on the surface and deteriorates the photocatalytic activity. For nanoparticles, the transportation length of e−/h+ from crystal interface to the surface is short, which helps to accelerate the migration rate of e−/h+ to the surface of the nanoparticle to participate the reaction process. For optimal photocatalysis efficiency there is a critical particle size below which the surface recombination of electron and hole becomes dominant because of the increased surface-to-volume ratio (3). Besides the effect of particle size on the photocatalytic activity, the role of dopant is important. Different dopants may not have the same effect on trapping electrons and/or holes on the surface or during interface charge transfer because of the different positions of the dopant in the host lattice. Consequently, the photocatalytic efficiency would be different for different types of dopants. In this article, we describe the synthesis of Pt4+ nanoparticles and the effect of transition metal ion Nd3+, Pd2+, Ti4+, and Fe3+ dopants on the photocatalytic activity of TiO2 nanoparticles. The photoactivity is investigated by performing the degradation experiments of 2-chlorophenols (2-CP). The comparison of photodegradation rates for TiO2 with and without dopants is presented. Synthesis of TiO2 anatase nanocrystals has been achieved by several methods including impregnation (4), coprecipitation (5), sol-gel (6– 8), hydrothermal method (9–11), and chemical vapor deposition (CVD) (12, 13). In our study, a metallorganic CVD (MOCVD) process was used to prepare TiO2 nanoparticles. MOCVD is preferred because it requires no postdeposition calcination, centrifugation, or hydrothermal processing to crystallize or refine particles. Also, the size distribution of particles can be simply controlled by the temperature of substrates and the flow rates of the precursors. The introduction of dopants into TiO2 nanoparticles can be realized either by a solid source, which is directly put in the proper position in the reactor, or by a solution mixed in with the liquid Ti precursor. EXPERIMENTAL PROCEDURES Fig. 1 shows the schematics of the MOCVD system used for the preparation of TiO2 nanoparticles. All powder samples were collected on 5-cm diameter discs made of several layers of 475-mesh stainless steel screen, which were designed to obtain high collection efficiency. Before deposition, acetone, methanol, and deionized water were used to remove any native contami
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC. This paper was submitted directly (Track II) to the PNAS office. Abbreviations: e−/h+, electron-hole; 2-CP, 2-chlorophenols; CVD, chemical vapor deposition; MOCVD, metallorganic CVD; TTIP, titanium tetraisopropoxide; XPS, x-ray photoelectron spectroscopy; XRD, x-ray diffraction; TEM, transmission electron microscopy; at%, atomic percent. ‡To whom reprint requests should be addressed. E-mail: [email protected].
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
STUDY OF ND3+, PD2+, PT4+, AND FE3+ DOPANT EFFECT ON PHOTOREACTIVITY OF TIO2 NANOPARTICLES
6483
nation from the surface of the screens. The screens were held perpendicular to the flow of the reactants in the MOCVD chamber by a 3-cm diameter and 12.5-cm long quartz tube. The MOCVD chamber consisted of a 5-cm diameter and 75-cm length stainless steel tube with a central region that could be externally and uniformly heated to 1,000°C by a resistive heater. Fig. 2 shows the temperature profile of the reactor wall at 600°C. There is a 20-cm long central region and the temperature decreased toward both the ends of the reactor. The base pressure obtained by a mechanical pump was in the mTorr regime. The precursor used for Ti was Ti[OCH(CH3)2]4 (titanium tetraisopropoxide, TTIP, 97%). TTIP is a liquid at room temperature with a boiling point of about 232°C. TTIP was placed in a bubbling chamber that was supplied with 99.999% pure Ar as TTIP carrier gas. The temperature of the bubbling chamber and the flow rate of Ar determined the precursor flow rate, which was adjusted by changing the temperature of the bubbling chamber and/or the flow rate of the Ar through the bubbling chamber. In addition to Ar and TTIP, O2 was introduced in the MOCVD chamber to participate in the chemical reaction to form TiO2. For the purpose of doping, neodymium (III) acetylacetonate, palladium (II) acetylacetonate, platinum (IV) acetylacetonate, and iron (III) acetylacetonate in powder forms were used for Nd3+, Pd2+, Pt4+, and Fe3+ ion doping, respectively. These dopant precursors were put in a Cu container, which was directly placed in the chamber. No Cu-related contamination peak was observed during the x-ray photoelectron spectroscopy (XPS) analysis. The use of Cu container does not introduce any Cu impurity because the melting point of Cu and its related oxide is much higher than the temperature at which the Cu container is used during synthesis. The incorporation rate of dopant was determined by the vapor pressure of the dopant precursor. The dopant concentration as a function of the dopant precursor temperature was measured. Thus, to obtain certain desired dopant concentration in TiO2, dopant precursor can be placed at a predetermined position of the chamber. We used the same deposition conditions for all sample preparations in this study (see Table 1). The partial pressures of O2 and Ar were 10 Torr and 1 Torr, respectively. The temperatures for substrate and TTIP precursor were 600°C and 220°C, respectively.
Fig. 1. Schematics of MOCVD system.
Fig. 2. Temperature profile of the reactor wall at 600°C.
X-ray diffraction (XRD) analyses were carried out for all samples. XRD θ–2θ scans were recorded by using Cu Kα radiation in a Rigaku (Tokyo) D-Max B diffractometer equipped with a graphite crystal monochromator for structural characterization of the polycrystalline doped and undoped TiO2 samples. XFIT software (freeware from http://www.ccp14.ac.uk/tutorial/xfit-95/) was used to measure the precise 2θ positions and the full width at half maxima of the diffraction peaks. XPS and energy dispersive x-ray spectroscopy were performed for composition determination of the samples. A SSI-M Probe XPS was used with Al Kα exciting radiation. High-resolution scans of the Ti 2p, dopant 3d, O 1s, and C 1s regions were obtained, in addition to the survey scans, to measure the composition of the nanoparticles and to verify the valance states of Ti and the dopants. Amray (Bedford, MA) 1810T scanning electron microscopy and a JEOL 2000 FX transmission electron microscopy (TEM) were used to observe the TiO2 surface morphology and to measure the average nanoparticle size and distribution, respectively. Photodegradation experiments were performed in the photocatalytic reactor system. This bench-scale system consisted of a cylindrical Pyrex-glass cell 20 cm in diameter and 30 cm in height with an inside reflective surface. A 100-W Hg lamp was used and immersed in the solution. A cold air-cooling jacket cooled the cell. The maximum energy emission at the wavelength of 365 nm
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
STUDY OF ND3+, PD2+, PT4+, AND FE3+ DOPANT EFFECT ON PHOTOREACTIVITY OF TIO2 NANOPARTICLES
6484
was achieved 4 min after the lamp was turned on. At the cooling jacket, an energy density of 5.3 mW/cm2 was measured. An aqueous solution (1,000 ml) of 2-CP and the stainless steel mesh with the TiO2 nanoparticles were placed in the photoreactor cell. After illumination, samples were collected at regular intervals in a test tube, and each sample solution was analyzed by HPLC (Waters model 6000 gradient system). The total organic carbon (TOC) of a sample solution was measured at constant irradiation time intervals by using a DC-190 high-temperature TOC analyzer. The Cl− ion was analyzed by ion chromatograph (Dionex) equipped with an electrochemical detector and a Dionex PAX-100 metalfree anion column (25 cm long, 4.6 mm i.d.). The eluent solution was a mixture of 80% H2O, 10% acetonitrile, and 10% 191-mM NaOH. The flow rate was 1 ml/min, and the injection loop volume was 50 µl Table 1. Deposition conditions, ionic radii of transition metals for a coordination number of 6(16), particle sizes obtained from XRD and TEM as well as its histogram, and photodegradation time to achieve 90% destruction of 2-CP for pure TiO2 and Nd3+-, Pd2+-, Pt4+-, Fe3+-doped TiO2 nanoparticles Pure TiO2 TiO2(Pt4+) TiO2(Fe3+) TiO2(Pd2+) TiO2(Nd3+) Sample Substrate temperature (°C) 600 600 600 600 600 TTIP temperature (°C) 220 220 220 220 220 10 10 10 10 10 O2 pressure (Torr) Ar pressure (Torr) 1.0 1.0 1.0 1.0 1.0 Position of dopant precursor (cm)* — 18 20 19 21.5 † — 0.625 0.645 0.86 0.983 Dopant ionic radii (Å) Particle size from XRD (nm) 27±3 28±3 28±3 27±3 27±3 Particle size from TEM (nm) 23±3 23±3 24±2 23±3 22±3 Particle size from histogram (nm) 23 23 23 22 22 60 50 90 40 25 Time for destruction of 90% 2-CP (min) *The position refers to the distance from center of the reactor (see Fig. 2 for the temperature profile of the reactor wall at 600°C). †Cited from ref . 16.
The activity of the photocatalytic decomposition of 2-CP was again estimated from the yield of carbon dioxide, determined, gravimetrically as BaCO3, from the yield of carbon dioxide as decreasing results of electric conductivity for Ba(OH)2 solution. HCO3− in a sample solution was measured by ion and liquid chromatography. RESULTS X-ray structure analysis for samples doped with different types of transition metal ions showed all of these samples had typical peaks of TiO2 poly crystalline anatase nanoparticle without any detectable dopant-related peaks. The dopants went either into the interstitial positions or substitutional sites of TiO2 crystal structure. Fig. 3 shows an x-ray diffraction pattern for Nd3+-doped TiO2 nanoparticles. The poly crystalline anatase structure was confirmed by (101) (004), (200), (105), and (211) diffraction peaks. Its tetragonal Bravais lattice type was also verified by lattice constant calculated from these peaks. Based on the full width at half maxima of the XRD peaks, the average particle diameter was calculated to be about 27±3 nm by using Scherrer's formula. There was no measurable effect on the size of TiO2 nanoparticles with the addition of different types of dopants.
Fig. 3. XRD pattern of ND-doped polycrystalline TiO2 nanoparticles.
XPS survey spectra for the same sample with Nd doping is shown in Fig. 4. Only peaks associated with Ti, O, and Nd were observed. The relative cation composition in the particles was determined by XPS and energy dispersive x-ray spectroscopy. The Nd concentration was measured to be about 1 atomic percent (at %). The magnified Ti 2p region is presented in Fig. 4 Inset. Two peaks located at 458.4 eV and 464.1 eV were identified as Ti (2p3/2) and Ti (2p1/2) (14), respectively. For metallic Ti0 these two peaks are expected at 455 and 459 eV (16). The peak shifts in Ti (2p3/2) and Ti (2p1/2) peak positions and the change in the separation between these two peaks is caused by the presence of tetravalent Ti4+, which is consistent with TiO2 formation (14, 15).
Fig. 4. XPS survey spectra of Nd-doped TiO2 nanoparticles and the magnified Ti 2p region.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
STUDY OF ND3+, PD2+, PT4+, AND FE3+ DOPANT EFFECT ON PHOTOREACTIVITY OF TIO2 NANOPARTICLES
6485
Fig. 5. Scanning electron microscopy image of TiO2 nanoparticles.
Fig. 5 shows a×10,000-magnified scanning electron microscopy image of Nd-doped TiO2 nanoparticles. The TiO2 particles are aggregated together and the rough measurements from this image indicate that the size of particles is in the nanoscale range. More precise size and distribution measurements were performed by TEM. A TEM bright-field image with the diffraction pattern as an inset is shown in Fig. 6. The particle size measured was about 22±3 nm, consistent with the grain size measured by XRD. The electron diffraction patterns correspond to that of anatase TiO2. Fig. 7 is the histogram of particle distribution. The distribution of the size of nearly spherical particles peaked at about 22 nm with a relatively small tail toward larger values. Fig. 8 is the comparison of the 2-CP photodegradation rates for pure TiO2 sample and Nd3+-, Pd2+-, Pt4+-, and Fe3+-doped TiO2 samples. In this study, the doping level in all four cases was kept constant at about 1 at %. The change of photocatalytic activity for doped samples is evident from the degradation curves. Moreover, the degradation rates for most doped samples have been enhanced with the exception of Fe3+doped sample. The time for 90% destruction of 2-CP has been reduced from 60 min for undoped TiO2 to 25 min for Nd3+-doped TiO2 nanoparticles. DISCUSSION The total number of free charge carriers on the TiO2 surface is determined by the rate of charge pair generation, charge trapping, charge release and migration, charge recombination as well as the rate of interfacial charge transfer. The complexities of the role of transition metal ion dopants are that they can participate in all of these processes. Acting as electron and/or hole traps is the most important function of dopants. The trap of charge carriers can decrease the recombination rate of e−/h+ pairs and consequently increase the lifetime of charge carriers. The process of charge trapping is as follows:
Fig. 6. TEM bright-field image of nanosized polycrystalline TiO2 particles and their diffraction pattern. Fig. 7. Size distribution histogram of TiO2 nanoparticles deposited at 600°C.
where Mn+ is the metal ion dopant. The energy level of Mn+/M(n−1)+ lies below the conduction band edge and the energy level of Mn+/M(n+1) lies above the valence band edge. Thus, the energy level of transition metal ions affects the trapping efficiency. The trapping of electrons makes it easy for holes to transfer onto the surface of TiO2 and react with OH− in the 2-CP solution and form active OH−, hydroxyl radicals to participate the destruction of 2-CP. +
Fig. 8. Photodegradation of 2-CP with undoped TiO2 and transition metal ion- (Nd3+, Pd2+, Pt4+, and Fe3+) doped TiO2 under an UV light source. C0=50 mg in 1,000 ml at pH 9.5.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
STUDY OF ND3+, PD2+, PT4+, AND FE3+ DOPANT EFFECT ON PHOTOREACTIVITY OF TIO2 NANOPARTICLES
6486
2-CP degradation is an oxidation reaction for which the lifetime of the holes is critical. The lifetime of holes can be enhanced by trapping electrons, thereby reducing the recombination rate and allowing holes to diffuse to the particle surface and participate in the oxidation reaction. If the energy level of dopant ions moves toward the conduction band edge, the efficiency of trapping becomes higher. In that case, the traps have larger tendency to act as shallow traps so that the holes generated by following photons cannot recombine with the already trapped electrons. Consequently, the lifetime of free holes can be extended. However, for transition metal ions in interstitial and substitutional positions, the change of their potential level is different and they have different trapping efficiencies for electrons. The effect of the interstitials in distorting the potential energy is larger than that of substitutional atoms. According to the XRD results of doped TiO2, only anatase phase was present in the TiO2 nanoparticles and no separate dopant-related phase was observed. This finding indicates that dopant ions Nd3+, Pd2+, Pt4+, and Fe3+ are either in the octahedral interstitial sites or the substitutional positions of TiO2. From the effective radii of ions for coordination number of 6 (16) in Table 1, one can see Fe3+ and Pt4+ have small ionic radii that are comparable to that of Ti4+, 0.605 Å. It is energetically favorable for these two ions to occupy Ti4+ sites. Nd3+, on the other hand, has an ionic radius that is much larger than that of Ti4+. It is energetically favorable for Nd3+ to reside in the octahedral interstitial site. The size of octahedral interstitial is large enough, 3.78 and 9.51 Å for a and c axes, respectively, to accommodate Nd3+. Pd2+ also has relatively larger ionic radius and can go into the interstitial positions. Substitutionally incorporated dopants are less useful for disturbing the dopant energy level, which in turn affects the electron trapping efficiency. Pt doping shows reduced enhancement compared with Nd and Pd doping, whereas Fe doping shows lack of enhancement. There have been previous studies that show similar results for Fe on the photodegradation of the vinyl chloride (5). Another contributing reason for the lack of photocatalytic activity enhancement by Fe may include nonoptimal valency of Fe. The accurate determination of the Fe oxidation state when it is present in such low concentrations is difficult because of the problem in satisfactorily resolving the Fe 3p region of the XPS spectrum. For the interstitial dopants, the situation is different and the potential energy level can be increased negatively toward the conduction band edge because of the oxygen affinity of dopant ions. This is the case for Nd3+ and Pd2+. The high oxygen affinities of interstitially located Nd3+ and Pd2+ ions effectively create a localized positive charge around Ti and/or form an oxygen vacancy. The potential energy of ion dopants was disturbed and electrons were efficiently trapped. Consequently, the oxidation process of chlorophenols was remarkably improved. Currently, the doped samples had 1 at% of dopant concentrations. TiO2 nanoparticles have applications in photovoltaics, paints, etc. The enhanced performance of doped TiO2 as a photocatalyst suggests that its performance for other applications also may be improved. However, these particles have to be analyzed specifically for the pertinent application. CONCLUSIONS Crystalline doped and undoped TiO2 anatase nanoparticles (22±3 nm) were successfully synthesized by using MOCVD at 600°C substrate temperature and 220°C precursor solution temperature. The effect of the different type dopants, Nd3+, Pd2+, Pt4+, and Fe3+, with similar concentration (≈1 at%) on the photocatalytic efficiency was investigated by performing the photodegradation of 2-CP under radiation of UV light. Results showed that the efficiency of photocatalyst was remarkably enhanced for Nd3+- and Pd2+-doped TiO2. In particular, the 90% destruction time of 2-CP was reduced from 60 min for the undoped TiO2 nanoparticles to 25 min for the 1 at% Nd-doped nano TiO2. On the other hand, Pt4+ caused only a slight decrease in the 2-CP oxidation time and Fe3+ did not help at all. Possible explanations could be based on the change of potential energy level for different positions of dopants in TiO2 lattice. The position of dopants is determined by the size differences between the host Ti4+ ionic radius and the dopant's ionic radii. Nd3+ and Pd2+ acted as interstitial dopants. Large disturbance of the potential energy resulted in the creation of localized positive charge around Ti and/or formation an oxygen vacancy, which consequently enhanced the electron trapping efficiency. However, Pt4+ and Fe3+ were presumably in substitutional positions with very little or no potential energy disturbance. They were less useful for the electron trapping. More experiments, including near edge x-ray absorption fine structure and XPS, are necessary to locate the position of dopants in TiO2 lattice.
1. Sclafani, A., Palmisano, L. & Davi, E. (1991) J. Photochem. Photobiol. A 56, 113–123. 2. Vidal, A., Herrero, J., Romero, M., Sanchez, B. & Sanchez, M. (1994) J. Photochem. Photobiol. A 79, 213–219. 3. Beydoun, D., Amal, R., Low, G. & McEvoy, S. (1999) J. Nanoparticle Res. 1, 439–458. 4. Litter, M.I. & Navio, J.A. (1994) J. Photochem. Photobiol. A 84, 183–193. 5. Palmisano, L., Augugliaro, V., Sclafani, A. & Schiavello, M. (1988) J. Phys. Chem. 92, 6710–6713. 6. Choi, W., Termin, A. & Hoffermann, M.R. (1994) J. Phys. Chem. 98, 13669–13679. 7. Zhang, Z., Wang, C.-C, Zakaria, R. & Ying, J.Y. (1998) J. Phys. Chem. B 102, 10871–10878. 8. Wang, C.-C., Zhang, Z. & Ying, J.Y. (1997) NanoStructured Mater. 9, 583–586. 9. Wang, Y., Cheng, H., Hao, Y., Ma, J., Li, W. & Cai, S. (1999) J. Mater. Sci. 34, 3721–3729. 10. Cheng, H., Ma, J., Zhao, Z. & Qi, L. (1995) Chem. Mater. 7, 663–671. 11. Wang, Y., Hao, Y., Cheng, H., Ma, J., Xu, B., Li, W. & Cai, S. (1999) J. Mater. Sci. 34, 2773–2779. 12. Ding, Z., Hu, X., Lu, G.Q., Yuc, P.-L. & Greenfield, P.F. (2000) Langmuir 16, 6216–6222. 13. Schrijnemakers, K., Impens, N.R.E.N. & Vansant, E.F. (1999) Langmuir 15, 5807–5813. 14. Wagner, C.D., Riggs, W.M., Davis, L.E., Moulder, J.F. & Muilenberg, G.E., eds. (1979) Handbook of X-Ray Photoelectron Spectroscopy (PerkinElmer, Eden Prairie, MN). 15. Sen, S.K., Riga, J. & Verbist, J. (1976) Chem. Phys. Lett. 39, 560–564. 16. Shannon, R.D. (1976) Acta Crystallogr. A 32, 751–767.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
ENTROPICALLY DRIVEN SELF-ASSEMBLY OF MULTICHANNEL ROSETTE NANOTUBES
6487
Colloquium Entropically driven self-assembly of multichannel rosette nanotubes
Hicham Fenniri*†, Bo-Liang Deng*, Alexander E.Ribbe*, Klaas Hallenga*, Jaby Jacob‡, and Pappannan Thiyagarajan‡ *1393 Herbert C.Brown Laboratory of Chemistry, Purdue University, West Lafayette, IN 47907–1393; and ‡Argonne National Laboratory, Intense Pulsed Neutron Source Division, 9700 South Cass Avenue, Argonne, IL 60439 Edited by Julius Rebek, Jr., The Scripps Research Institute, La Jolla, CA, and approved December 4, 2001 (received for review October 2, 2001) Rosette nanotubes are a new class of organic nanotubes obtained through the hierarchical self-assembly of low molecular weight synthetic modules in water. Here we demonstrate that these materials can serve as scaffolds for the supramolecular synthesis of multichannel nanotubular architectures and report on the discovery of their entropy-driven self-assembly process. Unidimensional nanotubular objects have captivated the minds of the scientific community over the past decade because of their boundless potential in nanoscale science and technology. The strategies developed to achieve the synthesis of these materials spanned the areas of inorganic (1–5) and organic (6–12) chemistry and resulted in, for instance, carbon nanotubes (1), peptide (9–11), and rosette nanotubes (12), as well as surfactant-derived tubular architectures (13–20). Although inorganic systems benefit from the vast majority of the elements of the periodic table and the rich physical and chemical properties associated with them, organic systems inherited the power of synthetic molecular (21, 22) and supramolecular (23, 24) chemistry. As such, the latter approach offers limitless possibilities in terms of structural, physical, and chemical engineering. Here, we present the design, self-assembly, and characterization of multichannel organic nanotubes in water and the discovery of their apparently entropy-driven self-assembly. DESIGN AND SYNTHESIS The heteroaromatic bicyclic base G∧C (Scheme 1), possessing the Watson-Crick donor-donor-acceptor of guanine and acceptor-acceptordonor of cytosine, was recently reported in the context of the self-assembly of helical rosette nanotubes (12). Because of the disymmetry of its hydrogen bonding arrays, their spatial arrangement, and the hydrophobic character of the bicyclic system, G∧ ∧C undergoes a hierarchical selfassembly process under physiological conditions to form a six-membered supermacrocycle maintained by 18 H-bonds (Fig. 1 Upper, thin solid bars). The resulting and substantially more hydrophobic aggregate then undergoes a second level of organization to produce a stack. The architecture thus generated defines an unoccluded central pore running the length of the stack with tunable inner and outer diameters (Fig. 1 Lower). The inner space is directly related to the distance separating the H-bonding arrays within G∧ ∧C, whereas the peripheral diameter and its chemistry are dictated by the choice of the functional groups conjugated to this motif.
Scheme 1. G∧ ∧C motif and compounds 1 and 2 synthesized. Detailed experimental procedures for the preparation of all of the intermediates and final compounds along with the corresponding spectroscopic characterizations are in the supporting Methods and Fig. 6.
In addition to their demonstrated synthetic accessibility and broad solvent compatibility, crown ethers are a very versatile class of receptors that display size, shape, and charge selectivity toward their guests (25). Furthermore, extensive investigations were carried out to establish their ionophoric properties (25) and incorporate them in artificial channel systems (26, 27), molecular photonic (28–30), and electronic (31, 32) devices. In the present article, 1 and 2 were synthesized and investigated to establish the rosette nanotubes as stable, yet noncovalent scaffolds for the self-organization of multichannel assemblies (Fig. 1). We also demonstrate that the presence of cations known to coordinate to crown ethers does not affect the one-dimensional organization, thereby reinforcing the potential of these materials in the design of iono-, photo-, and electro-active nanotubes and nanowires as well as in the preparation of selective ion channels. While investigating the mechanism of their formation, we also have uncovered their peculiar response to temperature, which strongly supports the entropic nature of the assembly process. A synthetic scheme was devised to allow for the preparation of 1 and 2 and the functionalization at virtually any position (Scheme 1). The versatility of the synthetic strategy allows for the incorporation of a wide range of crown ether hosts differing in dimensions and properties. All of the compounds were characterized by 1H NMR, 13C NMR, and MS. In addition, the key intermediates were characterized by high-resolution MS and
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC. This paper was submitted directly (Track II) to the PNAS office. Abbreviations: NOE, nuclear Overhauser effect; DLS, dynamic light scattering; SAXS, small angle x-ray scattering; TEM, transmission electron microscopy; 2D, two-dimensional; FT, Fourier transform; fb, flip-back. †To whom reprint requests should be addressed. E-mail: [email protected].
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
ENTROPICALLY DRIVEN SELF-ASSEMBLY OF MULTICHANNEL ROSETTE NANOTUBES
6488
elemental analysis (see additional Methods and Fig. 6, which are published as supporting information on the PNAS web site, www.pnas.org).
Fig. 1. Hierarchical self-assembly of compound 2 into a multichannel rosette nanotube (MACROMODEL, version 7.2) (see supporting Methods).
NMR STUDIES One-dimensional experiments (600 MHz, 90% H2O/D2O) between 2°C and 42°C indicated broadening at all but the lowest temperatures. A major and a minor population of imino protons coexist with a ratio of 3:1, which is independent of concentration (0.001 to 0.025 M) and temperature (2–42°C) (see supporting Methods and Fig. 7, which is published as supporting information on the PNAS web site). Flip-back total correlation spectroscopy (fb-TOCSY), nuclear Overhauser effect (NOE) spectroscopy (NOESY), and rotating frame Overhauser effect spectroscopy (ROESY) experiments were then carried out on the same solution to establish the self-assembly of the rosette supermacrocycle and to gain further information about the nature of the two coexisting species. fb-TOCSY experiments did not show any cross-peak between the imino protons, except an exchange NOE between the minor and major AH resonances. fb-NOESY experiments showed the expected crosspeaks between, respectively, the major and minor BH and CH resonances, which were confirmed by fb-ROESY as nonchemical exchange NOEs. Because BH and CH are too far apart to display any intramodular NOEs, the observation of an NOE between them corroborates the formation of intermodular H-bonds as highlighted in Fig. 1 (Upper, thick solid bars). The fb-NOESY also revealed the expected cross-peak between the minor AH and BH but not between the major AH and BH resonances, thereby suggesting that the major CH3AH protons are in a nonhydrogen-bonded conformation. Such conformation apparently would place CH3AH away from BH (see supporting Methods and Figs. 8 and 9, which are published as supporting information on the PNAS web site). VARIABLE TEMPERATURE UV-VISIBLE STUDIES The electronic spectra of 1 and 2 display a profile typical of the DNA bases, with two maxima at 238 nm and 287 nm. A strong indicator of DNA stability and tertiary structure is the hyperchromicity observed upon its denaturation (33). In this regard the supramolecular outcome of 1 and 2 was anticipated to behave similarly as was shown in the case of a lysine-substituted module (12). This hypothesis did not hold in the case of 1 and 2. The spectra displayed a linear, noncooperative, irreversible, and small hyperchromic effect in the range of 25°C to 95°C, regardless of the presence of sodium or potassium (up to 10 equivalents). Furthermore, the UV-visible spectra of 4-aminobenzo-18-crown-6ether (18C6) remain essentially unaffected in this temperature range with or without potassium (see supporting Methods, Fig. 10, and Table 2, which are published as supporting information on the PNAS web site). To rationalize this data we postulated that the supramolecular outcome of 1 and 2 must be unusually stable in the temperature range of 25°C to 95°C, despite its noncovalent nature. Variable temperature dynamic light scattering (DLS), small angle x-ray scattering (SAXS), and transmission electron microscopy (TEM) proved to be the ideal tools to probe this hypothesis. DLS AND SAXS STUDIES To assess the hydrodynamic dimensions (length and diameter) of the proposed nanotubes, DLS and SAXS measurements were carried out on dilute aqueous solution of 1 and 2 (see supporting Methods and Fig. 11, which is published as supporting information on the PNAS web site). A narrow mono-modal distribution (>90% in the range of 12 to 42 nm) with an average apparent hydrodynamic radius, RH, of 16.0 nm for 1 and 26.7 nm for 2 was recorded by DLS at 20°C. The 15C5 crown ether of compound 1 is known to bind Na+ selectively. DLS and SAXS studies (Table 1) performed on 1 in the absence and presence of Na+ (up to 10 equivalents) indicate that the latter does not alter the dimensions of the resulting assembly in solution (RH=16 nm, diameter ≈4 nm). The same conclusions can be drawn for compound 2 on the basis of the DLS, SAXS, and TEM data. Finally, owing to the hydrophobic character of the nanotubes' core, the consistently lower diameter derived from the SAXS data may reflect the more compact state of the assembly in water. Note also that the TEM measurements include a layer of the staining agent, whereas the computed diameter takes into consideration the energy-minimized (fully expanded) conformation of the crown ethers. The most impressive and unexpected observation was the peculiar and unprecedented response of these materials to temperature. As the latter increased so did the length of the nanotubes. In the range of 20°C to 40°C, the average hydrody
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
ENTROPICALLY DRIVEN SELF-ASSEMBLY OF MULTICHANNEL ROSETTE NANOTUBES
6489
namic radius grew over 320% for both 1 and 2 (Fig. 2). Above 40°C the increased scattering from larger assemblies saturated the detector of the instrument used. Table 1. Dimensions of the nanotubes generated from 1 and 2 determined by DLS, SAXS, and TEM DLS,* RH, nm SAXS,† diameter, nm TEM,‡ diameter, nm 1 16.0 3.6±0.66 3.9 (3.9) 1+NaCl 17.0 4.13±0.70 4.2 (4.1) 1+KCl 17.0 3.71±0.47 4.0 (3.9) 2 26.7 — 4.5 (4.4) 2+NaCl 27.0 3.5±0.09¶ 4.4 (4.5) 2+KCl 27.0 3.5±0.09¶ 4.4 (4.4)
Computed§ diameter, nm 4.0 4.0 4.0 4.3 4.3 4.3
*[1]=[2]=[NaCl]=[KCl]=2×10−3 M in water. †[1]=[2]=[NaCl]=[KCl]=1.3×10 −3 M in water at 20°C. ‡[1]=[2]=[NaCl]=[KCl]=2×10−4 M in water. The numbers in parentheses were recorded for the same aqueous samples after incubation at 70°C for 30 min in a thermostated bath. All the outer diameters include a layer of the staining agent and are each an average of 100 measurements made on randomly selected nanotubes. These numbers were also confirmed by using 2D-FT of densely packed nanotubes (see Fig. 5). §Computer models generated by using MACROMODEL 7.2. ¶Measurement performed in 10 mM Mes buffer, pH 5.5. This solution contains a 5-fold excess of sodium and potassium with respect to 2. In the absence of the Mes buffer the data was not reproducible.
As most self-assembly processes involving noncovalent interactions are enthalpically driven and entropically unfavorable, the increased and controlled level of aggregation as the temperature increases was the most significant observation in this study. This behavior is reminiscent of the hydrophobic effect, in which release of ordered water to the bulk solvent contributes to the solute's self-association. Entropically driven self-assembly processes in aqueous solutions are well known in natural systems (34–38), for colloidal microparticles (39), and in few instances for small molecules in organic solvents (40, 41). There is, however, no precedent of such process for well-defined nanoscale synthetic assemblies in water. In the present case, we believe that ordered water molecules on the hydrophobic surfaces of the bases located on each end of the nanotubes are released to the bulk solvent as the temperature increases while new rosette stacks are being recruited. This process would result in elongated nanotubes, in agreement with the DLS data (Fig. 2). TEM STUDIES TEM provided us with visual evidence of the formation of the proposed nanotubular assemblies, a confirmation of their dimensions, and the temperature effect recorded by DLS. Fig. 3 shows negatively stained samples of the nanotubes derived from 1 and 2. Detailed analysis of these micrographs led to the following conclusions: (i) all of the assemblies have the same outer diameter (≈4 nm), in agreement with the computed one (Table 1). (ii) Sodium or potassium did not alter the dimensions nor the tubular organization, also in agreement with the DLS and SAXS data. (iii) Temperature annealing does not affect the nanotubes' diameter. (iv) In agreement with the variable temperature DLS studies, the density of the nanotubular assemblies increased dramatically with temperature, thereby highlighting the entropic nature of the selfassembly process. In effect, as macromolecular systems enter the high nanometer to low micrometer regime, attractive long-range, interparticle forces (also called depletion forces or colloid interactions) (39) become relatively significant, thereby leading once again to release of ordered solvent molecules to the bulk and resulting in a new level of aggregation under the effect of temperature (phase separation). This property is fundamental to colloid science, namely, there is always an attractive force between like particles in solution, tending to induce aggregation (42). This is also the essence of van der Waals forces applied to aggregates of molecules, rather than between individual molecules.
Fig. 2. DLS regularization diagrams of 40×10−6 M solutions of 1 (Upper) and 2 (Lower) in H2O at 20°C, 30°C, and 40°C. The average hydrodynamic radii recorded are shown (see supporting Methods).
Concentrated samples of 1 or 2 generally result in higher density areas of nanotubes on the carbon-coated TEM grids that allowed us to extract information about their hollow nature, as well as their propensity to undergo a grid-like organization over multiple layers. Shown in Fig. 4 is an area of the TEM image where hollow, disk-shaped entities are identified. A remarkable feature of this image is that these objects appear to have been borne out of a parent nanotube as most of them seem to follow a linear organization. Furthermore, judging from their dimensions, we are indeed compelled to conclude that these entities are the rosette components that make up the nanotubular assemblies. Finally, two-dimensional (2D)-Fourier transform (FT) studies of the TEM images of densely packed rosette nanotubes (Fig. 5) allowed us to reveal their propensity to pack into multiple-layer thin films with quasi-perpendicular grid-like organization, thereby offering a new level of control over the supramolecular outcome of compounds 1 and 2. CONCLUSION The self-assembly of type I collagen fibrils (34, 35), the polymerization of the coat protein of the tobacco mosaic virus (36), and the selfassembly of bovine brain tubulin (37) or β-amyloids
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
ENTROPICALLY DRIVEN SELF-ASSEMBLY OF MULTICHANNEL ROSETTE NANOTUBES
6490
Fig. 3. Transmission electron micrographs of negatively stained samples of 1 and 2 (Top, left to right), 1−NaCl and 2+KCl (Middle, left to right), and 2 annealed in the absence and presence of KCl (Bottom, left to right). (Scale bar=40 nm.) See supporting Methods.
(38) are just a few classical examples of entropy-driven processes from nature. The discovery of the same behavior in the process of rosette nanotubes' self-assembly offers opportunities not only for in-depth structural and thermodynamic studies of this phenomenon, but also an additional tool in supramolecular engineering. Indeed, this study has many implications on our ability to control the elusive hydrophobic effect, as it is a demonstration that H-bonds can be recruited to orchestrate an entropically fueled process.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
ENTROPICALLY DRIVEN SELF-ASSEMBLY OF MULTICHANNEL ROSETTE NANOTUBES
6491
Fig. 4. Transmission electron micrograph featuring top views of the nanotubes and revealing their hollow nature. (Inset) Three-dimensional reconstructions of the gray value profiles derived from the TEM images by using the National Institutes of Health program IMAGE. (Scale bar=40 nm.) See supporting Methods.
This article establishes that nanotubular constructs resulting from the self-assembly of the G∧ ∧C motif can serve as noncovalent, yet very stable, scaffolds for the supramolecular synthesis of multichannel assemblies and unfolds the generality of the hierarchical self-assembly approach to functional nanotubular assemblies with predefined properties. In effect, further elaboration of the outer surface of this motif can result in unidimensional assemblies with predefined chemical and physical properties.
Fig. 5. 2D-FT analysis of a densely packed multiple layer thin film of nanotubes generated from 1. Transmission electron micrograph of multiple layers of nanotubes obtained from 1 (magnification: ×35,000) and original 2D-FT (Inset) (Upper Left). Inverse 2D-FT of masked 2D-FT (Inset) revealing the first densely packed nanotube layer oriented horizontally (Upper Right). Inverse 2D-FT of masked 2D-FT (Inset) revealing a second nanotube layer oriented vertically (Lower Right). 2D-FT profile along the x and y axes showing the quasi-perpandicularity of the two layers (Lower Left).
We acknowledge the support of the National Science Foundation (Career Award), the American Cancer Society (Grant ACS-IRG 58), the American Chemical Society-Petroleum Research Fund, the Showalter Foundation, the 3M Company, and Purdue University. H.F. is a Cottrell Scholar of Research Corporation. 1. Iijima, S. (1991) Nature (London) 354, 56–58. 2. Hamilton, E.J.M., Dolan, S.E., Mann, C.M., Colijin, H.O., McDonald, C.A. & Shore, S.G. (1993) Science 260, 659–661. 3. Brumlik, C.J. & Martin, C.R. (1991) J. Am. Chem. Soc. 113, 3174–3175. 4. Shenton, W., Douglas, T., Young, M., Stubbs, G. & Mann, S. (1999) Adv. Mater. 11, 253–256. 5. Chopra, N.G., Luyken, R.J., Cherrey, K., Crespi, V.H., Cohen, M.L., Louie, S.G. & Zettl, A. (1995) Science 269, 966–967. 6. Harada, A., Li, J. & Kamachi, M. (1993) Nature (London) 364, 516–518. 7. Nelson, J.C., Saven, J.G., Moore, J.S. & Wolynes, P.G. (1997) Science 277, 1793–1796. 8. Ashton, P.R., Brown, C.L., Menzer, S., Nepogodiev, S.A., Stoddart, J.F. & Williams, D.J. (1996) Chem. Eur. J. 2, 580–591. 9. Bong, D.T., Clark, T.D., Granja, J.R. & Ghadiri, M.R. (2001) Angew. Chem. Int. Ed. 40, 988–1011. 10. Ranganathan, D., Lakshmi, C. & Karle, I.L. (1999) J. Am. Chem. Soc. 121, 6103–6107. 11. Seebach, D., Matthews, J.L., Meden, A., Wessels, T., Baerlocher, C. & McCusker, L.B. (1997) Helv. Chim. Acta 80, 173–181. 12. Fenniri, H., Mathivanan, P., Vidale, K.L., Sherman, D.M., Hallenga, K., Wood, K.V. & Stowell, J.G. (2001) J. Am. Chem. Soc. 123, 3854–3855. 13. Schnur, J.M. (1993) Science 262, 1669–1676. 14. Nakashima, N., Asakuma, S. & Kunitake, T. (1985) J. Am. Chem. Soc. 107, 509–510. 15. Fuhrhop, J.-H., Spiroski, D. & Boettcher, C. (1993) J. Am. Chem. Soc. 115, 1600–1601. 16. Frankel, D.A. & O'Brien, D.F. (1994) J. Am. Chem. Soc. 116, 10057–10069. 17. Imae, T., Takahashi, Y. & Maramatsu, H. (1992) J. Am. Chem. Soc. 114, 3414–3419. 18. Kimizuka, N., Fujikawa, S., Kuwahara, S., Kunitake, T., Marsh, A. & Lehn, J.-M. (1995) Chem. Commun. 2103–2104. 19. Klok, H.-A., Joliffe, K.A., Schauer, C.L., Prins, L.J., Spatz, J.P., Möller, M., Timmerman, P. & Reinhoudt, D.N. (1999) J. Am. Chem. Soc. 121, 7154–7155. 20. Choi, I.S., Li, X., Simanek, E.E., Akaba, R. & Whitesides, G.M. (1999) Chem. Mat. 11, 684–690. 21. Corey, E.J. & Cheng, X.-M. (1989) The Logic of Chemical Synthesis (Wiley, New York). 22. Nicolaou, K.C. & Sorensen, E.J. (1996) Classics in Total Synthesis (VCH, New York). 23. Whitesides, G.M., Simanek, E.E., Mathias, J.P., Seto, C.T., Chin, D.N., Mammen, M. & Gordon, D.M. (1995) Acc. Chem. Res. 28, 37–44. 24. Prins, L.J., Reinhoudt, D.N. & Timmerman, P. (2001) Angew. Chem. Int. Ed. 40, 2382–2426. 25. Dietrich, B., Viout, P. & Lehn, J.-M. (1993) Macrocyclic Chemistry: Aspects of Organic and Inorganic Supramolecular Chemistry (VCH, Weinheim, Germany). 26. Biron, E., Voyer, N., Meillon, J.C., Cormier, M.-E. & Auger, M. (2000) Biopolymers Pep. Sci. 55, 3364–3372. 27. Roks, M.F.M. & Nolte, R.J.M. (1992) Macromolecules 25, 5398–5407. 28. Engelkamp, H., Middelbeek, S. & Nolte, R.J.M. (1999) Science 284, 785–788. 29. Meillon, J.-C., Voyer, N., Biron, E., Sanschagrin, F. & Stoddart, F. (2000) Angew. Chem. Int. Ed. 39, 143–145. 30. Balzani, V., Credi, A. & Venturi, M. (1998) Coord. Chem. Rev. 171, 3–16. 31. Balzani, V., Credo, A., Raymo, F.M. & Stoddart, J.F. (2000) Angew. Chem. Int. Ed. 39, 3348–3391. 32. Ballardini, R., Balzani, V., Credi, A., Gandolfi, M.T. & Venturi, M. (2001) Acc. Chem. Res. 34, 445–455.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
ENTROPICALLY DRIVEN SELF-ASSEMBLY OF MULTICHANNEL ROSETTE NANOTUBES
33. Cantor, C.R. & Schimmel, P.R. (1980) Biophysical Chemistry (Freeman, New York). 34. Kadler, K.E., Hojima, Y. & Prockop, D.J. (1987) J. Biol. Chem. 262, 15696–15701. 35. Kadler, K.E., Hojima, Y. & Prockop, D.J. (1988) J. Biol. Chem. 263, 10517–10523. 36. Riedhoff, P., Schneider, A., Mandelkow, E.-M. & Mandelkow, E. (1998) Biochemistry 37, 10223–10230. 37. Karr, T.L. & Purich, D.L. (1980) Biochem. Biophys. Res. Commun. 95, 1885–1889. 38. Shalaby, R.A. & Lauffer, M.A. (1985) Arch. Biochem. Biophys. 236, 390–398. 39. Yodh, A.G., Lin, K.-H., Crocker, J.C., Dinsmore, A.D., Verma, R. & Kaplan, P.D. (2001) Philos. Trans. R. Soc. London A 359, 921–937. 40. Kang, J. & Rebek, J., Jr. (1996) Nature (London) 382, 239–241. 41. Schmuck, C. (2001) Tetrahedron 57, 3063–3067. 42. Hemsley, A.R. & Griffiths, P.C. (2000) Philos. Trans. R. Soc. London A 358, 547–564.
6492
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
COMBINING CONSTITUTIVE MATERIALS MODELING WITH ATOMIC FORCE MICROSCOPY TO UNDERSTAND THE MECHANICAL PROPERTIES OF LIVING CELLS
6493
Colloquium Combining constitutive materials modeling with atomic force microscopy to understand the mechanical properties of living cells Mike McElfresh*†, Eveline Baesu*‡, Rod Balhorn*, James Belak*, Michael J.Allen§, and Robert E.Rudd* *Lawrence Livermore National Laboratory, Livermore, CA 94550; ‡Department of Engineering Mechanics, University of Nebraska, Lincoln, NE 68588; and §Biometrology, Incorporated, 851 West Midway Avenue, Alameda, CA 94501 Edited by Calvin F.Quate, Stanford University, Stanford, CA, and approved February 21, 2002 (received for review October 1, 2001) The goal of this work is to study the properties of living cells and cell membranes by using atomic force microscopy. During atomic force microscopy (AFM) measurement, there is a strong mechanical coupling between the AFM tip and the cell. The purpose of this paper is to present a model of the overall mechanical response of the cell that allows us to separate out the mechanical response of the cell from the local surface interactions we wish to quantify. These local interactions include recognition (or binding) events between molecules bound to an AFM tip (e.g., an antibody) and molecules or receptors on the cell surface (e.g., the respective antigen). The computational model differs from traditional Hertzian contact models by explicitly taking into account the mechanics of the biomembrane and cytoskeleton. The model also accounts for the mechanical response of the living cell during arbitrary deformation. The indentation of a bovine sperm cell is used to test the validity of this model, and further experiments are proposed to fully parameterize the model. The surface of the living cell is a highly complex heterogeneous structure containing a variety of lipid, protein, and carbohydrate components. The organization of the cell's exterior “sensing elements” and other specialized regions of the membrane is tailored to reflect the function of the cell and serves vital roles in cell-cell interactions, cell signaling, and cell-surface interactions. The changes that occur in these important chemical/mechanical phenotypes during the development of cancer and other diseases may be understood in much more detail, thereby allowing the relationships between specific phenotypes to cell and tissue normo- and pathophysiology, prognosis, and therapy to be discerned (1). Recent studies have shown that the components that comprise the membrane are segregated into domains that are dynamic and change in response to external and internal stimuli (2–5). This segregation appears to be controlled by a variety of factors, including the composition of the lipids, interactions with the cytoskeleton or extracellular matrix, and physical or structural barriers to diffusion (6–9). Although these barriers usually limit the random movement of receptors used in signaling and recognition and maintain them in a particular environment, proteins and carbohydrates are often relocalized and recruited into a particular region of the cell surface to facilitate cell function. In some cases, such as the sperm cell, dramatic changes in the composition of the membrane and the location and distribution of its proteins (receptors) occur throughout its development. In other cases, more subtle changes often occur later in the life of the cell and lead to cancer or other diseases, such as multiple sclerosis. Atomic force microscopy (AFM) has developed rapidly during the past decade, providing nanometer-scale resolution in the imaging of biological materials ranging in size from single molecules to intact cells. Although the best data have been obtained from studies of macromolecules (proteins, nucleic acids, and their complexes), AFM images of mouse and bull sperm have been obtained that rival the resolution of electron microscopy (EM) (10, 11). Unlike EM, however, AFM imaging can be performed in fluid on living cells. More recent developments in AFM now allow the detection of molecular recognition events between single molecules using ligands attached to AFM tips for the recognition of receptors bound on rigid surfaces (12–25). By monitoring the cantilever deflection during approachretraction cycles (i.e., force-volume/force-distance curves) at a constant (lateral) position on the sample, unbinding forces (i.e., the maximum force at the moment of receptor-ligand detachment) have been determined for various ligand-receptor pairs, including biotin-avidin (13, 14, 21), DNA bases (15), antibody-antigen (16–22), and cell-recognition proteins (23). This development has made it possible to use a single receptor molecule bound to the tip of an AFM cantilever to map the locations of ligands bound on solid surfaces (26). The goal of our project is to enable this “recognition mapping” method to be used in the study of the surfaces of living cells. Moving recognition microscopy onto living cell surfaces poses some particular challenges related to the fact that there is a mechanical coupling between the measuring system and the object to be observed. Difficulties arise because of the softness of the cell components, the size of the cell, and the need to work in an aqueous environment. Another view might attribute the difficulties to the size of cells. The net result of each of these challenges is that a rather large deformation of the cell may be necessary to measure its mechanical properties, even at a single receptor site. An immediate technical challenge then is to separate the interesting local characteristics of the receptor site from the gross deformation of the cell as a whole, requiring at the very least an understanding of how cellular anatomy translates into mechanical response. Here we approach this challenge by developing a computational model of the cell and design a set of experiments to parameterize the model. The framework for the
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC. This paper was submitted directly (Track II) to the PNAS office. Abbreviations: AFM, atomic force microscopy; CSG, coverslip glass; TSB, Tris-saline buffer; F(d), force-distance. †To whom reprint requests should be addressed at: Materials Research Institute; L-418, Lawrence Livermore National Laboratory, 7000 East Avenue, Livermore, CA 94550. E-mail: [email protected].
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
COMBINING CONSTITUTIVE MATERIALS MODELING WITH ATOMIC FORCE MICROSCOPY TO UNDERSTAND THE MECHANICAL PROPERTIES OF LIVING CELLS
6494
model is borrowed from solid mechanics, which has been developed to describe a broad range of materials, including rubbers and other soft materials. The modeling techniques we describe here are not based on a Hertzian analysis of the deformation, as has been common practice in AFM literature (27). It has been recognized that the material constants extracted through Hertzian analysis are, in fact, not characteristic constants of the system but rather depend on the way the measurement is made. The problem is that the techniques of AFM and nanoindentation have been developed for relatively hard materials such as metals, semiconductors, and ceramics. These systems typically offer a flat surface for analysis, and they allow only a small indentation before they yield plastically or fracture, which is the kind of system described well by the theory of elastic indentation developed by Hertz (28). The Hertz theory makes a few basic assumptions: • • • •
The material under study comprises a large system with a flat surface (a half-space). The material returns to its original shape when the load is released (elasticity). The material is linear: doubling the stress doubles the strain (Hooke's Law). There is no preferred direction or point in the bulk material (isotropy and homogeneity).
In addition, Hertz's original theory did not allow for adhesion of the indenter to the material, but the theory has been extended by Johnson, Kendall, and Roberts to account for the possibility that the surfaces would stick together (29). It is clear that the assumptions of the Hertzian theory are not applicable to living cells. The deformation of a cell during a typical experiment is a considerable fraction of the size of a cell, so the half-space assumption is incorrect (finite size is important); the deformation may depend on the rate at which the force is applied (viscoelasticity may be important); large deformations lead to a nonlinear response (hyperelasticity); and the structure within the cell can lead to some regions being harder than others (inhomogeneity and perhaps anisotropy). Here we focus on the problems of finite domain and nonlinear elasticity. We have selected the bovine sperm cell for these studies. This cell was selected for several features, including the discreteness of the cell and the reproducible well-defined shape. We are primarily interested in the anterior of the cell, and the tail's effect may be safely neglected. The interior of the cell is composed primarily of chromatin (protein-coated DNA) and water and, for the purposes of this discussion, it is modeled as a homogeneous incompressible medium. The chromatin is understood to play an important role in determining the cell shape, but its elastic properties are not expected to influence small-to-moderate deformations of the dorsal region of the cell. Experiment. Previously frozen bovine sperm cells were plated onto 0.170-mm-thick coverslip glass (CSG) that was pretreated with a coating of 1% poly-L-lysine. The CSG was then transferred to the bottom of a Petri dish containing Tris-saline buffer (TSB) (150 mM NaCl/10 mM Tris, pH 7.2), and freshly suspended cells were added to the perimeter of the Petri dish such that the cells were not introduced directly above the CSG. The cells were incubated in the Petri dish for approximately 15 min or until adequate adsorption of living cells to the CSG had occurred (as monitored with a light microscope). The CSG was then carefully removed from the Petri dish such that a dome of TSB fluid was retained over the adsorbed cells and the backside of the CSG wicked completely dry with a tissue. The CSG was then mounted to the AFM stub by using a small piece of double-sided sticky tape and loaded into the AFM. The model DNP AFM probe from Digital Instruments (Santa Barbara, CA) was chosen for these measurements. It had been selected because the large radius of curvature of its probe tip and the ultra-low spring constant of its cantilever allowed us to interrogate the living cells to suit our objectives without damage to the cells. Before measurements, the size and shape of AFM probe were characterized by using a titanium reference sample from Digital Instruments/Veeco and the TGG01 silicon grating from MikroMasch (Tallinn, Estonia). This probe's cantilever is made of silicon nitride, is triangular, 200 microns in length, and has a spring constant of 0.06 N/m. Cantilever sensitivity (i.e., cantilever deflection signal vs. voltage applied to move the z-piezo) was first determined by using an extremely hard reference sample made of sapphire. The probe was then used to make force measurements (under TSB) in three predetermined subregions of the bovine sperm cell. The force curves were taken by using a total z-scan size of 600–800 nm with a penetration depth into the cell of about 350 nm. The z-scan rate used was approximately 10 Hz. After the force curve analysis, the probe was again characterized by using the titanium reference.
Fig. 1. AFM measurement of topography of the bovine sperm cell.
Fig. 1 shows the topography of a bovine sperm cell analyzed by contact mode AFM under TSB. Several regions of the cell are distinguishable, including the acrosome, midpiece, postacrosomal segments, and flagellum. The three segments are distinguished by the amplitude of the local height variations, with the acrosomal region having variations on the order of 100 nm, the midpiece exhibiting 5-nm variations, and the postacrosomal region 15-nm variations in local height. In addition, a fairly clearly defined 30-nm depression running across the short axis of the cell body identifies the boundary between the midpiece and postacrosomal segments. These amplitude variations are consistent with the numbers of membrane layers present in each region. The total cell thickness at each of the three regions as measured by AFM under TSB is as follows: acrosomal region, 797 nm; midpiece, 689 nm; postacrosomal region, 610 nm. The 100-nm height variations seen in the acrosomal region are because of the presence of the inner and outer acrosomal membranes that encapsulate the anterior end of the cell. The 5-nm height variations measured for the midpiece are because of a flat belt-like structure corresponding to the equatorial segment. The postacrosomal region's 15-nm height variations are because of a delicate highly porous layer corresponding to the perinuclear material. The pores are approximately 80 nm in diameter. Fig. 2 shows force-distance [F(d)] curves for each of the three regions: acrosomal, midpiece, and postacrosomal segments, which are evident in the topographic image. The F(d) curves
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
COMBINING CONSTITUTIVE MATERIALS MODELING WITH ATOMIC FORCE MICROSCOPY TO UNDERSTAND THE MECHANICAL PROPERTIES OF LIVING CELLS
6495
show three regimes: (i) essentially flat on initial approach toward the cell surface (i.e., from ≈625 to ≈260 nm), (ii) a shallow nearly linear slope for over 100 nm once the tip is engaged with the cell, and (iii) the slope increases nonlinearly for the final approach of the tip as the cell is further compressed. These three behaviors correspond, respectively, to (i) the tip moving through the fluid with nominal resistance, (ii) a stiffness of about 0.03 N/m, and then (iii) a continually increasing stiffness. Hysteresis is exhibited; however, because the approach and retraction curves show changes at the same d values, we conclude that the hysteresis is probably associated with irreversible displacements of the media between membrane layers.
Fig. 2. Force vs. distance curves for the acrosome.
Modeling. The goal of the modeling described here is to obtain a mathematical relation between the deformation of the cell and the applied forces. This relation will predict the deformation under a given force or alternatively allow the determination of the applied forces once the deformation (shape) is known. The modeling has two stages: first we model the mechanical properties of a membrane, and then we model the AFM experiment. The membrane is modeled as a nonlinear elastic medium, whereas the AFM experiment is considered to be a point-load problem (a force applied at just one point) on a sector of a sphere. We assume the response observed in our AFM experiments on living cells is elastic, which means that the cell will return to its original shape once the applied force is removed. There is some evidence of a viscoelastic response under certain conditions that will not be addressed here (27). Because of the small thickness of the membrane with respect to the size of the cell, a direct two-dimensional continuum model is used to model the combination of membrane and associated cytoskeleton. The third dimension, the thickness of the membrane, is essentially atomistic in nature and is regarded as negligible from the point of view of continuum deformations. Each phospholipid bilayer is ≈5 nm thick, and the composite membrane is roughly 30 nm thick, varying somewhat from site to site on the cell, which is very small in comparison with the cell length of 10,000 nm. Similarly, the tip of the AFM cantilever used in our experiments is approximately 50 nm in radius, which is comparable to the thickness of the membrane and negligible compared with the cell dimensions, justifying the use of a point-load treatment in our model. We return to this point below, showing that the model is internally consistent, because the estimated curvature of the membrane is larger than the radius of the tip, so the point load is a reasonable first approximation. In this model, we neglect the multilayer nature of the membrane. Modeling the Membrane. The characteristic quantity of our model of the nonlinear elastic fluid membrane is the strain energy/unit mass, w, of the membrane, i.e., the energy required to deform the membrane. Once the strain energy is determined, the stress function can be easily obtained with the methods of elasticity by taking the derivatives of the strain energy with respect to strain and curvature. When the membrane is treated as a two-dimensional (closed) nonlinearly elastic fluid continuum, w must be a scalar function of two quantities, J and H, which characterize the local stretching and bending of the membrane, respectively (30). More precisely, J characterizes the local change in area, and H is the mean curvature. There have been widespread attempts to use classical Kirchhoff linear plate bending theory for solids to model biomembranes and surfactant systems, (see refs. 30–33 and refs. therein). The invariant H used there was the invariant of a tensor characterizing the change in curvature, say k, which is not appropriate for describing fluidity, for reasons discussed in ref. 30. Nevertheless, when the plate model is used to describe small deformations, k approximates well the curvature tensor used to describe fluids (whose invariant is H) and therefore to this order of approximation, the two theories would yield the same result for small deformations. But the deformation of soft tissue, especially the deformation of a cell membrane in an AFM experiment, would certainly yield very large deformations, for which a form of w appropriate for fluids (that depends on J and H) is used. The dependence of w on the two quantities mentioned above
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
COMBINING CONSTITUTIVE MATERIALS MODELING WITH ATOMIC FORCE MICROSCOPY TO UNDERSTAND THE MECHANICAL PROPERTIES OF LIVING CELLS
6496
is considered in general, in an additive way, through two coefficients T and A, respectively, [1]
where T is the stretching modulus and A is the bending modulus. The invariant H appears squared in the expression for w to ensure the physical requirement that the membrane's local response is the same for upward as well as downward bending. The bending moduli are materials constants specific to this cell for the membrane. A more detailed discussion of this energy function will be given elsewhere (E.B., R.E.R., M.M., J.B., and R.B., unpublished work; ref. 34). Here we just mention that A can be related to the moments necessary to deform a plane membrane into a cylindrical surface. Modeling the AFM Experiment. To model the AFM experiment, initially the cell is considered to be a sector of a sphere that is subjected to a point force F, at the pole. The actual shape that the cell takes on under an applied force is given as a solution to a set of partial differential equations (the Euler-Lagrange equations) that are derived on the basis of energy minimization considerations. The axisymmetry in the problem reduces these partial differential equations to ordinary differential equations with eight associated boundary conditions, which can be solved numerically. Considering half the cell as a sector of a sphere is particularly appropriate to the bovine sperm cell, which has a definite shape. The media outside of the membrane (the liquid droplet surrounding it and the cytoplasm) are assumed to be incompressible, acting on the membrane through a net pressure. This assumption seems to be reasonable because more than 50% of the volume inside the cell is water. By enforcing this constraint, the net pressure through the membrane appears in the equations as a Lagrange multiplier to be calculated after the equations are solved, from the condition of preserved volume. Therefore, it is not necessary to directly measure the pressure. The input parameter to this model is the applied force, and what is sought is the deformation (the profile of the deformed cell), that is, the coordinates of each point of the membrane as functions of the arc length along the meridian of the sphere. The dependence of w on J can be avoided if we use the assumption that the local area of the membrane is preserved. This constraint yields one relation between the variables in question. In that case, T can be calculated after the fact, as another Lagrange multiplier. Alternatively, we can leave it in the strain energy and actually test the assumption that the area is preserved (see ref. 35). Therefore, change in the strain energy, ∆w, is due only to change in H2. For a given value of the parameter A, the shape of the cell can be computed by minimizing the strain energy. The optimal value of A gives the closest agreement between the computed and measured shape of the cell. What is required from experiment is information about the deformed surface of the cell. The parameterized strain energy is then used to compute the mechanical properties of the membrane. Further details of this model will be presented elsewhere (E.B., R.E.R., M.M., J.B. and R.B., unpublished work; ref. 34). Although we currently have no direct measurement of the deformed cell shape, we can estimate the curvature within the context of our model assuming the membrane behaves like a liquid crystal (3). Within this assumption, A≈10−12 erg (1 erg= 0.1 µJ). From our preliminary force-displacement preliminary plot in Fig. 2, we see that the maximum force F=6.6×10−9 N corresponds to the displacement d=60×10−9 m. The total work done by external forces is (0.5)Fd≈4×10−16 Nm, which is balanced by the work stored in the membrane, i.e., [2]
where da is the area element. The integral can be estimated by using an average value of ∆H2, say ∆h2, where h2=∫ total surface H2 da. With is of the order of 10−3, demonstrating these numerical values, and using a membrane thickness, t=10−8 m, the quantity that the length scales in the problem are well within the range of applicability of our two-dimensional continuum model. Further, these estimates give an average radius of curvature of 10−6 m for the deformed membrane. The quantity h2 is approximately equal to 4/R2, where R is the average radius of curvature. The radius of curvature at the pole, where the force is applied, gives most of the contribution to this average value. Using Eq. 2 and the experimental force-displacement curve, we estimate the maximum radius of curvature of the membrane to be 10−6 m. This value is two orders of magnitude greater than the radius of the tip (10−8 m) and the thickness of the membrane (10−8 m), justifying the approximation of modeling the AFM-cell interaction as a point load. DISCUSSION We have selected a particular strategy for moving recognition microscopy onto a living cell surface. To separate the local characteristics of the receptor sites, which we plan to study with recognition microscopy, from the gross deformation of the cell as a whole, we have developed a computational model to help us understand how cellular anatomy translates into mechanical response. This approach is because mechanical response associated with the softness of the “materials” comprising the cell will be convoluted with the mechanical response associated with the recognition events that we want to study. In the present work, we have measured both a highly detailed topology of the bovine sperm cell and force vs. distance curves. In combination with the computational model, only the net force was used here. By using this net force with a published value of the parameter A, the model was used to derive a prediction for the deformation of the membrane. We are presently developing an experimental method that will allow the direct measurement of the cell deformation under a point load. With a quantitative knowledge of the point load force and the cell deformation profile the strain energy/unit mass, w, of the membrane can be parameterized. With w parameterized, we then know the mechanical response of the cell membrane under any combination of forces. The combination of model plus experiment envisioned here might be refined in a number of ways to accommodate observed phenomena that have been neglected in this first phase. For example, it is known that there is a force generated in response to the dragging of entities through, or across, the cell membrane. To accommodate this in the theory, one may add terms that depend on the time derivative of the surface metric to the membrane part of the stress and on the derivative of the curvature to the bending part. The coefficients associated with these additional terms are the viscosity associated with strain rate and with flexure rate, respectively. The resulting constitutive equations may then be used in the existing equations of motion for the surface to predict the coupling between deformation and deformation rate generated by nonequilibrium processes associated with cell response. This evolution of the theoretical model can be tested then with suitable experiments that will emphasize its new features. We thank Stuart Lindsay, David Steigman, Michael Ortiz, and Carl Melius for useful discussions. This work was performed under the auspices of the U.S. Department of Energy by the University of California, Lawrence Livermore National Laboratory (LLNL), under Contract No. W-7405-Eng-48. We gratefully acknowledge funding through the LLNL Laboratory Directed Research and Development Grant 01ERI-001. E.B. thanks the Department of Applied Science, University of California, Davis, and the LLNL Materials Research Institute for their hospitality during her visit.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
COMBINING CONSTITUTIVE MATERIALS MODELING WITH ATOMIC FORCE MICROSCOPY TO UNDERSTAND THE MECHANICAL PROPERTIES OF LIVING CELLS
6497
1. Lekka, M., Lekki, J., Marszalek, M., Golonka, P., Stachura, Z., Cleff, B. & Hrynkiewicz, A.Z. (1999) Appl. Surf. Sci. 141, 345–349. 2. Primakoff, P. & Myles, D.G. (1983) Dev. Biol. 98, 417–428. 3. Friend, D.S. (1989) Ann. NY Acad. Sci. 567, 208–221. 4. Edidin, M. (1997) Curr. Opin. Struct. Biol. 7, 528–532. 5. Retvield, A. & Simons, K. (1998) Biochim. Biophys. Acta 1376, 467–479. 6. Gumbiner, B. & Louvard, D. (1985) Trends Biochem. Sci. 10, 435–438. 7. Hadjiconstantinou, N.G. & Patera, A.T. (1997) Int. J. Modern Phys. 8, 967–976. 8. Edidin, M. (1993) J. Cell Sci. Suppl. 7, 165–169. 9. Zhang, F., Lee, G.M. & Jacobsen, K. (1993) BioEssays 15, 579–588. 10. Allen, M.J., Lee, C., Pogany, G.C., Balooch, M., Siekhaus, W.J. & Balhorn, R. (1993) Chromosoma 102, 623–630. 11. Allen, M.J., Bradbury, E.M. & Balhorn, R. (1995) J. Struct. Biol. 114, 197–208. 12. Moy, V.T., Florin, E.L. & Gaub, H.E. (1994) Colloids Surf. 93, 343–348. 13. Lee, G.U., Kidwell, D.A. & Colton, R.J. (1994) Langmuir 10, 354–357. 14. Florin, E.-L., Moy, V.T. & Gaub, H.E. (1994) Science 264, 415–417. 15. Boland, T. & Ratner, B.D. (1995) Proc. Natl. Acad. Sci. USA 92, 5297–5301. 16. Hinterdorfer, P., Baumgartner, W., Gruber, H.J., Schlcher, K. & Schindler, H. (1996) Proc. Natl. Acad. Sci. USA 93, 3477–3481. 17. Dammer, U., Hegner, M., Anselmetti, D., Wagner, P., Dreier, M., Huber, W. & Guntherodt, H.-J. (1996) Biophys. J. 70, 2437–2441. 18. Allen, M.J. (1997) IEEE Eng. Med. Biol. March/April, 34–41. 19. Zhang, P.-C., Bai, C., Ho, P.K.H., Dai, Y. & Wu, Y.-S. (1997) IEEE Eng. Med. Biol. March/April, 42–46. 20. Ros, R., Schwesinger, F., Anselmetti, D., Kubon, M., Schafer, R., Pluckthun, A. & Tiefenauer, L. (1998) Proc. Natl. Acad. Sci. USA 95, 7402–7405. 21. Wong, S.S., Joselevich, E., Woolley, A.T., Cheung, C.L. & Lieber, C.M., (1998) Nature (London) 391, 52–55. 22. Willemsen, O.H., Snel, M.M.E., van der Werf, K.O., de Grooth, B.G., Greve, J., Hinterdorfer, P., Gruber, H.J., Schindler, H., van Kooyk, Y. & Figdor, C.G. (1998) Biophys. J. 75, 2220–2228. 23. Fritz, J., Katopodis, A.G., Kolbinger, F. & Anselmetti, D. (1998) Proc. Natl. Acad. Sci. USA 95, 12283–12228. 24. Frisbie, D., Rozsnyai, L.F., Noy, A., Wrighton, M.S. & Lieber, C.M. (1994) Science 265, 2071–2074. 25. Moy, V.T., Jiao, J., Hillmann, T., Lehmann, H. & Sano, T. (1999) Biophys. J. 76, 1632–1638. 26. Raab, A., Han, W., Badt, D., Smith-Gill, S.J., Lindsay, S.M., Schindler, H. & Hinterdorfer, P. (1999) Nat. Biotechnol. 17, 902–905. 27. A-Hassan, E., Heintz, W.F., Antonik, M.D., D'Costa, N.P., Nageswaran, S., Schoenenberger, C.-A. & Hoh, J.H. (1998) Biophys. J. 74, 1564–1578. 28. Hertz, H. (1882) J. Reine Angew. Math. 92, 156–171. 29. Johnson, K.L., Kendall, K. & Roberts, A.D. (1971) Proc. R. Soc. London Ser. A 324, 301–321. 30. Steigmann, D.J. (1999) Arch. Rational Mech. Anal. 150, 127–152. 31. Helfrich, W. (1973) Naturforschung 28c, 693–703. 32. Fung, Y.C. (1966) Proc. Fed. Am. Soc. Exp. Biol. 25, 1761–1772. 33. Kleman, M. (1976) Proc. R. Soc. London A 347, 387–404. 34. Rudd, R.E., McElfresh, M., Balsu, E., Balhorn, R., Allen, M.J. & Belak, J. in Proc. Int. Conf. on Computational Nanoscience ICCN '02, San Juan, PR, April 21–25, 2002, in press. 35. Evans, E.A. & Skalak, R. (1980) Mech. Thermodyn. Biomembr. (CRC, Boca Raton, FL).
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGNING SUPRAMOLECULAR PORPHYRIN ARRAYS THAT SELF-ORGANIZE INTO NANOSCALE OPTICAL AND MAGNETIC MATERIALS
6498
Colloquium Designing supramolecular porphyrin arrays that self-organize into nanoscale optical and magnetic materials Charles Michael Drain*†, James D.Batteas†‡, George W.Flynn†§, Tatjana Milic*, Ning Chi‡, Dalia G.Yablon§, and Heather Sommers‡ *Department of Chemistry and Biochemistry, Hunter College and Graduate Center of City University of New York, 695 Park Avenue, New York, NY 10021; ‡Department of Chemistry, College of Staten Island and Graduate Center of City University of New York, 2800 Victory Boulevard, Staten Island, NY 10314; and §Department of Chemistry, Columbia University, 3000 Broadway, MC 3167, New York, NY 10027 Contributed by George W.Flynn, October 2, 2001 Tessellation of nine free-base porphyrins into a 3×3 array is accomplished by the self-assembly of 21 molecular entities of four different kinds, one central, four corner, and four side porphyrins with 12 trans Pd(II) complexes, by specifically designed and targeted intermolecular interactions. Strikingly, the self-assembly of 30 components into a metalloporphyrin nonamer results from the addition of nine equivalents of a first-row transition metal to the above milieu. In this case each porphyrin in the nonameric array coordinates the same metal such as Mn(II), Ni(II), Co(II), or Zn(II). This feat is accomplished by taking advantage of the highly selective porphyrin complexation kinetics and thermodynamics for different metals. In a second, hierarchical self-assembly process, nonspecific intermolecular interactions can be exploited to form nanoscaled three-dimensional aggregates of the supramolecular porphyrin arrays. In solution, the size of the nanoscaled aggregate can be directed by fine-tuning the properties of the component macrocycles, by choice of metalloporphyrin, and the kinetics of the secondary self-assembly process. As precursors to device formation, nanoscale structures of the porphyrin arrays and aggregates of controlled size may be deposited on surfaces. Atomic force microscopy and scanning tunneling microscopy of these materials show that the choice of surface (gold, mica, glass, etc.) may be used to modulate the aggregate size and thus its photophysical properties. Once on the surface the materials are extremely robust. With the increasing demand for the ability to sculpt matter into precise functioning devices of nanoscale dimensions, the molecular level design of functional materials is an overarching theme in much of the synthetic materials literature (1–7). Inspired by biological systems, the introduction of specific interactions is a route toward using the facile and energetically favorable production capabilities to self-assemble materials (8). Exploitation of nonspecific intermolecular interactions has resulted also in the formation of molecular electronic devices (9–11). We have used self-assembly to form a square planar array of nine porphyrins mediated by coordination of exocyclic pyridyl groups on three different porphyrins to 12 trans-palladium dichlorides (12, 13). In addition to modulating the size and distribution on surfaces, metalation of the porphyrin macrocycle enables one to design nanoscale systems with a host of photonic, magnetic, redox catalytic, and sensor capabilities. These functions have been well studied on metalloporphyrin monomers. Substitution of the peripheral R groups with long-chain hydrocarbons enables the design of nanoscale aggregates that, using nonspecific interactions, organize into two-dimensional arrays (refs. 14–20 and references therein). Here we present an overview of the design capabilities for materials and devices by using porphyrin supramolecular arrays. FORMATION OF THE FREE-BASE NONAMER SUPRAMOLECULAR ARRAY Earlier (12, 13), we reported the synthesis and characterization of a porphyrin nonamer that is self-assembled by the coordination of exocyclic pyridyl groups on nine porphyrins to 12 PdCl2 (Fig. 1). The predefined geometry of either the metallo or free-base porphyrins as well as the coordination geometry of the metal ion linker, all in the correct stoichiometry, dictates the final structure of the self-assembled arrays. In the case of the square planar nonamer, the 180° (trans) coordination geometry of the 12 PdCl2 species combines with four “L-shaped” 5,10-bis (4-pyridyl)-15,20-bis(4-alkylphenyl)porphyrins that serve as the corners, four “T-shaped” 5,10,15-Tris(4-pyridyl)-20-(4-alkylphenyl) porphyrins that constitute the sides, and one “X-shaped” tetrakis(4-pyridyl)porphyrin that resides in the center. Alternatively, the combination of two X-shaped and two L-shaped porphyrins with the 90° (cis) coordination geometry of six PtCl2 linkers results in linear tapes substantially weighted toward the tetramer (12, 13). All these self-assembled arrays were well characterized in solution. This earlier report also contained a preliminary atomic force microscopy (AFM) study that revealed nanoscaled aggregates of the nonameric arrays may be deposited on glass surfaces. As defined elsewhere (1), specific intermolecular forces are directional and have predefined donor-acceptor interactions such as hydrogen bonds and metal ion coordination, whereas nonspecific intermolecular forces usually have less directionality and no specifically designed donor-acceptor interactions such as electrostatic interactions. Herein we use the former to design and self-assemble discrete supramolecular entities and the latter to self-organize the supramolecules into hierarchical structures. DIFFERENTIAL METALATION When nine equivalents of a divalent transition metal acetate [Mn(II), Ni(II), Co(II), or Zn(II)] dissolved in methanol is added simultaneously with 12 equivalents of PdCl2Bn2 in toluene to a <10 µM mixture of porphyrins in toluene at 40–50°C, the self-assembly of the 30-component metalloporphyrin nonamer is
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC. Abbreviation: AFM, atomic force microscopy. †To whom reprint requests may be addressed. E-mail: [email protected], [email protected], or [email protected].
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGNING SUPRAMOLECULAR PORPHYRIN ARRAYS THAT SELF-ORGANIZE INTO NANOSCALE OPTICAL AND MAGNETIC MATERIALS
6499
accomplished in ≈90% yield in 1–3 h, depending on the metal ion. This in situ is possible for several reasons. First it is well known that the kinetics and thermodynamics of porphyrin metalation by the first-row transition metals is substantially different than that of Pd(II) (21). The size of the metal ion, the lability of the counter ions, and the energetics of the metal-porphyrin interaction all affect the yield and rate of the reaction. Thus for most porphyrins the first-row transition metals can be inserted into the core in a few hours or less at temperatures <80°C by using an excess of the acetate salt dissolved in methanol, whereas Pd(II) insertion takes 8–12 h at >120°C. Yet, the efficiency observed for the self-assembly of the metalloporphyrin nonamer also suggests that there are additional considerations, because a stoichiometric quantity of the metal to be inserted is used. As we and others have observed, the insertion of Zn(II) into meso pyridyl porphyrins increases the basicity of the pyridyl nitrogen (12, 13, 22). This increased bacisity is indicated by a concomitant increase in the coordination bond energy between the pyridine moiety and both Pt(II) and Pd(II). The stability of dimers formed from free-base monopyridylporphyrins and either PtCl2 or PdCl2 in toluene is less than the same dimers composed of the zinc-metalated porphyrins by about a factor of two. Inversely, it is reasonable to expect that the exocyclic coordination of Pd(II) or Pt(II) by the pyridyl moieties may increase the binding constant and/or lower the barrier to Zn(II) insertion into the macrocycle via similar electronic effects. Therefore, there is a synergy, or cooperativity, between the metalation of the porphyrins and the formation of the nonameric array (23). Under these conditions, UV-visible spectra indicate that the nonamer is formed before complete metalation of the free-base porphyrins. This in situ result is supported by the observation of the same kinetics for porphyrin metalation of the preassembled nonamer under the same conditions. The electronic perturbation of the macrocycle by exocyclic ligand coordination to these square planar metals is demonstrated also by 2–5 nm red shifts observed in the visible spectra (12, 13, 22).
Fig. 1. The self-assembled 30-component metalloporphyrin nonamers (Left) subsequently self-organize into columnar stacks (Right, not to scale). For example, when r=t-butyl and the cobalt porphyrin is in the nonamer, the average height of the stacks on glass is nearly twice that of the stacks formed from the free-base nonamer
SECONDARY SELF-ORGANIZING PROCESSES The initial self-assembly process, the formation of the nonamer, is directed by specifically designed intermolecular forces, stoichiometry, and an understanding of the complexities of the thermodynamics of the self-assembly process. The secondary, hierarchical self-organization of the porphyrin nonamers into columnar aggregates of low polydispersity arises from nonspe
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGNING SUPRAMOLECULAR PORPHYRIN ARRAYS THAT SELF-ORGANIZE INTO NANOSCALE OPTICAL AND MAGNETIC MATERIALS
6500
cific interactions inherent in the supermolecule. The formation of the nanoscaled aggregates of porphyrin nonamers, as in the case of the formation of many nanoscaled colloidal particles, is driven partially by the minimization of surface area and surface energetics. The limited particle size is expected and is manifested in the fact that the aggregates have roughly the same dimensions in all three axes. Thus, the ≈6×6nm square nonamer forms ≈4–6-nm tall columnar stacks (12, 13). Small stacking differences are observed by altering the various pendant R groups, -H, -CH3, and -t-butyl, which can be attributed to packing and somewhat to electronic effects on the π system. The energetics of self-organization of the columnar stacks of porphyrin nonamers arise from the complex interplay between a variety of intermolecular forces. The π stacking of porphyrins to form face-to-face (H) or edge-to-edge (J) aggregates is a well known and understood phenomenon (12, 13). Electronic spectra indicate that both types of arrangements are present in the columnar aggregates (ref. 20 and references therein). Under the conditions described herein, however, neither solutions of individual porphyrins or a mixture of them nor the palladium complex results in any discernable aggregation as shown by absorption and emission spectra, dynamic light scattering, or AFM. Estimates of πstacking interactions are between 3 and 5 kcal·mol−1 per porphyrin face for mesotetraphenylporphyrin (ref. 20 and references therein). Because the aryl groups are not coplanar with the porphyrin, these substituents also sterically prevent exact alignment of the two macrocycles. A second internonamer interaction comes from the PdCl2 groups. The electrostatic and steric interactions arising from the Pd2+ and the Cl− ions would tend to place the Cl− from one nonamer over the vacant axial positions of the Pd2+ ions of the adjacent nonamer, over the pyrrole NH, or in the case of the metalloporphyrins over the Zn(II). The nature of the R groups, and their position on the aryl ring, also influences the kinetics and size of the secondary assembly; small substituents on the para-position favor π stacking, whereas those on ortho and meta positions would prevent significant π stacking. In general, the kinetics are faster, and the resultant hierarchical assemblies are larger as one goes from r=t-butyl to methyl to H. After reaching equilibrium the respective average columnar heights by dynamic light scattering are 6, 7, and 10 nm, and by AFM on glass surfaces 5, 7, and 8 nm. Although the exact geometry of the nanoscaled aggregates is still under investigation, docking experiments suggest that one planar array may stack on top of another in one of either two ways: (i) one is rotated 30–60° relative to the next, or (ii) one is off set diagonally by a little more than half the diameter of a single porphyrin (≈0.6 nm). FORMATION OF SURFACE-BOUND STRUCTURES AS DEVICE PRECURSORS To convert these materials into useful nanoscale devices, their organization and stability after surface deposition must be evaluated. A myriad of devices can be envisioned by using these materials as a platform for their formation. These devices include (i) chemical sensor arrays using changes in the optoelectronic properties of the materials for signaling, (ii) complex three-dimensional storage devices based on the ability of Co(II)-porphyrin nonamers to form organized nanoscale stacks of magnetic materials with varying magnetic properties depending on the number of nonamers in a stack, and (iii) photo-gated magnetic materials wherein the magnetism is gated on or off depending the metalloporphyrin used. As a precursor to the formation of such devices, we have investigated the adsorption of both free-base and metalloporphyrin nonamers on a variety of surfaces including glass, mica, graphite, and Au(111) thin films. The adsorption properties of the nonamers have been characterized by using a combination of AFM and scanning tunneling microscopy. The free-base materials have shown remarkable stability on glass surfaces and are found to remain intact for more than 1 year under ambient conditions from deposition with no apparent change in the photonic properties by UV-visible or fluorescence spectra.
Fig. 2. The size of the columnar stack of free-base porphyrin nonamers (r= t-butyl) is determined also by the nature of the surface. On glass (a) aggregates of ≈5–27-nonamer-high stacks are predominantly found, on mica (b) the aggregate density increases as the size decreases (≈5–8-nonamer-high stacks), whereas on Au(111) (c) single nonameric arrays are found typically on the terraces with stacks containing 2–3 arrays at step defects.
Earlier we reported that the free-base nonamer forms aggregates of ≈5–6 nm3 in size in solution, observed by dynamic light scattering, that remain intact when deposited on the glass surface (12, 13). The aggregate structure on surfaces has been observed by AFM where (caused by the 10–50-nm radius of curvature of the AFM tip), only the aggregate height may be reported. Nonamer aggregates on glass (Fig. 2a) are deposited fairly uniformly in density over the surface with heights of 2–6 nm (Fig. 3a), which corresponds to ≈4–12 stacked nonamer arrays. When the free-base nonamer aggregate is deposited on graphite, the stacks are found to cluster together and are generally shorter (data not shown). On surfaces such as mica (Fig. 3b) or the electron-rich gold surface (Fig. 3c), the nonamer aggregates break apart and form smaller structures of ≈3 nm in height on mica, down to single nonameric arrays (Fig. 3c) when the aggregates are deposited on gold. Scanning tunneling microscopy of the nonamer on gold (Fig. 4) shows that the array is ≈5× 6×0.4 nm in size. Interestingly, single nonameric arrays appear almost exclusively on Au(111) terraces, whereas 2–3-nonamer
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGNING SUPRAMOLECULAR PORPHYRIN ARRAYS THAT SELF-ORGANIZE INTO NANOSCALE OPTICAL AND MAGNETIC MATERIALS
6501
high stacks are found to bind to defects such as step edges on the gold surface. The appearance of stacks at defect sites suggests that the aggregate height, thus its optoelectronic properties, may be tuned at the level of single arrays by controlling the strength of the surface dipole of the metal arising from the Smoluchowski effect (24–26). Thus one can envision that the organization of these materials on metal nanoparticles of varying dimensions will exhibit tuned photonic properties because of the number of supermolecules stacked together is controlled by the varying surface dipoles.
Fig. 3. Histograms of the stacking heights taken from topographic AFM measurements for the free-base nonamer (r=t-butyl) on glass (a), mica (b), and gold (c). In an experiment that used 20% less Co(II) than needed to fully metallate the porphyrin nonamer (d), one observes two populations of columnar stacks on glass surfaces as expected. One height is nearly equivalent to the free base, and one is about twice this height, which suggests that the Co(II) nonamers aggregate into larger columnar stacks because of an additional magnetic or electrostatic attraction between the layers.
Fig. 4. Scanning tunneling microscope constant current (I=250 pA, V= 1.2V) image (29×29 nm) of a single nonamer on the Au(111) surface. The line trace (Inset) shows that the nonamer is ≈5–6 nm wide and ≈0.4 nm high.
With metalation, the Zn(II) and Co(II) porphyrin nonamers deposited on glass exhibit divergent properties. In general the Zn(II) nonamers show smaller stack heights on glass than the free base. Preliminary results on Co(II) nonamers, however, show a height distribution nearly double the stack height of the free base (≈9±1 nm), suggesting that the Co(II) porphyrin nonamers have an increased proclivity for stacking. Perhaps the formation of the larger aggregates is caused by the addition of a magnetic attraction between the nonamer layers in the aggregates (Fig. 3d). Preliminary magnetic force microscopy measurements employing cobalt-coated AFM tips suggest that the cobalt nonamers have net magnetic dipoles oriented perpendicular to the surface with the dipole facing out from the surface. Here again, it is expected that the tunability of the stack height dictated by the surface chemistry and electronic properties will afford the controlled formation of nanoscale aggregates with designed magnetic properties. CONCLUSIONS The height of the columnar aggregates is determined by a balance between the porphyrin π-stacking forces, the electrostatics of the PdCl2 units, and the steric interactions between the meso substituents. These aggregates are remarkably robust and may be deposited on a variety of surfaces, which also affords a means to modulate the particle size, where they have been characterized by scanning probe microscopy. The metalation of the porphyrins in the nonamer can be accomplished either in the self-assembly process or in situ after the formation of the array.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGNING SUPRAMOLECULAR PORPHYRIN ARRAYS THAT SELF-ORGANIZE INTO NANOSCALE OPTICAL AND MAGNETIC MATERIALS
6502
Because there are three different porphyrins in the nonameric array, four T, four L, and one X, each type of macrocycle can contain a different metal ion. The advantage of metalating the porphyrins before self-assembly is that up to three different metals can be used and they will reside in predetermined sites in the supermolecule. The metalation state and type of metal ion in the porphyrin are powerful means to modulate the photophysical properties and the chemical reactivity of the nonamers and the nanoscale aggregates. This work dovetails well with other recent work on nanoscaled materials based in inorganic, organic, and organometallic systems (27–33). We gratefully acknowledge support from National Science Foundation Grants CHE-9732950, DGE-9972892, CHE-0095649, and DMR-9809687 and the Professional Staff Congress-City University of New York. Hunter College chemistry department infrastructure is supported partially by National Institutes of Health RCMI program GM3037. H.S. was supported by the National Science Foundation-Research Experience for Undergraduates Program in Polymers and Biopolymers at College of Staten Island (CHE-0097446). 1. Alivisatos, A.P., Barbara, P.F., Castleman, A.W., Chang, J., Dixon, D.A., Klein, M.L., McLendon, G.L., Miller, J.S., Ratner, M.A., Rossky, P.J., Stupp, S.I. & Thompson, M.E. (1998) Adv. Mater. 10, 1297–1336. 2. Aviram, A. & Ratner, M. (1998) Ann. N.Y. Acad. Sci. 852, 1–349. 3. Lehn, J.-M. (1990) Angew. Chem. Int. Ed. Engl. 29, 1304–1319. 4. Stang, P.J. & Olenyuk, B. (1997) Acc. Chem. Res. 30, 502–518. 5. Lindsey, J.S. (1991) New J. Chem. 15, 153–180. 6. Fujita, M., ed. (2000) Structure and Bonding (Springer, New York), Vol. 96. 7. Reed, M.A. (2001) MRS Bull. 26, 113–120. 8. Ball, P. (2001) Nature (London) 409, 413–416. 9. Drain, C.M. & Mauzerall, D. (1992) Biophys. J. 63, 1556–1563. 10. Drain, C.M. & Mauzerall, D. (1990) Bioelectrochem. Bioenerg. 24, 263–266. 11. Drain, C.M., Christensen, B. & Mauzerall, D. (1989) Proc. Natl. Acad. Sci. USA 86, 6959–6962. 12. Drain, C.M., Nifiatis, F., Vasenko, A. & Batteas, J.D. (1998) Angew. Chem. Int. Ed. Engl. 37, 2344–2347. 13. Drain, C.M., Nifiatis, F., Vasenko, A. & Batteas, J.D. (1998) Angew. Chem. 110, 2478–2481. 14. Fox, M.A. (1999) Acc. Chem. Res. 32, 201–207. 15. Sarno, D.M., Jiang, B., Grosfeld, D., Afriyie, J.O., Matienzo, L.J. & Jones, W.E., Jr. (2000) Langmuir 16, 6191–6199. 16. Qiu, X., Wang, C., Zeng, Q., Xu, B., Yin, S., Wang, H., Xu, S. & Bai, C. (2000) J. Am. Chem. Soc. 122, 5550–5556. 17. Liu, C.-Y., Pan, H.-I., Fox, M.A. & Bard, A.J. (1993) Science 261, 897–899. 18. Chambron, J.-C, Heitz, V. & Sauvage, J.-P. (2000) in The Porphyrin Handbook, Kadish, K.M., Smith, K.M. & Guilard, R., eds. (Academic, New York), Vol. 6, pp. 1–41. 19. Chou, J.-H., Kosal, M.E., Nalwa, H.S., Rakow, N.A. & Suslick, K.S. (2000) in The Porphyrin Handbook, Kadish, K.M., Smith, K.M. & Guilard, R., eds. (Academic, New York), Vol. 6, pp. 43–133. 20. Hunter, C.A. & Sanders, J.K.M. (1990) J. Am. Chem. Soc. 112, 5525–5534. 21. Buchler, J.W. (1975) in Porphyrins and Metalloporphyrins, Smith, K.M., ed. (Elsevier, New York), pp. 157–224. 22. Drain, C.M. & Lehn, J.-M. (1994) Chem. Commun. 2313–2315. 23. Sharma, C.V.K., Broker, G.A., Huddleston, J.G., Baldwin, J.W., Metzger, R.M. & Rogers, R.D. (1999) J. Am. Chem. Soc. 121, 1137–1144. 24. McCarty, G.S. & Weiss, P.S. (1999) Chem. Rev. (Washington, D.C.) 99, 1983–1990. 25. Smoluchowski, R. (1941) Phys. Rev. Lett. 60, 661–674. 26. Schenning, A.P.H.J., Benneker, F.B.G., Geurts, H.P.M., Liu, X.Y. & Nolte, R.J.M. (1996) J. Am. Chem. Soc. 118, 8549–8552. 27. Puntes, V.F., Krishnan, K.M. & Alivisatos, A.P. (2001) Science 291, 2115–2117. 28. Banfield, J.F., Welch, S.A., Zhang, H.Z., Ebert, T.T. & Penn, R.L. (2000) Science 289, 751–754. 29. Zubarev, E.R., Pralle, M.U., Sone, E.D. & Stupp, S.I. (2001) J. Am. Chem. Soc. 123, 4105–4106. 30. Diaz, J.D., Storrier, G.D., Bernhard, S., Takada, K. & Abruña, H.D. (1999) Langmuir 15, 7351–7354. 31. Jung, T.A., Schlitter, R.R. & Gimzewski, J.K. (1997) Nature (London) 386, 696–698. 32. Wurthner, F., Thalacker, C. & Sautter, A. (1999) Adv. Mater. 11, 754–758. 33. Kurth, D.G., Lehmann, P. & Schutte, M. (2000) Proc. Natl. Acad. Sci. USA 97, 5704–5707.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
NANOSCALE SURFACE CHEMISTRY
6503
Colloquium Nanoscale surface chemistry
Theodore E.Madey*†, Kalman Pelhos*, Qifei Wu‡, Robin Barnes*, Ivan Ermanoski*, Wenhua Chen*, Jacek J.Kolodziej*, and John E.Rowe§ *Department of Physics and Astronomy and Laboratory for Surface Modification, and ‡Department of Chemistry, Rutgers, The State University of New Jersey, Piscataway, NJ 08854–8019; and §United States Army Research Office, PO Box 12211, Research Triangle Park, NC 27709–2211 Edited by Robert Gomer, The University of Chicago, Chicago, IL, and approved January 9, 2002 (received for review October 10, 2001) We report evidence in several experiments for nanometer-size effects in surface chemistry. The evidence concerns bimetallic systems, monolayer films of Pt or Pd on W(111) surfaces. Pyramidal facets with {211} faces are formed on annealing on physical monolayer of Pt, Pd on a W(111) substrate, and facet sizes increase with annealing temperature. We used synchrotron radiation-based soft x-ray photoemission to show that monolayer films of Pt, Pd, on W “float” on the outer surface, whereas multilayer films form alloys on annealing. Acetylene reactions over bimetallic planar and faceted Pd/W surfaces exhibit size effects on the nanometer scale, that is, thermal desorption spectra of reactively formed benzene and ethylene (after acetylene adsorption) change systematically with facet size. In the second case, the decomposition of C2H2 over planar and faceted Ir(210) surfaces also exhibits structure sensitivity; temperature programmed desorption of H2 from C2H2 dissociation depends on the nanoscale surface structure. Finally, we have characterized interactions of Cu with the highly ordered S(4× 4)/W(111) surface. The substrate is a sulfur-induced nanoscale reconstruction of W(111) with (4×4) periodicity, having broad planar terraces (≈ ≈30 nm in width). Fractional monolayers of vapordeposited Cu grow as threedimensional clusters on the S(4×4) surface over a wide coverage range. At low Cu coverage (≤0.1 ML), Cu nanoclusters nucleate preferentially at characteristic 3-fold hollow sites; we find a clear energetic preference for one type of site over others, and evidence for self-limiting growth of nanoclusters. An important issue in surface chemistry and catalysis is how surface structures and features with nanometer dimensions affect reactivity in heterogeneous systems (1–3). The focus of our work has been on several aspects of nanoscale phenomena that influence surface chemistry, including faceting of metallic and model bimetallic catalyst surfaces, and nucleation of subnanometer metallic clusters on sulfided surfaces. We study atomically rough substrates [bcc (111) surface of W, and fcc (210) surface of Ir] that are morphologically unstable, that is, the initially planar substrate becomes covered with nanoscale facets when covered with monolayer films of gases or other metals, and heated to elevated temperature. Major objectives of this work have been (i) to determine how the surface transition from planar to faceted affects the surface reactivity of metallic and bimetallic systems, and (ii) to characterize the nucleation and growth of metals on sulfided W surfaces. The three main components of this effort are surface structure, surface chemistry, and surface electronic properties. The importance of bimetallic catalysts based on Pt-group metals has been increasing in recent decades (4, 5). These catalysts display important advantages over classical reforming catalysts, including better stability, as well as improved activity and selectivity. In particular, refractory metals (W, Mo, Re,…) in combination with Pt-group metals are active catalysts for hydrogenation and hydrogenolysis reactions (6–9). Our previous work in the area of bimetallic surfaces is summarized in two review articles (10, 11). In brief, morphologically unstable W (111) and Mo(111) coated with a single physical monolayer (ML) of certain metals or other elements (Pd, Rh, Ir, Pt, Au, O) undergo massive reconstruction from a planar morphology to a microscopically faceted surface on heating to T>700 K. Three-sided nanometer-sized pyramids form in which the facet sides are mainly film-covered {112} facets. The faceting transition in these systems is believed to be thermodynamically driven but kinetically limited: annealing is needed to achieve sufficient surface atom mobility for mass transport. The overlayer film increases the anisotropy in surface free-energy and enhances the relative stability of the faceted morphology. Striking evidence for structure sensitivity is seen in catalytic hydrogenolysis of butane over planar and faceted Pt/W. Recent synchrotron radiation studies using soft x-ray photoemission spectroscopy (SXPS) have provided insights into the electronic properties and thermal stability of the bimetallic systems. In the following paragraphs, we describe recent results and focus on three main aspects of the studies: surface structure and morphology, surface chemistry, and surface electronic properties. We begin with atomistic studies of faceting and reconstruction in bimetallic systems based on W(111) and discuss evidence that acetylene reactions over Pd/W(111) surfaces exhibit size effects on the nanometer scale. We use SXPS to show that interfacial mixing, even at 300 K, is observed for multilayers of Pt, Rh, Ir on W(111), whereas monolayer films are stable at high T. We report the oxygen-induced faceting of Ir(210), and describe structure sensitivity in C2H2 chemistry over clean planar and clean faceted Ir (210). We describe studies of metals on sulfided W(111), including self-limiting growth of Cu nanoclusters on specific surface sites. This program includes detailed microscopic studies of faceting of metal substrates induced by monolayer overlayer films. We believe the results of these studies are important in understanding the mechanisms of possible dynamic structural rearrangements at the surfaces of high area metallic and bimetallic catalysts under high-temperature operation. METHODS In this work we have used an array of ultrahigh vacuum surface science methods, including high-resolution scanning tunneling microscopy (STM), low energy electron diffraction (LEED), and
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC. This paper was submitted directly (Track II) to the PNAS office. Abbreviations: ML, monolayer; TPD, temperature programmed desorption; STM, scanning tunneling microscopy; LEED, low-energy electron diffraction; LEEM, low-energy electron microscopy; SXPS, soft x-ray photoemission spectroscopy; HRSXPS, high-resolution SXPS; L, Langmuir. †To whom reprint requests should be addressed. E-mail: [email protected].
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
NANOSCALE SURFACE CHEMISTRY
6504
low-energy electron microscopy (LEEM) for structure and morphology; temperature-programmed desorption (TPD) for surface chemistry; and high-resolution SXPS (HRSXPS) using synchrotron radiation for electronic properties.
Fig. 1. STM image of fully faceted W(111) surface: Pt coverage is ≈1.1 physical ML, annealed at 1,200 K for 1 min. Field of view is 2,000 Å×2,000 Å. Sample bias: +1.5 V, tunneling current: 1.5 nA. STM images of Figs. 1 and 2 are displayed in XSLOPE mode, which gives a three-dimensional top view perspective of the pyramids with {211} facets. [Reproduced with permission from ref. 13 (Copyright 1999, World Scientific).]
RESULTS AND DISCUSSION Faceting of W(111) Induced by Ultrathin Metal Films. We have used two microscopic methods to provide insights into the mechanism of metal film-induced faceting of bcc(111) surfaces, and have found evidence for the nucleation and growth of faceted regions (12, 13). LEEM and STM have been used to investigate the faceting of W(111) as induced by Pt. The atomically rough W(111) surface, when fully covered with a monolayer film of Pt and annealed to temperatures higher than ≈750 K, experiences a significant morphological restructuring: the initially planar surface undergoes a faceting transition and forms three-sided pyramids with {211} faces as seen clearly in the STM image of Fig. 1. In complementary studies, LEEM can distinguish between planar and faceted surfaces based on the different types of diffraction of lowenergy electrons on surfaces with different morphologies, with up to ≈70-Å lateral resolution. LEEM measurements (12, 13) of Pt/W(111) demonstrate that the transition from planar to faceted structure proceeds through the nucleation and growth of spatially separated faceted regions. The surface remains planar for Pt coverages less than 2/3 ML (1 ML=1.7×1015 atoms per cm2). As the Pt coverage increases above 2/3 ML on the heated W surface, local islands of Pt with coverage of 1 ML are able to nucleate, and it is there that facets form. When the entire surface has a coverage of 1 ML, the surface is fully faceted. STM data (Fig. 2) confirm the LEEM observations that a partially faceted Pt/W surface is a combination of large planar regions with scattered faceted regions. The faceted regions include pyramids of two distinct size distributions (Fig. 2): large individual pyramids or clusters of large pyramids, surrounded by smaller size satellite pyramids. LEEM and STM prove to be excellent complementary microscopic techniques in the study of faceting. STM provides structural information down to atomic scale. Although the resolution of LEEM does not match that of an STM, LEEM has several advantages that make it a most useful tool in studying the kinetics of large scale morphological transformations: it is capable of imaging surfaces at very high temperatures, and the surface is easily observed during metal deposition.
Fig. 2. STM image (5,000 Å×5,000 Å) of a partially faceted surface. Pt coverage is ≈0.8 physical ML, annealed at 1200 K for 1 min. Relatively large pyramids (200–700 Å in size) are scattered on the otherwise smooth planar surface, sometimes standing individually, sometimes forming large clusters that are several thousand Angstroms in extent. [Reproduced with permission from ref. 13 (Copyright 1999, World Scientific).]
There is ample documentation that the atomically rough W(111) surface develops nanometer-scale facets with W{211} orientation when covered by 1 ML films of Pd, Pt, Rh, or Ir, followed by annealing (10, 11). In recent experiments, we have found that the metal film-covered W (211) surface itself may undergo a faceting transition (14). A W(211) surface covered with a thin film (between 0.5 and 1 physical monolayer) of Rh, Pt or Pd is found to exhibit an n×1 superstructure when annealed above a threshold temperature of ≈900 K (500 K for Pd). The superstructure is observed by using low-energy electron diffraction: phase diagrams have been measured to indicate the coverage range and the temperature threshold where the new structure appears. Scanning tunneling microscopy results indicate that in the case of Pd/W(211) the superstructure phase is caused by missing overlayer rows (Fig. 3), but in the case of Rh/W(211) it is more likely caused by a microfaceting of the surface into {110} faces (14). Overlayer coverages >1 ML are thermally unstable, and form ultrathin alloy films on annealing. Alloy formation for thermally annealed multilayers of Rh, Pd, Pt on W(211) is confirmed by synchrotron radiation studies (see below). To provide guidance for future faceting studies of planar surfaces, we have been collaborating with the group of R. Blaszczyszyn to conduct field-emission microscopy studies (15). In this work, a nearly hemispherical emitter tip, coated with metallic overlayers, develops facets on annealing. The typical
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
NANOSCALE SURFACE CHEMISTRY
6505
emitter has a diameter of ≈200 nm; the shape and dimensions are a good approximation to a single catalyst particle, although a bit larger. Recent studies have focused on ultrathin films of Pt on W, Pt on Ir, and Pd on Mo (16). In general, on annealing the metal film-covered emitter tip, the shape changes from nearly hemispherical to more polyhedral. The faceting effect, i.e., growth of facets, is particularly pronounced for the Pt/W and Pd/Mo systems, and less so for Pt/Ir [consistent with the fact that we have not yet found evidence for metal-induced faceting of planar Ir(210); see below]. The facets that form on bcc W and Mo are {112}, {123} and {178}, whereas {116} and {115} grow on fcc Ir.
Fig. 3. STM scan ≈0.8 ML of Pd on W(211), annealed at 1000 K (1,000 Å× 1,000 Å). Dark vertical lines are identified as missing rows of Pd atoms; bright spots decorating the missing rows are attributed to adsorbed background gases. The average separation between missing rows is ≈50 Å, consistent with the ≈(11×1) superstructure seen using LEED. [Reproduced with permission from ref. 14 (Copyright 2001, Am. Chem. Soc.).]
Surface Chemistry of Bimetallic Pd/W. The surface chemistry of small hydrocarbons on transition metal surfaces is relevant to many important catalytic processes. We focus here on acetylene chemistry, for which reaction pathways are known to be sensitive to surface structure (ref. 17; for a review see ref. 18). Acetylene cyclization, 3C2H2→C6H6, is catalyzed by palladium and has been studied extensively on clean and modified surfaces, under ultrahigh vacuum as well as high-pressure conditions. Acetylene cyclization is extremely structure-sensitive, so it is an ideal reaction for probing the effects of surface electronic and geometric structure. Isotopic studies indicate that benzene forms from acetylene in a stepwise fashion on Pd(111) and Cu(110); the reaction proceeds via a C4 intermediate species, and the reactants do not undergo C —C or C—H bond scission (17–19). Bimetallic systems can improve both activity and selectivity. For example, we have shown in a combination high-resolution electronenergy-loss-spectroscopy (HREELS) and TPD study that a single ML of Pd on W(211) decreases the high reactivity of W for C2H2 decomposition, and catalyzes self-hydrogenation of C2H2 to C2H4 and cyclization of C2H2 to C6H6 (20). Evidence for finite size effects in benzene formation has been reported by Goodman et al. (21) for Pd clusters supported on alumina films; Lambert et al. (22) have shown that there is a minimum ensemble on Pd(111) (≈7 atoms) necessary for benzene formation. Although it is clear that electronic structure plays an important role in the cyclization reaction, these studies suggest questions: can electronic structure alone be responsible for these known size effects? What are the effects of geometrical parameters, e.g., size and shape? Zhdanov and Kasemo (3) have recently performed Monte Carlo analyses on the model catalytic reaction 2A+2B→2AB, which indicate that the reaction kinetics on a faceted nanocrystal can be different from those on a single crystal surface. The simulations have identified a kinetic “structure gap” not associated with special electronic effects or properties of small particles, but with their size and shape. They argue that geometric structure alone can play an important role in reaction kinetics. It is this idea that we are testing. We have found that acetylene reactions over Pd/W(111) surfaces exhibit size effects on the nanometer scale (23). In these studies, we have characterized the self-hydrogenation of C2H2 to form C2H4 and cyclization of C2H2 to form C6H6 on planar and faceted Pd-covered W(111) and on Pd-covered W(112). The substrates ranged from 1 ML of Pd on planar W(211), to faceted Pd/W(111) surfaces containing different facetsize distributions. The goal is to probe the influence of facet size on reactivity. Below ≈700 K annealing temperature, the Pd/W(111) surface remains planar. Above ≈700 K, Pd-covered W(112) facets are formed, which grow in size with increasing annealing temperature. We have measured TPD spectra for self-hydrogenation of adsorbed C2H2 to form C2H4 product (Fig. 4). The ethylene TPD spectra for Pd/W(111) evolve in a systematic fashion as the surface is thermally converted into a faceted substrate, and the facets grow in size. For the largest facets (annealing T≈1,000 K), the ethylene TPD spectrum is similar to that for C2H4 formation over planar Pd/W(112) (Fig. 4 Upper). Similar facetsize effects are seen for acetylene cyclization to form benzene (Fig. 4 Lower). The data provide clear evidence that size effects at the nanometer scale (3) should be considered in evaluation of reactivity data for faceted bimetallic surfaces. Synchrotron Radiation Studies of Metals on W. During the last several years, we have found that certain ultrathin films (≈1 physical ML) on W(111) and Mo(111) substrates can induce faceting (e.g., Rh, Pd, Ir, Pt, Au, as well as O, Cl), whereas others do not [e.g., Ti, Co, Ni, Cu, Ag, Gd (10, 11)]. We noted a correlation, namely, that the elements that cause faceting have Pauling electronegativity >2.0, whereas the elements that do not cause faceting have electronegativities <2.0; W and Mo each have negativity=1.7. This observation suggested that electronic structure plays a role in faceting. To search experimentally for possible electronic factors that influence faceting, we use HRSXPS based on synchrotron radiation. We study (mainly) the sharp W 4f, Pt 4f, Ir 4f, etc., and measure surface core level shifts (SCLS). Based on an extensive set of measurements, we can draw a number of conclusions. (i) Although the SCLS is an extremely sensitive and useful indicator of interface formation, there is no clear correlation between Pauling electronegativity and surface or interfacial 4f7/2 binding energies (24). This finding was not unexpected, because the Pauling electronegativity is a measure of initial-state charge transfer effects, whereas the measured 4f-binding energies are influenced by a combination of initial and finalstate effects. (ii) Metals with the highest heats of adsorption (e.g., Pt, Pd), and the highest W4f interfacial binding energies are those which cause faceting (24). This finding is consistent with a first-principles theory that indicates that a higher heat of adsorption provides both a strong thermodynamic driving force for faceting and a lower kinetic barrier to faceting (25, 26). (iii) For most systems studied [Cu, Ni, Rh, Pd, Ir, Pt, Au on
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
NANOSCALE SURFACE CHEMISTRY
6506
W(111) and W(112); refs. 27–29], we find evidence that a 1 ML of overlayer metal is thermally stable, and floats on the outer surface without significant alloy formation, for all temperatures up to the onset of desorption. For all systems except Au/W, we find that multilayer films form alloys on annealing: invariably, W atoms from the substrate diffuse into the overlayer film, rather than vice versa. In certain cases (Pt/W, Ir/W) we can measure sharp 4f levels of both overlayer and substrate, and find evidence for alloy formation in multilayer films (Fig. 5). In general, the alloying behavior of the bimetallic systems investigated here is consistent with the known bulk-phase diagrams (e.g., Pt is not soluble in W, but W is soluble in Pt to a maximum of ≈60% W). Moreover, we have used Born-Haber cycles and the equivalent core approximation to extract thermochemical data concerning energetics of adhesion, segregation and alloying in these systems (28).
Fig. 4. (a and b) Surface chemistry of acetylene on planar and faceted Pd/W surfaces. (Top) TPD spectra for C2H4 reactively formed after adsorption of a saturation coverage of C2H2 at 100 K onto Pd-covered W (C2H2 exposure ≈3× 10−6 torr·s). In each case, the metal is precovered with 1 ML of Pd and annealed to the indicated temperature, before deposition of C2H2. The sizes of {211} facets increase with annealing temperature >700 K. (Bottom) TPD spectra for C6H6 reactively formed after adsorption of C2H2 on Pd/W surfaces, as in a.
(iv) We find evidence for intermixing at the interface for multilayer deposition of several metals (Pt, Ir, Rh) on W(111) at room temperature (29)! This finding is surprising to those in the thin film community (e.g., for those studying magnetic thin films) who generally believe that thin films of high-melting-temperature materials form abrupt interfaces. Faceting and Surface Chemistry of Ir(210). As part of a larger effort to study the morphological stability of adsorbate covered metallic surfaces, we have investigated the influence of various adsorbates on fcc Ir(210). The structure of the atomically rough Ir(210) surface is similar to that of bcc W(111), but with reduced symmetry (2-fold). Based on studies using LEED and STM, we find that oxygen overlayers induce substantial facets on Ir(210), whereas the metal overlayer studied to date (Au) does not cause faceting.
Fig. 5. Evidence for formation of interfacial alloy in Pt 4f and W 4f SXPS spectra associated with an annealing sequence for an 8 ML Pt film on W(211). Annealing time is 1 min at each temperature. Features on high binding energy sides of each 4f peak indicate that W atoms diffuse through the interface to form a Pt-W alloy film. [Reproduced with permission from ref. 28 (Copyright 2000, Am. Phys. Soc.)]
When Ir(210) is exposed to more than ≈0.9 L (1 L=1×10−6 torr·s; 1 torr=133 Pa) of oxygen and annealed to T>600 K, it experiences significant morphological restructuring: nanometer-scale pyramid-like structures (facets) are formed on the initially planar surface (I.E., K.P., and T.E.M., unpublished data.). LEED measurements show that the pyramids have three sides with mirror symmetry (two {311} facets and one {110} facet), and that the faceted surface exhibits a quasi-reversible behavior on annealing to higher temperatures. The surface reverts to its planar state at temperatures above 850 K but, provided the maximum annealing temperature is below the desorption temperature of oxygen, facets reappear on cooling to temperatures below 800 K. Furthermore, we are able to remove the oxygen from the surface by means of catalytic oxidation of CO at ≤580 K or by means of the H2+O reaction at ≤400 K, while preserving the faceted structure. TPD and Auger electron spectroscopy have shown that residual adsorbed O and CO (or H) are negligible after this procedure. The faceted clean surface is stable up to 600 K, but irreversibly reverts to the planar state above 600 K. These experiments indicate that the clean, faceted, metastable Ir(210) surface provides an ideal substrate to study thermal relaxation of nanometer-scale surface features. In a previous section we reported that C2H2 reacts on planar and faceted Pd/W to form C2H4 and C6H6 products. In contrast, the clean Ir (210) surface is considerably more aggressive in
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
NANOSCALE SURFACE CHEMISTRY
6507
dissociating C2H2 than the Pd/W surfaces. We find that adsorption of C2H2 on Ir(210) at either 100 K or 300 K leads to dissociation when the surface is heated, and that the dominant desorption product (>99%) in TPD spectra is H2. Traces of C2H4 are seen at low temperature in TPD.
Fig. 6. TPD spectra for H2 reactively formed after adsorption of a saturation coverage of C2H2 onto planar Ir(210) (curve a) and onto the clean, faceted Ir(210) surface (curve b). In each case, the C2H2 dose is ≈3×10−6 torr·s (3 L) onto the substrate at 300 K. The differences in the TPD spectra illustrate the structure-sensitivity of C2H2 surface chemistry.
TPD spectra for desorption of H2 from hydrogen-dosed Ir(210), and desorption of H2 from acetylene-dosed Ir(210) exhibit differences that indicate clearly that desorption of H2 from C2H2-dosed Ir is reaction-rate limited (W.C., I.E., Q.W., and T.E.M., unpublished data). This reactivity is almost certainly caused by formation of intermediate fragments (30, 31) that may dissociate in a stepwise fashion on heating (−CCH? −CCH3?). Fig. 6 provides vivid evidence for structure sensitivity in C2H2 decomposition over Ir surfaces. Here, we compare TPD spectra of H2 after C2H2 deposition on clean planar Ir(210) and on clean faceted Ir(210); the clean faceted surface was prepared as described above. There are striking differences in the TPD spectra that arise from the nanoscale surface structure on the faceted surface. We are using HREELS and HRSXPS in an attempt to identify the stoichiometry and concentration of the intermediates on the planar and faceted surfaces. Nucleation of Nanoscale Clusters of Cu on S(4×4) W(111). Studies of sulfided surfaces of W may have interesting implications for hydrodesulfurization (HDS) catalysis, a process of great importance for removing S-impurities from petrochemicals. The HDS process is typically catalyzed by Mo or W sulfides that are promoted with group VIII transition metals. In earlier studies of the sulfided W(111) surface, we found that S induces an unusual reconstruction with (4×4) periodicity (32). The sulfurcovered W(111) surface (S(4×4)/W(111) is characterized by broad, planar terraces ≈30 nm in width. We recognized the possibility that this highly corrugated sulfided surface might offer the opportunity for site-dependent nucleation of metal nanoclusters; such arrays of nanoclusters may, in turn, have unusual chemical properties. We have now studied the interactions of vapor-deposited Cu with the highly ordered S(4×4)/W(111) surface, by means of STM, LEED, and Auger electron spectroscopy (33). We found that fractional monolayers of Cu grow homogeneously as clusters on S(4×4)/W(111) over a wide Cu coverage range. At low Cu coverages (<0.1 ML), Cu nanoclusters are found to nucleate preferentially at characteristic 3-fold hollow sites on the sulfided surface; there is a clear energetic preference for one type of site over others (Fig. 7). The formed Cu nanoclusters are uniform in size as coverage increases, indicating self-limiting growth (the clusters are ≈0.5 nm in diameter, and contain ≈3 atoms). The selflimiting growth may be attributed to a repulsive interaction between diffusing Cu atoms and stable nanoclusters arising from lattice mismatch between the Cu metal and the sulfided substrate. As Cu coverage increases ≥0.1 ML), cluster formation and growth occur without limitation on other adsorption sites (atop and vacancy).
Fig. 7. STM image (100 Å×100 Å) of 0.05 ML of Cu on S(4×4)/W(111). Nanoclusters with uniform size are found to nucleate preferentially in characteristic 3-fold hollow sites. Nanoclusters are estimated to be ≈5 Å in diameter; a few examples are highlighted with circles in the image.
The observation of unusual growth for a metal on sulfided W(111) is an impetus for further studies of catalytically active metals (Co, Ni) that have direct relevance to the fundamentals of hydrodesulfurization catalysis. Concluding Remarks. The main factor that distinguishes this work from other studies of model bimetallic and sulfide catalysts is our emphasis on atomically rough, high-surface-energy surfaces that may be morphologically unstable during reaction conditions. For example, the observation of faceting demonstrates that the surface is not a rigid template, but it may undergo massive structural rearrangements when covered by a metallic film under reaction conditions. The exposed surfaces may be quite different from those present in the absence of the overlayer metal. We believe that these results are important for understanding dynamic structural rearrangements at the surfaces of high area bimetallic and sulfide catalysts, and in clarifying the role of nanometer-scale size effects in surface reactions. We acknowledge valuable discussions with and contributions of various collaborators, including R.Blaszczyszyn, J.G.Chen, C.-T.Chan, G.L. Kellogg, and J.B.Hannon. This work has been supported in part by the U.S. Department of Energy, Division of Chemical Sciences, and by U.S. Army Research Office, Durham, NC.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
NANOSCALE SURFACE CHEMISTRY
6508
1. Valden, M., Lai, X. & Goodman, D.W. (1998) Science 281, 1647–1650 2. Heiz, U., Sanchez. A., Abbet, S. & Schneider, W.-D. (1999) J. Am. Chem. Soc. 121, 3214–3217. 3. Zhdanov V.P. & Kasemo, B. (1998) Surf. Sci. 405, 27–37. 4. Sinfelt, J.H. (1983) Bimetallic Catalysts (Wiley, New York). 5. Anonymous (1982) Advances in Multimetallic Catalysts (Catalytica Associates, Inc., Santa Clara, CA). 6. Kuznetsov, B.N., Yermakov, Yu.I., Boudart, M. & Collman, J.P. (1978) J. Mol. Catal. 4, 49–57. 7. Leclercq, C., Romero, S., Pietrzyk, T.S., Grimblot, J. & Leclercq, L. (1984) J. Mol. Catal. 25, 67–86. 8. Yermakov, Yu.I., Kuznetsov, B.N. & Ryndin, Yu.A. (1976) J. Catal. 42, 73–78. 9. Trunschke, A., Ewald, H., Gutschick, D., Miessner, H., Skupin, M., Walther, B. & Böttcher, H.-C. (1989) J. Mol. Catal. 56, 95–106. 10. Madey, T.E., Guan, J., Nien, C.-H, Tao, H.-S., Dong, C.-Z. & Campbell, R.A. (1996) Surf. Rev. Lett. 3, 1315–1328. 11. Madey, T.E., Nien, C.-H., Pelhos, K., Kolodziej, J.J., Abdelrehim, I.M. & Tao, H.-S. (1999) Surf. Sci. 438, 191–206. 12. Pelhos, K., Hannon, J.B., Kellogg, G.L. & Madey, T.E. (1999) Surf. Sci. 432, 115–124. 13. Pelhos, K., Madey, T.E., Hannon, J.B. & Kellogg, G.L. (1999) Surf. Rev. Lett. 6, 767–774. 14. Pelhos, K., Abdelrehim, I.M., Nien, C.- H. & Madey, T.E. (2001) J. Phys. Chem. B 105, 3708–3717. 15. Pelhos, K., Madey, T.E. & Blaszcszyszyn, R. (1999) Surf. Sci. 426, 61–68. 16. Antczak, G., Madey, T.E., Blaszcszyszyn, M. & Blaszcszyszyn, R. (2001) Vacuum 63, 43–51. 17. Omerod, R.M., Lambert, R.M., Bennett, D.W. & Tysoe, W.T. (1995) Surf. Sci. 330, 1–10. 18. Abdelrehim, I.M., Pelhos, K., Eng J., Jr., Chen, J.G. & Madey, T.E. (1998) J. Mol. Catal. A 131, 107–120. 19. Lambert, R.M. & Ormerod, R.M., (1994) in Surface Reactions, Springer Series in Surface Science, ed. Madix, R.J. (Springer, Berlin), Vol. 34, pp. 89–134. 20. Abdelrehim, I.M., Pelhos, K., Madey, T.E., Eng, J., Jr. & Chen, J.G. (1998) J. Phys. Chem. B 102, 9697–9707. 21. Holmblad, P.M., Rainev, D.R. & Goodman, D.W. (1997) J. Phys. Chem. B 101, 8883–8886. 22. Baddeley, C.J., Tikhov, M., Hardacre, C., Lomas, J.R., Lambert, R.M. (1996) J. Phys. Chem. 100, 2189–2194. 23. Barnes, R. Abdelrehim, I.M. & Madey, T.E. (2001) Topics Catal. 14, 53–61. 24. Tao, H.- S., Rowe, J.E. & Madey, T.E. (1998) Surf. Sci. 407, L640–L646. 25. Che, J.G., Chan, C.T., Kuo, C.H. & Leung, T.C. (1997) Phys. Rev. Lett. 79, 4230–4233. 26. Nien, C.-H., Madey, T.E., Tai, Y.W., Leung, T.E., Che, J.G. & Chan, C.T. (1999) Phys. Rev. B 59, 10335–10340. 27. Kolodziej, J.J., Pelhos, K., Abdelrehim, I.M., Keister, J., Rowe, J.E. & Madey, T.E. (1999) Prog. Surf. Sci. 59, 117–134. 28. Kolodziej, J.J., Madey, T.E., Keister, J.W. & Rowe, J.E. (2000) Phys. Rev. B 62, 5150–5162. 29. Kolodziej, J.J., Madey, T.E., Keister, J.W. & Rowe, J.E. (2002) Phys. Rev. B, in press. 30. Eng, J., Jr., Chen, J.G., Abdelrehim, I.M., Madey, T.E. (1998) J. Phys. Chem B 102, 9687–9696. 31. Marinova, Ts.S., & Kostov, K.L. (1987) Surf. Sci. 181, 573–585. 32. Nien, C.-H., Abdelrehim, I.M. & Madey, T.E. (1999) Surf. Rev. Lett. 6, 77–96. 33. Wu, Q., Chen, W. & Madey, T.E. (2002) J. Phys. Chem., in press.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
MAGNETIC NANODOTS FROM ATOMIC FE: CAN IT BE DONE?
6509
Colloquium Magnetic nanodots from atomic Fe: Can it be done?
E.te Sligte†, R.C.M.Bosch, B.Smeets, P.van der Straten, H.C.W.Beijerinck, and K.A.H.van Leeuwen Department of Applied Physics, Eindhoven University of Technology, P.O. Box 513, 5600 MB, Eindhoven, The Netherlands Edited by Jack Halpern, University of Chicago, Chicago, IL, and approved February 7, 2002 (received for review October 5, 2001) Laser focusing of Fe atoms offers the possibility of creating separate magnetic structures on a scale of 10 nm with exact periodicity. This can be done by using the parabolic minima of the potential generated by a standing light wave as focusing lenses. To achieve the desired 10-nm resolution, we need to suppress chromatic and spherical aberrations, as well as prevent structure broadening caused by the divergence of the incoming beam. Chromatic aberrations are suppressed by the development of a supersonic Fe beam source with speed ratio S=11±1. This beam has an intensity of 3×1015 atoms sr−1 s−1. The spherical aberrations of the standing light wave will be suppressed by aperturing with beam masks containing 100-nm slits at 744-nm intervals. The beam divergence can be reduced by application of laser cooling to reduce the transverse velocity. We have constructed a laser system capable of delivering over 500 mW of laser light at 372 nm, the wavelength of the 5D4→5F5 atomic transition of 56Fe we intend to use for laser cooling. Application of polarization spectroscopy to a hollow cathode discharge results in a locking system holding the laser continuously within 2 MHz of the desired frequency. Laser focusing of atoms is being studied with increasing interest as it proves a viable technique to produce periodic nanostructures, especially because the period of the structures is known with great precision. The two basic lithographical processes are etching and deposition. Etching with the use of metastable rare gases, such as He* (1, 2), Ne* (3), and Ar* (4), as well as with alkali atoms like Na (5) and Cs (6) in combination with self-assembling monolayer (SAM) resists has produced arrays of lines and dots. Deposition of similar structures has been achieved with metal atoms such as Cr (1, 7, 8) and Al (9) and is being planned for other group III metals (10). The present work aims to deposit magnetic nanostructures made of Fe atoms. This method could provide a fascinating new experimental approach to the field of one- and zerodimensional magnetism (11). For this scheme to work, we would have to deposit discrete, isolated structures of Fe. Atom lithography is practiced in general by focusing atoms in the periodic potential created by a standing laser light wave, as depicted in Fig. 1. Atoms exposed to a nearly resonant light field experience a dipole force as a result of the electric field of the standing wave. This standing wave results in a sinusoidal potential. The “sharpness” of the focused image is determined in atom optics as in conventional optics by the incoming beam collimation and chromatic aberration. These two aspects are determined by the transverse and axial velocity spreads, respectively. A third factor, which causes the contrast of the image to be degraded (i.e., atoms are also deposited in between the focus positions), is that the potential is sine-shaped, and thus not purely parabolic. This effect is called spherical aberration in analogy with conventional optics. Having optimized the properties of the incoming atom flux, we are still limited by the surface diffusion and the growth pattern of the iron on the surface. We suspect that the effects of surface diffusion are restricted (i.e., an increase of the full width at half maximum of the deposited structures below 2 nm), provided the contrast of the image is good and the substrate material is chosen such that the mobility of Fe atoms on the substrate is low. The problem of surface diffusion is thus coupled with the problem of spherical aberration. All of these problems will have to be solved if an array of truly separate, nanometer-sized magnets is to be created. Atom beam properties can be manipulated through laser cooling (12). In laser cooling, an atom absorbs photons and reemits them in a random direction. If the photons are part of a laser beam, they all have the same momentum and thus will transfer a net momentum to the atom. If the laser frequency is slightly below that of the transition, atoms moving along the direction of the laser will absorb fewer photons than atoms moving in the opposite direction because of the detuning induced by the Doppler effect. Two counterpropagating laser beams thus have the net effect of transversely cooling an atomic beam. Laser cooling normally requires a closed, two-level transition. The Fe atom has no closed twolevel transition available for laser cooling. We attempt laser cooling via the 5D4→5F5 transition, at a wavelength of 372 nm. The total “leak” in this transition is 1/243, i.e., when an atom is excited, there is a chance of 1 in 243 that it will not decay back to the ground state from where it came. If this occurs, we cannot apply any further laser cooling or focusing to the atom. First, the chromatic aberration is reduced by decreasing the axial velocity spread. Zeeman slowing (6) cannot be applied in this case because that technique requires very large numbers of absorptions. Therefore, the atoms have to leave the source with a very narrow velocity distribution. Supersonic sources have such a characteristic narrow velocity distribution (13). The second obstacle is the beam divergence. We intend to reduce the transverse velocity spread by transverse laser cooling (14), as shown schematically in Fig. 1. Our simulations show that a beam divergence of 0.2 mrad can be achieved in this way, whereas the atoms undergo less than 100 absorptions. Third, we need to suppress the spherical aberrations resulting from the potential not being parabolic in most places. We intend to achieve this by means of the use of separately produced beam masks that block the beam in places except where the potential is harmonic in good approximation. This approximation is shown in Fig. 1. This article aims to give an experimental overview of the progress made so far. We will begin by describing the design of our experiment in Methods. This description will be followed by a discussion of the operating characteristics of the experimental apparatus in Results and by a brief discussion of these results, as well as an outlook into the near future, in Discussion. METHODS In this section, we give an overview of the design of our experiment. The three main components are a supersonic atomic
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC. This paper was submitted directly (Track II) to the PNAS office. †To whom reprint requests should be addressed. E-mail: [email protected].
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
MAGNETIC NANODOTS FROM ATOMIC FE: CAN IT BE DONE?
6510
Fe source, a laser system capable of continuously generating over 500 mW of UV light, and beam masks with a periodicity to match the wavelength of the light used to focus the Fe atoms.
Fig. 1. Principle of Fe atom lithography. The Fe atoms exit a supersonic beam source. They are then collimated by laser-cooling techniques. After blocking of the unsuitable atoms by a beam mask, they will be focused in a periodic potential generated by a standing wave. We intend to use this scheme to create one- or zero-dimensional ferromagnetic nanostructures.
Beam Source. The source design reviewed in this section has been described more thoroughly by Bosch et al. (15). We use a design based on similar sources developed by Hagena (16). The prime requirement of our source design is that the Fe atoms must have a uniform velocity distribution to reduce the chromatic aberrations in focusing. The uniformity of the velocity distribution is expressed in terms of the speed ratio S, defined as the ratio of the final flow velocity u and the parallel velocity spread α∥. This speed ratio is of the order unity for thermal sources; supersonic sources can have far greater speed ratios. We therefore use a supersonic source. Unfortunately, supersonic sources require inlet gas pressures of 10–104 mbar (millibar; 1 mbar=100 Pa). Fe would require unrealistically high temperatures to have such a vapor pressure. We have therefore chosen to use Fe as a seed gas in a supersonic expansion of Ar. With the use of a seeded supersonic source, we have drastically reduced the Fe vapor pressure required. We also demand that the Fe flux out of the source be sufficient to achieve a reasonable deposition rate. This rate can be achieved with operating temperatures around 2000 K. Several crucible designs have been tested, made of three different materials. High-density graphite turned out to dissolve in the liquid Fe. Boron nitride (BN) sources were corroded by the molten Fe. The only material that proved resistant to Fe and the high temperatures required was highly purified alumina (Al2O3), capable of withstanding temperatures up to 2200 K, its sublimation temperature. However, alumina is extremely hard and difficult to machine.
Fig. 2. Supersonic Fe atom source, with the oven and the gas source mounted on two separate flanges (1 and 2), along with copper connectors for the heating current (3). The Fe is seeded in a supersonic Ar expansion. The Ar flows into the source from a Ta gas inlet (4). A thermocouple (5) measures the temperature of the source. The source (6) is made of alumina and heated by a graphite heating coil (7). The oven is insulated by 20 layers of Ta foil (8) and its exterior is water-cooled (9). The beam is extracted from the source chamber by a skimmer (10).
The source currently in operation is depicted in Fig. 2. Because of the material it is made of, the crucible design has been kept extremely simple. Heating is applied externally by inserting the crucible into an oven, which consists of a doubly wound graphite heating coil. To prevent clogging, the last winding, located at the nozzle, has been made thinner: 1.5×5 mm2 instead of 3×5 mm2 for the other windings. Power losses are minimized by application of extensive heat shielding: 20 layers of tantalum foil around the circumference of the cylindrical oven and 5 layers at the nozzle side. A hole is left at the nozzle side to allow for expansion of the gas. Radiation through this hole is the dominant power loss process. The alumina crucible itself consists of two parts, an inner tube and an outer tube. The Ar flows into the inner tube from the gas inlet. As it flows through the inner tube, it is heated to the source operating temperature. It leaves the inner tube through a 1-mm orifice, entering the source chamber. In the source chamber, the Fe vapor and the Ar gas mix. Upstream diffusion of Fe vapor into the source system is prevented by Ar flowing through the small space between the inner and outer tubes. The Ar/Fe mixture then exits through a nozzle 230 µm in diameter. This departure causes a supersonic expansion of the argon gas. The argon expands into a chamber at 10−1 mbar; this pressure cannot be lower because of the large atomic flux into the chamber. Collision with the background gas causes the expansion to end in strong shock waves after 20–30 mm, called a Campargue expansion (17). To extract a steady beam from the source chamber, a conical skimmer is placed at a distance of 10–15 mm from the nozzle. The skimmer extracts part of the beam, which then expands into a far better vacuum (10−4 mbar) and does not encounter a shock front. The Fe in this beam will be used to deposit structures in a separate deposition chamber held at a background pressure of 10−8 mbar. Laser System. For laser cooling and focusing of Fe, we estimate that a laser power of about 500 mW is necessary. There are no commercial systems capable of delivering 500 mW at 372 nm. To obtain light of the right wavelength, we have frequency-doubled a commercially available Ti:S laser operating at 744 nm by using a doubling system built at the Free University of Amsterdam.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
MAGNETIC NANODOTS FROM ATOMIC FE: CAN IT BE DONE?
6511
Fig. 3. The laser system to be used for laser-cooling of Fe. Laser light at 744 nm is generated in a tunable Ti:S laser. It is fed into a ring cavity with a lithium triborate crystal in it. In the crystal, light at 372 nm is generated through second harmonic generation (SHG), which can be used for laser cooling. A small portion of the laser light is diverted to a hollow cathode discharge (HCD) where polarization spectroscopy (PS) generates the error signal used to keep the laser tuned to the wavelength of the 5D4→5F5 transition.
Frequency doubling is based on the nonlinear susceptibility of some materials. A full mathematical treatment (19) of the problem yields a quadratic dependence of second harmonic output power on the input power:
[1]
The nonlinear crystal we use to obtain the second harmonic is lithium triborate (LBO). To maximize the power passing through the crystal, the crystal is placed in a ring cavity as shown in Fig. 3. The ring cavity consists of four mirrors, three with a reflection coefficient R=0.999 and one with R=0.99. The Ti:S laser light enters the cavity through the R=0.99 mirror. The cavity length was locked to the incoming laser light wavelength with the use of the Pound-Drever-Hall technique (20). The finesse of the cavity was calculated at 469. Having obtained the necessary output power at 372 nm, we need a way to lock the laser wavelength to the wavelength of the 5D4→ 5F5 atomic transition. To do this, it is necessary to perform polarization spectroscopy on this transition (21). We need atomic Fe to observe this transition. In our setup, we generate the Fe atoms by sputtering from the cathode of a hollow cathode discharge. The discharge runs on He at 0.2 mbar; although this may seem a strange choice, because of the low sputtering probability for such a light element, it was chosen because other noble gases have relatively strong atomic transitions at or very near 372 nm. We applied polarization spectroscopy to this discharge. In polarization spectroscopy, Doppler broadening of the absorption profile is compensated for (21). This method theoretically enables us to obtain an error signal with a peak to peak width determined by the natural linewidth of the transition, Γ/2π=2.58 MHz for our target transition. This laser system should enable us to successfully perform laser cooling and focusing of Fe. Beam Masks. The potential induced by the standing light wave is a sine-type function. We intend only to use the near-parabolic minima and, to create structures with a zero background, block the rest of the potential. We intend to accomplish this with the use of beam masks. These beam masks would have holes or slits significantly less than half a wavelength-wide at a spacing of an integer number of wavelengths. The slit or hole dimension was chosen at 100 nm. The thickness of the masks has to be on the order of the slit size for reasons of manufacturability. This means that the mask pattern has to be etched into a 100-nm thick membrane. The membrane is suspended on a section of Si wafer. The choice of material is limited by the fact that after deposition on Si, most materials are under an internal strain. This strain must not be compressive to prevent buckling and subsequent deformation of the grids. The tensile strain must also be small enough to prevent snapping of the structures, and the material itself must be stiff enough to prevent tensile deformation. The material that meets these criteria best is SiN (22, 23). Table 1. Design parameters and standard operating conditions for the source Supersonic source Operating temperature T0 Ar inlet pressure p0 Flow rate
300–2200 K 133–1600 mbar 1019–1020s−1
Nozzle diameter Skimmer diameter Nozzle-skimmer distance Beam attenuation parameter
0.2–0.25 mm 0.5–0.8 mm 10–15 mm 10.7·10−4 mbar−1
d q
The masks are mounted on a section of standard Si[100] wafer. The etching of the pattern into the masks is done in two steps. First, the SiN film is covered with a resist layer. The desired pattern is etched into the resist layer by electron beam lithography. The second step is the actual etching itself of the SiN film by reactive ion etching. We expect the masks to allow sufficient suppression of spherical aberrations. RESULTS We continue with a discussion of the operating properties of the parts of the setup completed thus far, starting with the atomic Fe beam source. We continue to discuss the laser system and the beam masks. The robust source design used has lasted for over 2 years without noticeable deterioration. The crucible can operate without reloading for up to 200 h. The typical operating conditions of the source are summarized in Table 1. We have made a study of the properties of the Fe/Ar beam this source produces. The easiest way to determine the beam properties of an atomic Fe beam is by a time of flight method. We measured the intensity and velocity distribution of the Fe and Ar atoms with a mass spectrometer. In determining the optimum expansion characteristics, we must take into account the imperfect extraction of the beam by the skimmer. in the supersonic expansion is known from fluid dynamics theory (13). As the flow passes The ideal center line argon beam intensity through the skimmer, it is attenuated exponentially (18):
[2]
where q is the beam attenuation parameter, p is the operating pressure of the crucible, and I0 is the actual center line beam intensity. This behavior has been measured by monitoring the argon mass spectrometer signal at varying source pressure, and thus q has been determined. The value found was q=10.7·10−4 mbar−1. The same value was found by measuring the Fe signal. Given the Ar beam intensity and attenuation, there are two ways to deduce the iron beam intensity (15). One is to assume that the mass spectrometer detector sensitivity is the same for both species and multiply the Ar flux by the ratio of the Fe signal and the Ar signal. The other way is by multiplying the Ar flux with the ratio of the Fe and Ar source pressures. Both methods give results that agree to within a factor of 2. For both estimates, the Fe beam intensity lies between I (0)=1015s−1sr−1 and I (0)= 1016s−1sr−1, depending on the operating conditions. The time of flight setup has also been used to measure the velocity distribution. The average velocity was u=1400 m/s under standard
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
MAGNETIC NANODOTS FROM ATOMIC FE: CAN IT BE DONE?
6512
operating conditions (see Table 2). The speed ratio was found to be S=11±1 in this case. Table 2. Typical operating conditions and properties of the Ar-Fe beam Operating characteristics Source temperature T0 Ar inlet pressure p0 Fe vapor pressure pv Ar beam intensity I0,Ar Fe beam intensity I0,Fe Ar speed ratio SAr Fe speed ratio SFe Virtual source radius Rv Rv/z(z=1 m) Beam divergence
1930 K 1050 mbar 0.1 mbar 2·1019s−1 sr−1 3·1015 s−1 sr−1 11 11 0.25 mm 0.25 mrad
The UV output power of the laser system critically hinges on the finesse of the cavity, which was determined from transmission measurements to be 177±6. The cavity increases the laser power inside it by a factor of 90, and thus increases the 372-nm output power by a factor of 8100. The power output at 372 nm proved to depend quadratically on the input power into the cavity, with a conversion efficiency coefficient K=2.20±0.05×10−4 mW−1. This value enables the laser system to produce over 800 mW of 372-nm laser light if pumped with 2 W of red light. On a regular basis, 300 mW of light is produced at 1.4-W pump power. The iron density and temperature inside the hollow cathode discharge were measured by absorption spectroscopy. The absorption dip had an FWHM of 1.00 GHz and an amplitude of about 40%. The temperature was deduced from the Doppler width of the absorption dip to be 673 ±6 K. From the intensity of the absorption dip, the iron atomic density in the discharge was estimated at 3.2±0.2×1016m−3, corresponding to a partial Fe vapor pressure of 3×10−6 mbar. From our polarization spectroscopy setup, we obtained a dispersive error signal with a peak to peak width of 40 MHz. By using this error signal, we are able to keep our laser system in continuous lock within 2 MHz of the desired frequency. The beam masks produced have a period of 744.2±0.7 nm, twice the wavelength we intend to use. They have been made with a slit pattern as well as with a dot pattern. The lines are 100±4-nm-wide, and the dots have a diameter of 100±4 nm. The transmission masks cover a 250×250 µm area. An SEM picture of part of a sample beam mask for the deposition of lines is shown in Fig. 4. DISCUSSION The values for average speed and speed ratio in the source expansion agree well with the values predicted by H.C.W.B. and Verster (13). The speed ratio is high enough to suppress chromatic aberrations in the focusing section to a minimum focus width of about 10 nm, sufficient to deposit the Fe nanostructures we aim to construct. The Fe beam intensity of at least 1015s−1sr−1 is sufficient to deposit nanostructures in a reasonable deposition time, and the speed ratio of around 11 is sufficient to effectively suppress chromatic aberrations. Our laser system is capable of producing the 500 mW of continuous wave laser power needed and can be frequency-stabilized to within 1 line width. The difference between the theoretical and experimental value for the cavity finesse can be accounted for by stating that the calculation does not take into account absorption by the crystal and mirrors, and the possibility of imperfect transmission at the crystal faces. The 40-MHz-wide error signal (far greater than the predicted value of 2.58 MHz) can be explained with effects that increase the homogeneous linewidth, e.g., pressure broadening, Stark broadening, etc.
Fig. 4. SEM picture of a mask to be used for Fe line deposition. The lines are situated 744.2±0.7 nm apart and are 100±4-nm-wide.
The design of the masks leaves two problems. One is the finite lifespan of the masks; we intend to deposit significant layers of iron, which will also be deposited onto the masks. The internal stresses of the iron film can cause the masks to deform or even break when the iron film thickness and mask thickness become comparable. This effect limits the lifetime of the masks to several depositions. An ample supply of masks has been produced, therefore this is no serious impediment to our experiments. The second problem is the alignment of the masks relative to the standing light wave. This alignment would have to be stable to about 10 nm. We plan to achieve this stability by monitoring the light scattered by the iron atoms in the standing light wave. The fluorescence will be minimal if the atoms passing through the slit traverse the standing light wave at the nodes. This scheme allows the position and the orientation of the mask with respect to the light wave to be adjusted and stabilized by using piezo-electric translators. We conclude that currently our project has achieved three main results: an operational supersonic Fe beam source, an operational laser system delivering 500 mW of UV laser light, and beam masks that are ready for use. All three have been developed especially for this project. With the laser and Fe source operational, we have achieved laser cooling of Fe. To our knowledge, the laser cooling of Fe has never been done before. We thank W.Hogervorst and coworkers at the Laser Lab of the Free University of Amsterdam for the design and construction of the frequency-doubling cavity. We also thank R.Navarro, M.F.A.Eurlings, and J.T.M.van Beek for design and production of the beam masks. This work is financially supported by the Dutch Foundation for Fundamental Research on Matter (FOM). 1. Brezger, B., Schulze, Th., Drodofsky, U., Stuhler, J., Nowak, S., Pfau, T. & Mlynek, J. (1997) J. Vac. Sci. Technol. B15, 2905–2911. 2. Petra, S.J.H., Feenstra, L., Vassen, W. & Hogervorst, W., International Quantum Electronics Conference/Conference on Lasers and Electron Optics, 12–14 September, 2000, Nice, France. 3. Engels, P., Salewski, S., Levsen, H., Sengstock, K. & Ertmer, W. (1999) Appl. Phys. B 69, 407–412. 4. Berggren, K.K., Bard, A., Wilbur, J.L., Gillaspy, J.D., Helg, A.G., McClelland, J.J., Rolston, S.L., Phillips, W.D., Prentiss, M. & Whitesides, G.M. (1995) Science 269, 1255–1257. 5. Timp, G., Behringer, R.E., Tennant, D.M. & Cunningham, J.E. (1992) Phys. Rev. Lett. 69, 1636–1639. 6. Lison, F., Adams, H.K., Schuh, P., Haubrich, D. & Meschede, D. (1997) Appl. Phys. B 65, 419–421.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
MAGNETIC NANODOTS FROM ATOMIC FE: CAN IT BE DONE?
6513
7. McClelland, J.J., Scholten, R.E., Palm, E.C. & Celotta, R.J. (1993) Science 262, 877–880. 8. Andersson, W.R., Bradley, C.C., McClelland, J.J. & Celotta, R.J. (1999) Phys. Rev. A 59, 2476–2485. 9. McGowan, R.W., Giltner, D. & Lee, S.A. (1995) Opt. Lett. 20, 2535–2537. 10. Rehse, S.J., McGowan, R.W. & Lee, S.A. (2000) Appl. Phys. B 70, 657–660. 11. Himpsel, F.J., Ortega, J.E., Mankey, G.J. & Willis, R.F. (1998) Adv. Phys. 47, 511–597. 12. Metcalf, H.J. & van der Straten, P. (1999) Laser Cooling and Trapping (Springer, Berlin). 13. Beijerinck, H.C.W. & Verster, N.F. (1981) Physica 111C, 327–352. 14. Hoogerland, M.D., Driessen, J.P.J., Vredenbregt, E.J.D., Megens, H.J.L., Schuwer, M.P., Beijerinck, H.C.W. & van Leeuwen, K.A.H. (1996) Appl. Phys. B 62, 323–327. 15. Bosch, R.C.M., Beijerinck, H.C.W., van der Straten, P. & van Leeuwen, K.A.H. (2002) Eur. Phys. J. Appl. Phys., in press. 16. Hagena, O.F. (1991) Z. Phys. D 20, 425–428. 17. Campargue, R. (1964) Rev. Sci. Instrum. 35, 111–112. 18. Shen, Y.R. (1984) The Principles of Nonlinear Optics (Wiley, New York). 19. Drever, R.W.P., Hall, J.L. & Kowalski, F.V. (1983) Appl. Phys. B 31, 97–105. 20. Demtröder, W. (1981) Laser Spectroscopy, Basic Concepts, and Instrumentation (Springer, Berlin). 21. Cardinale, G.F. & Tustison, R.W. (1992) Thin Solid Films 207, 126–130. 22. Ekstrom, C.R., Keith, D.W. & Pritchard, D.E. (1992) Appl Phys. B 54, 369–374. 23. Beijerinck, H.C.W., van Gerwen, R.J.F., Kerstel, E.R.T., Martens, J.F.M., van Vliembergen, E.J.M., Smits, M.R.Th. & Kaashoek, G.H. (1985) Chem. Phys. 96, 153–173.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DISTRIBUTED RESPONSE ANALYSIS OF CONDUCTIVE BEHAVIOR IN SINGLE MOLECULES
6514
Colloquium Distributed response analysis of conductive behavior in single molecules Marc in het Panhuis*†, Robert W.Munn‡, Paul L.A.Popelier‡, Jonathan N.Coleman†, Brian Foley§, and Werner J.Blau† Ireland Polymer Research Centre, Department of Physics, and §Department of Electronic and Electrical Engineering, Trinity College, Dublin 2, Ireland; and ‡Department of Chemistry, UMIST, Manchester M60 1QD, United Kingdom Edited by Arthur C.Gossard, University of California, Santa Barbara, CA, and approved March 11, 2002 (received for review September 27, 2001) The ab initio computational approach of distributed response analysis is used to quantify how electrons move across conjugated molecules in an electric field, in analogy to conduction. The method promises to be valuable for characterizing the conductive behavior of single molecules in electronic devices. If Moore's law (every 12–24 months the number of transistors on a silicon chip doubles) is to continue to apply until the year 2020, transistor features will have to shrink to molecular scale. This small size will lead to problems that spell the end of Moore's law in classical system architecture. These problems are related to the way dopants are distributed in each device. As transistor size decreases, dopants may aggregate, or statistical deviations in dopant density from device to device may become important. Equally serious is that very small transistors must have very small insulator gates. As dimensions decrease, quantum mechanical tunneling across the gate becomes important. At very small scales, this tunneling acts to decrease device efficiency, presenting a significant limitation: as processing power increases, computationally intensive fields such as virtual reality, complex image recognition, nanorobotics, and real-time holography develop and demand increases in step. In recent years, this need for new transistor architecture has stimulated the emerging field of molecular-scale electronics (1–10). It has been demonstrated that these impediments can be overcome by using a nonclassical device architecture that does not rely on doping or inversion layer-conduction channel formation. Thorough overviews of the concepts, prospects, and expected impact of molecular electronic devices have been given in the literature (7, 11–14). The work of Tour, Reed, and colleagues (5) on two-terminal self-assembled monolayer (SAM) devices has advanced the technology of molecular electronic devices. Their nanoscale device uses charge flow in the conjugated molecule 2′-amino-4-ethynylphenyl-4′ethynylphenyl-5′-nitro-1-benzenethiol, which has polar functional groups that can be used to switch the device. Applying a voltage to the gate electrode sets up an electric field, to which the polar groups respond by changing their orientation, breaking the effective conjugation between adjacent carbon atoms and hence limiting current flow, corresponding to switching from the ON to the OFF state (1, 3–5, 15, 16). Currentvoltage measurements at 60 K showed an ON-OFF peak-to-valley ratio of 1,030:1 (5). Three-terminal molecular devices, such as the SAM organic field effect transistor (SAMFET) as reported by Schön et al. (9, 10), in contrast to two-terminal devices have the ability to modulate the conductance and achieve gain in logic circuits. A schematic of the SAMFET device using the molecule 4,4′-biphenyldithiol (BPDT) as reported in refs. 9 and 10 is shown in Fig. 1, with a SAM connected to source and drain electrodes. It is reported that the drain current can be modulated by ≈5 orders of magnitude by an applied gate voltage. The gate voltage affects only the molecules close to the gate (ON current), whereas the OFF current samples all of the molecules in the SAM. This switching results in a conductance change of ≈7 orders of magnitude at room temperature. Schön et al. estimated a conductance of ≈5 µS per molecule. In the present paper, we show that the recently developed ab initio computational approach of distributed response analysis (17) can be used to quantify the conduction behavior of single molecules. In principle, this technique can be used to identify superior active molecules for electronic devices. We are currently exploring how well the conductive behavior calculated in this paper predicts experimental results on molecular conductance like those reported in refs. 9 and 10. Distributed Response Analysis. Since its earliest days, molecular electronics has been concerned with electrical conduction, rectification, and switching in single molecules (18). Much progress has now been made in studying these processes both experimentally and theoretically, and it is apparent that for a single molecule, the perturbation caused by the electrical contacts is significant (19–21). Nevertheless, it is desirable to explore means of characterizing separately the propensity of molecules to conduct or switch. The distributed polarizability (22) quantifies (among other things) the tendency for charge to flow between different regions of a molecule, which is analogous to conduction, whereas the distributed hyperpolarizability (23) quantifies how the distributed polarizability depends on electric field, which is analogous to switching. Distributed response can be calculated rigorously for both linear (17, 24) and nonlinear (23) coefficients. Hence, although they characterize charge flow within rather than through molecules, distributed linear polarizability and quadratic hyperpolarizability appear suitable means to assess molecules for use as conductors and switches. Distributed molecular response is calculated by using ab initio techniques described in detail in ref. 17. The distributed polarizability relate the change in electron density in an atomic region A to the electrical potential in atomic region B. This approach allows components us to calculate various atomic and molecular properties such as dipole moment and (hyper)polarizability. †Materials
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001 at the National Academy of Sciences in Washington, DC. This paper was submitted directly (Track II) to the PNAS office. Abbreviations: SAM, self-assembled monolayer; SAMFET, SAM organic field effect transistor; pNA, para-nitroaniline; mNA, meta-nitroaniline; FB, fluorobenzene; CN, conduction number; BPDT, 4,4′-biphenyldithiol. *To whom reprint requests should be addressed. E-mail: [email protected].
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DISTRIBUTED RESPONSE ANALYSIS OF CONDUCTIVE BEHAVIOR IN SINGLE MOLECULES
6515
In addition, charge flow into an atomic region can be calculated rigorously.
Fig. 1. Schematic representation of SAMFET, according to Schön et al. (9). The self-assembled monolayer consists of BPDT molecules.
The distributed multipole polarizability components are calculated according to ref. 17. [1]
where is a multipole moment operator, i and j denote occupied molecular orbitals, and σ and τ are virtual molecular orbitals obtained are transition multipole moment matrix elements in an atomic basis, through closed-shell Hartree-Fock calculations. The quantities as defined in the atoms in molecules context (25, 26) with rA as origin. The matrix G is defined in terms of two-electron integrals, [2]
where εσ and εi are orbital energies. Charge flow into an atomic region A is calculated from the distributed polarizability components according to ref. 17, [3]
where i=q, x, y, z. Thus x is the electrostatic potential at the origin of an atomic basin B and component of the electric field at the origin of atomic basin B.
is the x
[4]
RESULTS We define the charge flow number for atomic region A as per unit field magnitude V′, both in a.u.. Fig. 2 shows the calculated values for para-nitroaniline (pNA) across the symmetry axis, for meta-nitroaniline (mNA) across the equivalent axis from the center of the ring to the amino, and for fluorobenzene (FB) across the symmetry axis. We further define the total conduction number (CN) as the sum of the charge flows on one side of the molecular axis perpendicular to the field direction, taking advantage of the molecular symmetry in pNA and FB, which fixes a nodal plane across which the charge flow numbers change sign. For mNA, with no symmetry axis, we calculate CN similarly as the sum of the positive charge flow numbers, allowing the molecular response itself to determine the surface across which the charge flow numbers change sign; in this case, the charge flow is not rigorously parallel to the field. For pNA, we find CN=11.7, whereas mNA and FB yield 11.1 and 8.3, respectively. Now the distributed polarizability components yield the usual molecular polarizability according to refs. 17 and 22,
where α, β=x, y, z, and hence CN and the average polarizability Trα share a common origin. The calculated values of CN and the average polarizability are collected in Table 1, which reveals for the three molecules investigated the new empirical result that 1 unit of CN is approximately equivalent to 1/6 a.u. of polarizability. The ability of an electron to move across a molecule in the direction of an external field is analogous to conduction. Current flow refers to flow of positive charge, so that current is opposite in direction to electron flow. Our method allows us to study the ease of conduction as a function of the direction of the external field; in particular, charge flow is not permitted if the external field is perpendicular to the molecular plane. Distributed response analysis yields atomic polarizabilities (17) and atomic charge flow numbers. These can be combined to calculate polarizabilities and charge flow numbers for
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DISTRIBUTED RESPONSE ANALYSIS OF CONDUCTIVE BEHAVIOR IN SINGLE MOLECULES
6516
functional groups, such as the donor, π-bridge, and acceptor parts of pNA and mNA, but Table 2 shows that these functional group properties are not related in the same way as their whole-molecule counterparts. The complete distributed response analysis of pNA, mNA, and FB is described elsewhere (M.i.h.P., S. O'Flaherty, P.L.A.P., R.W.M. and W.J.B., unpublished work).
Fig. 2. Charge flow numbers for (A) pNA, (B) FB, and (C) mNA. The numbers are dimensionless representations of the reorganization of charge per unit field among the atomic regions. Table 1. Calculated average polarizability αav=1/3Trα and conduction numbers CN for pNA, mNA, and FB CN αav/a.u. Molecule pNA 11.7 70.5 mNA 11.1 68.8 FB 8.3 48.7
CN/αav 0.166 0.161 0.170
The new result that 1 CN is equivalent to 1/6 a.u. of polarizability can be used to predict the CNs from the polarizability for molecules for which the conductance has been measured experimentally to explore the relationship between CN and conductance. As already noted, the experimental conductance has been estimated for BPDT (9). We therefore calculated the average molecular polarizability for four molecules related to BPDT (see Table 3), using the same level of ab initio calculation as for pNA, mNA, and FB. Benzenedithiol has recently been studied in detail between nanoscale metal contacts (27). The thiol groups (used to connect to source and drain) increase the polarizability along the molecular axis, and we predict that the dithiols each have a CN over 40% larger than the parent molecules benzene and biphenyl, with the CN for BPDT reaching 25.6. We observe that introducing side groups such as amino or nitro and increasing the number of phenyl rings both increase the CN. On the basis of these observations, we predict that SAMFETs based on 2′-amino-4-ethynylphenyl-4-ethynylphenyl-5′-nitro-1-benzenethiol [as used by Tour, Reed, and colleagues (5)] should have a much higher CN than those based on BPDT [as used by Schön et al. (refs. 9 and 10)]. The best materials known today to serve as active molecules are carbon nanotubes. The static polarizability of single walled carbon nanotubes has been estimated around 800 a.u. (28), which would lead us to predict a CN of approximately 150, far superior to any other conjugated molecule. If conductance were proportional to CN, then from the BPDT conductance of 5 µS, we would predict a conductance of some 30 µS for single walled carbon nanotubes, compared with the quantum of conductance e2/h of about 77 µS. The conductance for multi-walled carbon nanotubes (with only the outer shell contributing to conduction) has been measured as 77 µS (1 unit of quantum conductance) (29). However, these predictions remain to be validated by making a connection between calculated CNs and experimentally measured conductance. CONCLUSION In conclusion, we have shown that the ab initio computational technique of distributed response analysis can, in principle, be used to quantify the conduction behavior of a single molecule acting as a switching region in a nanoscale field effect transistor design. In an analogy to conduction, it is shown that once an electron is introduced on one side of the molecule, it has the ability to move across the molecule. The calculations demonstrate in detail how charge flow occurs in pNA, mNA, and FB. On the basis of our calculations, we predict that pNA has a more efficient conductive behavior than mNA and FB. We Table 2. Calculated average polarizability αav=1/3Trα and conduction numbers CN for the functional groups in pNA Functional group CN αav/a.u. Acceptor (−NO2) 0.64 16.8 7.46 33.9 Bridge (−C6H4−) 3.60 19.8 Donor (−NH2)
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DISTRIBUTED RESPONSE ANALYSIS OF CONDUCTIVE BEHAVIOR IN SINGLE MOLECULES
6517
established the empirical result that 1 unit of molecular CN is approximately equal to 1/6 a.u. of (static) molecular polarizability. This result has been used to estimate the conductive behavior of similar conjugated molecules by using the average molecular polarizability. It was predicted that BPDT has a CN of 25.6 and therefore should have the most effective conduction behavior of the molecules investigated (see Tables 1 and 3). Table 3. Calculated average polarizability αav=1/3Trα and predicted CN for benzene, 1,4-benzenedithiol (BDT), biphenyl, and BPDT αav/a.u. Predicted CN Molecule Benzene 49.8 8.3 BDT 86.6 14.5 Biphenyl 108.3 18.1 BPDT 155.1 25.6
We are currently investigating how accurately our CN predicts the results of experiments on molecular conductance such as those reported by Schön et al. (9, 10). We envisage that distributed response analysis could then be used to identify superior molecular switches from molecular CN before the molecules are experimentally investigated. We gratefully acknowledge R.H.Baughman for stimulating discussions. M.i.h.P., J.N.C., and W.J.B. thank the MultiMedia Research Program of the Irish Higher Education Authority and the European Community Research Training Network Coupled Mechanical and Electrical Properties of Carbon Nanotube Based Systems (COMELCAN) for assistance. 1. Tans, S.J., Verschueren, A.R.M. & Dekker, C. (1998) Nature (London) 393, 49–52. 2. Reed, M.A. & Tour, J.M. (2000). Sci. Am. 286, 86–93. 3. Reed, M.A. (2001) MRS Bull. 26, 113–120. 4. Tour, J.M. (2000) Acc. Chem. Res. 33, 791–804. 5. Chen, J., Reed, M.A., Rawlett, A.M. & Tour, J.M. (1999) Science 286, 1550–1552. 6. Chen, J., Wang, W., Reed, M.A., Rawlett, A.M., Price, D.W. & Tour, J.M. (2000) Appl. Phys. Lett. 77, 1224–1226. 7. Joachim, C., Gimzewski, J.K. & Aviram, A. (2000) Nature (London) 408, 541–548. 8. Rueckes, T., Kim, K., Joselevich, E., Tseng, G.Y., Cheung, C.-L. & Lieber, C.M. (2000) Science 289, 94–97. 9. Schön, J.H., Meng, H. & Bao, Z. (2001) Nature (London) 413, 713–716. 10. Schön, J.H., Meng, H. & Bao, Z. (2001) Science 294, 2138–2140. 11. Goldhaber-Gordon, D., Montemerlo, M.S., Love, J.C., Optipeck, G.J. & Ellenbogen, J.C. (1997) Proc. IEEE 85, 521–540. 12. Ratner, M.A. & Jortner, J. (1997) in Molecular Electronics, eds Jortner, J. & Ratner, M.A. (Blackwell, London), pp. 5–72. 13. Ellenbogen, J.C. & Love, J.C. (2000) Proc. IEEE 88, 386–426. 14. Wada, Y. (2001). Proc. IEEE 89, 1147–1171. 15. Donhauser, Z.J., Mantooth, B.A., Kelly, K.F., Bumm, L.A., Monnell, J.D., Stapleton, J.J., Price, D.W., Jr., Rawlett, A.M., Allara, D.L., Tour, J.M. & Weiss, P.S. (2001) Science 292, 2303–2307. 16. Seminario, J.M., Zacarias, A.G. & Tour, J.M. (2000) J. Am. Chem. Soc. 122, 3015–3020. 17. in het Panhuis, M., Popelier, P.L.A., Munn, R.W. & Ángyán, J. (2001) J. Chem. Phys. 114, 7951–7961. 18. Aviram, A. & Ratner, M.A. (1974) Chem. Phys. Lett. 29, 277–283. 19. Kergueris, C., Bourgoin, J.P., Palacin, S., Estève, D., Urbina, C, Magoga, M. & Joachim, C. (1999) Phys. Rev. B 59, 12505–12513. 20. Yaliraki, S.N., Roitberg, A.E., Gonzalez, C., Mujica, V. & Ratner, M.A. (1999) J. Chem. Phys. 111, 6997–7002. 21. Hall, L.E., Reimers, J.R., Hush, N.S. & Silverbrook, K. (2000) J. Chem. Phys. 112, 1510–1521. 22. Stone, A.J. (1985) Mol. Phys. 56, 1065–1082. 23. Reis, H., Papadopoulos, M.G., Hättig, C., Ángyán, J. & Munn, J.W. (2000) J. Chem. Phys. 112, 6161–6172. 24. Ángyán, J., Jensen, G., Loos, M., Hättig, C. & Hess, B.A. (1994) Chem. Phys. Lett. 219, 267–273. 25. Bader, R.W.F. (1990) Atoms in Molecules-A Quantum Theory (Oxford Univ. Press, Oxford). 26. Popelier, P.L.A. (2000) Atoms in Molecules-An Introduction (Pearson Education, Harlow, U.K.). 27. Emberly, E.G. & Kirczenow, G. (2001) Phys. Rev. B. 64, 235412–1– 235412–8. 28. Jensen, L., Schmidt, O.H., Mikkelsen, K.V. & Åstrand, P.-O. (2000) J. Phys. Chem. B 104, 10462–10466. 29. Frank, S., Poncharal, P., Wang, Z.L. & de Heer, W.A. (1998) Science 280, 1744–1746.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGN OF PROTEIN STRUTS FOR SELF-ASSEMBLING NANOCONSTRUCTS
8488
Design of protein struts for self-assembling nanoconstructs
Paul Hyman*†‡, Regina Valluzzi§, and Edward Goldberg† *NanoFrames LLC, Boston MA; †Department of Molecular Biology and Microbiology, Tufts University School of Medicine, Boston MA; and §Department of Chemical Engineering, Tufts University, Medford MA Edited by Jack Halpern, University of Chicago, Chicago, IL, and approved April 29, 2002 (received for review October 12, 2001) Bacteriophage T4 tail fibers have a quaternary structure of bent rigid rods, 3×160 nm in size. The four proteins which make up these organelles are able to self-assemble in an essentially irreversible manner. To use the self-assembly domains of these proteins as elements in construction of mesoscale structures, we must be able to rearrange these domains without affecting the self-assembly properties and add internal binding sites for other functional elements. Here we present results on several alterations of the P37 component of the T4 tail fiber that change its length and add novel protein sequences into the protein. One of these sequences is an antibody binding site that is used to inactivate phage carrying the modified gene. To realize the potential for nanotechnology, methods for practical directed assembly of mesoscale structures must be developed. We use a biological paradigm to develop the science and engineering needed to implement a practical bottom-up manufacturing system. Living cells normally assemble mesoscale structures (e.g., muscle fibers, mitotic spindles, flagella, virus particles) following well-studied mechanisms, including vectorial assembly and specific interaction moieties. Our approach is to create a set of nanoscale subunits of precise size, shape, and functionality that can be assembled in a massively parallel manner. Our subunits are based on the tail fiber proteins of bacteriophage T4. These proteins make up a self-assembling, precisely defined, highly stable structure (1, 2) and, as we show below, are readily amenable to reengineering without losing these properties. Bacteriophage (phage) T4 is one of the archetypical members of the family Myoviridae or T-even phage. These viruses are characterized by a large, elongated icosohedral head (which contains the phage DNA), a contractile tail (to stabilize the phage perpendicular to the cell and penetrate the cell wall), and tail fibers (which contain the phage receptors and trigger infection) (3, 4) (Fig. 1A). The tail fiber proteins have an unusual quaternary structure of long, thin (3 nm×160 nm), rigid rods (5). Their function is to transduce chemical recognition of the Escherichia coli host into a mechanical force on the phage base plate, essentially acting as a set of cooperative levers. This mechanical stress triggers a series of protein conformational changes that lead to entry of the phage DNA into the cell (6, 7). The three main tail fiber proteins, P34, P36 and P37¶, are thought to be principally composed of dimeric∥ antiparallel β-sheets (8). Gp35, which forms the angle in the tail fiber, probably has a more complex structure. The joints between the homodimeric segments are also likely to have a more complex structure but there is no evidence that the central rod regions have any tertiary structure (i.e., interactions between distant amino acid residues) at all (5). The extended antiparallel β-sheet secondary structure should directly support the rigid rod quaternary structure. If correct, we thought that deletions or additions to the central rod regions which maintain the β-sheet structure should permit alteration in tail fiber length without greatly affecting overall structural integrity. Furthermore the binding domains at the ends of the proteins should form separate functional domains from the central, rigid rod domain. Finally the β-sheet structure should contain turns and loops that can be expanded with functional peptides without disrupting the quaternary structure. In this paper we describe modifications of P37 that support some of these hypotheses. MATERIALS AND METHODS E. coli and Phage Strains and Reversion Assay. T4 37amA481 (11) was the mutant used to derive all phage strains discussed in this paper. E. coli B40 (suI) (lab strain, courtesy of P.Strigini, Harvard Medical School, Cambridge, MA) was used to grow and titer phage containing an amber mutation, and E. coli BB (su0) (12) was used for all non-amber phages. T4 37amA481 pseudorevertants were identified by their ability to form plaques on BB, and stocks were prepared by standard techniques (13). Plasmids were produced, and recombined with phage using E. coli MC1061 (F− araD139 ∆(ara-leu)7696 galE15 galK16 ∆(lac)X74 rpsL (Strr) hsdR2 (rK− mK+) mcrA mcrB1) (14) as the host strain. PCR Primers and Product Cloning. Primers cysF (CTATTAACGGACTTTTGAGA) and cysR (TTCAATACGTCCAATAGTTT) amplify the central rod region of phage T4 gene 37 including the location of the S∆1 deletion and we used them to screen pseudorevertant phage as well as for sequencing. These primers amplify a 1.4-kb fragment from wild-type T4 DNA but only a 0.36-kb product from T4 37S∆1 DNA. Primers recF (GACGAGCTCCTTCGGGTTCCCTTTTTCTTTA) and 37B-2R (TTGGGTAACTCGACATGA) amplify a 3.2-kb segment of the tail fiber gene cluster including the 3′ end of gene 35, gene 36, and the first two-thirds of gene 37. When these primers are used to amplify T4 37S∆1, a 2.1-kb fragment is produced in which the deletion junction is approximately in the middle. We cloned this 2.1-kb PCR product into pGEM-T (Promega) for sequencing, further modification (see below), and to transfer modified genes into T4 phage by recombination between the plasmid and infecting phage. The construct containing this 2.1-kb insert was designated p37S∆1. Recombination of Phage and Plasmid. We transferred modified genes into phage by infecting plasmid bearing cells with T4 37amA481 (whose amber mutation is located in the segment of
This paper results from the Arthur M.Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC. This paper was submitted directly (Track II) to the PNAS office. ‡To whom reprint requests should be addressed. E-mail: [email protected] ¶GpX (gene product) refers to the monomeric product of gene X, whereas PX refers to the matured, multimeric complex of gpX's that has assembled into the structure that is found in the phage T4 virion. ∥There is conflicting evidence as to whether the gene 34, 36, and 37 segments of the tail fibers are homodimers (8, 9) or homotrimers (10). For the purposes of this work either case is possible because the monomers are arranged in an overall parallel fashion (i.e., N termini together at one end of the mature protein and C termini at the other end). This means that insertions, deletions, or modifications in the protein monomers will be located at identical positions in the mature dimer/trimer. For simplicity, we will refer to the proteins as forming dimers.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGN OF PROTEIN STRUTS FOR SELF-ASSEMBLING NANOCONSTRUCTS
8489
DNA that is missing in T4 37S∆1 and its derivatives) and growing the phage to produce a stock. Because MC1061 is not an amber-suppressing strain, only cells where recombination between the plasmid and phage genome occurred would produce viable pseudorevertant phage. We selected recombinant phage from the lysates by plating on BB (su0) and screened plaques by PCR to identify which plaques contained the 37S∆1 deletion.
Fig. 1. Phage images. (A) Structure of bacteriophage T4. (B) Electron micrograph of T4 37S∆1 phage. The white line next to the distal portion of the tail fiber indicates the length of a wild-type distal tail fiber at the same magnification. (C) Electron micrograph of T4 37S∆1 treated with mAb and secondary antiserum.
Measuring Adsorption Rates. Adsorption rates were measured by using a single time point method (15). Briefly, phage were incubated with log phase cells for a fixed time, usually 5 or 10 min at 37°C (within the phage eclipse period). At that time we diluted the phage/cell mixture into buffer saturated with chloroform to lyse the cells. The number of infectious phage remaining is determined and the adsorption constant is calculated as Kads= (2.3/Ct)log(Po/Pt), where C is the cell concentration (ml−1), t is the incubation time (in minutes), P0 is the infectious phage concentration (ml−1) at time 0, and Pt is the infectious phage concentration (ml−1) at time t. Construction of S∆1G5, S∆1UCS, S∆1ras1, and S∆1ras2. The pentaglycine coding segment in S∆1G5 was added to the cloned DNA in p37S∆1/T by using overlapping PCR primers (16). Primers 37S≥1–1F (GGCGATGGTGGCGGTGGCGGCAATGTACAATTTTACGCTG) and 37S∆1–1R (TACATTGCCGCCACCGCCACCATCGCCATTTAATCTCAA) contain complementary sequences corresponding to the Gly-5-containing S∆1 junction. They were used with the flanking recF and 37B-2R primers to produce two modified half segments that were then recombined using the complementary ends to fuse the two segments and the flanking primers to amplify the whole segment. The entire segment was then cloned into pGEM-T. S∆1UCS (universal cloning site) was creating by amplifying half segments of the S∆1 clone with primers 37S∆1–2F (GGCGATGAGACGGTACCGTCTCAATGTACAATTTTACGCTG) and 37S∆1–2R (TACATTGAGACGGTACCGTCTCATCGCCATTTAATCTCAA). Each primer contains a BsmBI and KpnI site. The two half segments were joined using the KpnI site to create a single segment with two BsmBI sites around the central KpnI site inserted into the S∆1 junction. BsmBI cuts at positions 7/11 outside of the recognition site and the two BsmBI sites in p37S∆1UCS/T are arranged so that the two cuts drop out the center segment (containing both BsmBI sites) leaving the original construct sequence with two different cohesive ends. This arrangement allows for the insertion (with an unambiguous orientation) of any double-stranded oligonucleotide with the correct cohesive ends. Thus, any oligopeptide can be cloned into junction of the S∆1 deletion. We inserted the ras1 control (nonepitopic for Y13–259; see below) sequence by combining the oligonucleotides S∆1R-1F (GCGATGGTGGCGGTGGCGCCCGCGGCGTGGGAAAGAGTGCCCTGACCATCCAGCTGATCGGTGGCGGTGGCA) and S∆1R-1R (ACATTGCCACCGCCACCGATCAGCTGGATGGTCAGGGCACTCTTTCCCACGCCGCGGGCGCCACCGCCACCA). Similarly, the ras2 mAb epitope coding sequence was inserted by using the oligonucleotides S∆1R-2F (GCGATGGTGGCGGTGGCGAAGAATACTCCGCAATGCGCGACCAGTACATGCGCACCGGTGAAGGTGGCGGTGGCA) and S∆1R-2R (ACATTGCCACCGCCACCTTCACCGGTGCGCATGTACTGGTCGCGCATTGCGGAGTATTCTTCGCCACCGCCACCA). To anneal each oligonucleotide pair, we mixed the appropriate oligonucleotides in equimolar amounts, boiled the mixtures briefly, and cooled the mixtures slowly to form the appropriate double-stranded oligonucleotides with the correct single-stranded extensions. These oligonucleotides were ligated directly into BsmBI-digested p37S∆1UCS/T. The insertions were confirmed by sequencing with the cysF primer. mAb Inactivation Experiments. We purchased mAb Ab-1 (Y13– 259; ref. 17) and inactivating peptide from Calbiochem and rabbit antirat whole IgG serum from Sigma. The mAb and peptide were resuspended in Dulbecco's PBS and the anti-serum was used as supplied. For inactivation experiments, we diluted phage to 1010 cells/ml in 10 mM phosphate pH 7.4/10 mM MgSO4. We added mAb (from a 0.1 mg/ml stock) to 500 µl of diluted phage and incubated the mixture for 30 min (unless otherwise indicated) at room temperature on a rotisserie mixer. Then we added 4 µg of secondary antiserum (from a 2 mg/ml stock) and incubated for 30 min at room temperature. For the initial experiments shown in Fig. 2A we used 1 µg of mAb, whereas 3 µg or the indicated amount was used for the remaining experiments. For the free epitope inhibition experiment shown in Fig. 2D, we mixed the peptide (EEYSAMRDQVMRTGE) and mAb at a 10:1 molar ratio and incubated for 30 min at room temperature. The mAb/peptide mixture was then added to phage as described above. Electron Microscopy of Phage. Phage and phage/antibody complexes were stained with 1% phosphotungstate (pH 7) on carbon grids. Grids were examined at 100 kV by using a Philips CM10 transmission electron microscope. The final micrograph images were at a magnification of ×73,000. RESULTS AND DISCUSSION Identification and Characterization of a Large Deletion in P37. The tail fiber acts as a trigger to signal initiation of the tail sheath contraction process that precedes phage DNA injection. The
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGN OF PROTEIN STRUTS FOR SELF-ASSEMBLING NANOCONSTRUCTS
8490
reversible, noncovalent binding of a number of tail fiber distal ends to their specific receptor sites on the cell surface leads to a cooperative mechanical stress in the base plate. This stress triggers base plate expansion and initiates the tail sheath contraction, which extends the tail core through the cell wall (18, 19). The tail fibers' critical function for phage viability provides a sensitive assay for rigidity in tail fiber structure because any major loss of rigidity in the structure should impair the tail fibers' triggering function.
Fig. 2. Inactivation of phage by a monoclonal antibody. (A) Treatment of phage with mAb and secondary anti-serum. Each phage type was treated with 1 µg of mAb as described in Materials and Methods. (B) Time course of mAb treatment. S∆1 ras2 phage were treated with 3 µg of mAb for the indicated time before a 30 min incubation with secondary anti-serum. (C) Dose-response study of S∆1ras2 phage with varying amounts of mAb. (D) Effect of treating mAb with free epitope before inactivation of S∆1ras2.
We used PCR analysis to screen spontaneous pseudorevertants of a gene 37 amber mutation (amA481), and identified a phage that appeared to have approximately 1 kb of DNA deleted from the middle of gene 37. This gene codes for the protein forming the distal end of the tail fibers, and its C terminus forms the phage receptor. Sequence analysis confirmed that a single contiguous segment of DNA coding for 346 of 1,026 amino acid residues (34%) was deleted in this phage, which was designated S∆1 (spontaneous deletion 1). Table 1 shows the protein sequences of the deletion junctions and the corresponding wild-type protein. The deleted region begins at amino acid 73, which is 23 residues downstream from the conserved N-terminal domain of P37. This conserved region is thought to form the stiff butt end joint with the P36 Cterminal conserved domain (20). Thus, this deletion falls completely within the P37 rod-like region. Phage carrying the S∆1 mutation produce plaques of normal size and appearance indicating that they are able to infect Table 1. Partial protein sequences of naturally occurring and engineered gene 37 proteins Phage Partial protein sequence of gene 37 at S∆1 junction Wild-type T4 GLLRLNGDYVQ//GSNNVQFYADG 37 S∆1 GLLRLNGD|NVQFYADG GLLRLNGDGGGGGNVQFYADG 37 S∆1G5 37 S∆1ras1 (control) GLLRLNGDGGGGARGVGKSALTIQLIGGGGNVQFYADG GLLRLNGDGGGGEEYSAMRDQYMRTGEGGGGNVQFYADG 37 S∆1ras2 (mAb epitope) Sequences flanking the S∆1 junction are in italics, double slash represents 340 deleted amino acid residues, vertical line marks the position of the junction, inserted sequences are in boldface.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGN OF PROTEIN STRUTS FOR SELF-ASSEMBLING NANOCONSTRUCTS
8491
and grow normally. We also measured the adsorption rate of the S∆1 phage (a measure of the rate of irreversible binding to the cell surface) and found that it was the same as wild-type phage (9.2 vs. 9.5×10–10 ml/min; S∆1/wild-type=0.97). It was possible that the deletion mutation is compensated for by a second (duplication/insertion) mutation so that the overall tail fiber length was unchanged. To test this possibility, we cloned a 2-kb segment of DNA from the S∆1 phage that surrounds the deletion site and placed it in a nonexpressing plasmid. Restriction and sequence analysis confirmed that this clone contained the expected DNA segments surrounding the 1,038 bp S∆1 deletion and no additional DNA sequences. Homologous recombination was used to transfer the S∆1 deletion into T4 phage containing the A481 amber mutation (which is located in the segment corresponding to the S∆1 deletion segment). The S∆1 deletion transferred at high efficiency, indicating that there is no other suppressor mutation needed to produce a viable phage. We examined phage carrying the S∆1 mutation by electron microscopy. The shortened distal portion of the tail fiber is clearly visible in the electron micrograph of S∆1 phage (Fig. 1B). To compare wild-type tail fiber to S∆1 tail fiber we calculated the ratio of the lengths of the distal half fiber/proximal half fiber (D/P) by using measurements from enlarged electron micrographs. We found that for wild-type fibers D/ P=0.99±0.06 (n=11) and for S∆1 fibers D/P=0.54±0.14 (n=6). This finding confirms that the viable S∆1 phage have shortened but otherwise functional tail fibers. Inserting a Peptide Into a β-loop In P37. In the β-sheets forming the central rod regions of the tail fibers, the loop regions contribute little to maintaining the H-bond network, nor to the van der Waals interaction in the hydrophobic layer within the rod (21, 22). We postulate that the loops can be more variable and flexible than other regions of the tail fiber proteins. This postulate suggests that the junction of the S∆1 deletion is in a loop (rather than in a β-strand) of the rod portion of gene 37. Surface loops in proteins can often be expanded to include additional peptide sequences with minimal effects on protein structure, function or stability (23). Thus, if the S∆1 junction is in a loop, we should be able to insert additional sequences into the junction, expanding the loop, without severely disrupting the structural integrity of the tail fiber. To test this we added DNA sequences encoding a pentaglycine peptide into the S∆1 junction (Table 1) in the cloned gene segment. This modified sequence also transferred readily into phage by homologous recombination. [The S∆1G5 phage produced poorer stocks, although the adsorption constant was almost the same as for the wild-type and S∆1 phage (12×10−10 ml/min; S∆1G5/wild type=1.3). Poorer stocks might indicate a mild interference with phage development.] This finding confirms that the S∆1 junction is able to accept peptide insertions without any significant loss of structural integrity and fits our hypothesis that the junction identifies a loop in the β-structure. Inserting and Characterizing an Antibody Epitope into a Tail Fiber Protein. To use tail fiber derived proteins as mesoscale assembly units, we will also need to attach specific functions to the assembled arrays of structural units. They may be attached before or after maturation of the final structure or at an intermediate step. The attachment may be covalent (e.g., disulfide bridges) or noncovalent (e.g., his tags). Incorporation of a peptide epitope may also be used to attach a functionality linked to the appropriate antibody. Fusions between antibodies and functional peptides have been extensively developed (24, 25). In the case of our nanoarchitectures, the compound would be fused to a mAb that is specific for an epitope in the structural unit. Thus, we set out to show that an antibody epitope could be incorporated into a tail fiber protein. Next, we inserted two different 15 aa sequences from the human H-ras gene into the putative loop at the S∆1 fusion junction (Table 1). Both peptides were flanked by four glycines on each side. One construct, S∆1ras1, containing a nonepitope segment of H-ras, was created as a control, whereas the other, S∆1ras2, contains the epitope specifically recognized by the rat monoclonal IgG antibody Y13–259 (17). Each of these modified genes readily transferred into phage by homologous recombination. To test whether the epitope was accessible for interactions with the exogenous antibody, we treated S∆1, S∆1ras1, and S∆1ras2 phage with the anti-ras mAb. If the mAb can bind to the H-ras epitope, it might inactivate the phage by linking together tail fibers on a single phage, thereby preventing proper binding to the cell surface. Alternatively, several phage might be linked together to form large noninfectious complexes. However, as Fig. 2A shows, mAb treatment alone (gray bars) did not result in phage inactivation. When the phage/mAb mixtures were further treated with an anti-rat IgG serum (striped bars) (which binds to the Fc region of the mAb), 85% of the S∆1ras2 phage were inactivated. Because the S∆1ras1 control phage were unaffected and because the anti-rat IgG antiserum alone has no effect on the S∆1ras2 phage (data not shown), this finding demonstrates that the ras2 epitope is exposed on the surface of the tail fiber and accessible to the mAb. The requirement for the secondary antibody for phage inactivation may reflect the axial symmetry of P37. Because each mature fiber contains more than one epitope in close proximity on each tail fiber, it is likely that both binding sites in the mAb become bound to a single fiber. This would not be expected to inactivate the phage. Hence, the need for the secondary antibodies to crosslink tail fibers by binding to two mAbs bound to two different fibers and inactivate the phage. Regardless of the specific mechanism of inactivation, these experiments show that a functional peptide can be added to the rod region of a tail fiber protein without disrupting the tail fiber structure or function. We further investigated the interaction of the S∆1ras2 phage with the mAb. Fig. 2B shows that inactivation depends on the time allowed for mAb binding before addition of the secondary antiserum, reaching a maximum of 99.9% by 120 min. Fig. 2C shows that inactivation also has a simple dose-response relationship with the amount of Y13–259 mAb used. Fig. 2D shows that the S∆1ras2 phage could be protected from inactivation by pretreating the mAb with a free 15-aa peptide of the same sequence as the 15-residue epitope inserted into the tail fiber protein. Although 99.8% of the phage were inactivated in the control treatment (with buffer only), there was no significant inactivation when the mAb was pretreated with the peptide. This finding demonstrates that the inactivation requires a specific interaction of the antibody with its specific epitope sequence. We examined how the mAb interacts with the tail fiber by imaging mAb-treated phage. Fig. 1C shows a typical cluster of inactivated phage. The phage form a “bouquet” with the tail fibers linked together. It is unlikely that phage in such a bouquet could orient properly on the cell surface to allow the tail fibers to function cooperatively and trigger infection. Taken together, these results demonstrate that rearrangements, fusions, and insertions can be made to a tail fiber protein without disrupting the functional integrity of the mature protein structure. They also support our hypothesis that fusion sites can be used for insertion of foreign peptides in such a way that they are available for binding. Further, these results support our hypothesis that the binding domain at the Nterminal end of P37 (and, presumably, the binding domains of other tail fiber proteins) is functionally separable from the central rod region. This finding suggests that chimeric proteins composed of the P37-
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGN OF PROTEIN STRUTS FOR SELF-ASSEMBLING NANOCONSTRUCTS
8492
binding domain of P36 joined by a central rod domain to the P36-binding end of P37 will form homo-polymeric fibers as shown in Fig. 3. The fusion site of these chimeric proteins should accept a functional peptide just as the S∆1ras2 junction does. This will provide the potential for attaching immunoconjugated functional moieties at precise locations along the fiber. The length of the chimeric proteins can be adjusted by using more or less of the rod region from either of the parent proteins, allowing the spacing of the functional moieties to be controlled. Furthermore, other β-loops within the central rod domain can be used as insertion sites for the addition of antigenic peptides that can subsequently be recognized by antibodies to add either functional or structural capabilities including crosslinking of the polymeric fibers into open two- and three-dimensional arrays.
Fig. 3. Self-assembly of fibers from tail fiber derived chimeric proteins. (A) Protein domains used to create a chimeric protein subunit (matured protein complexes shown). (B) Self-assembly of chimeric proteins onto a P37 initiator. The initiator is required to allow the gp36 monomer domains to combine to form the mature P36 domain. As in the wild-type fiber, this assembly propagates along the axis of the central rod domain. In the case of the chimeric proteins, this results in the formation of another P37 domain, supporting the next round of polymerization.
This approach will let us engineer protein fibers to place functional moieties in predesigned positions relative to one another to construct nanocomponents and nanodevices that exhibit functions not attainable with single nanoparticles or nanostructured materials. Taken together, the capability demonstrated here provides great potential for fabrication of a broad range of nanostructures that will exhibit anticipated, and in many cases, still unanticipated functions. Note Added in Proof. In collaboration with Dr. W.Stafford (Boston Biomedical Research Institute), we have recently used analytical ultra-centrifugation to confirm the trimeric nature of the gene 34, 36, and 37 tail fiber segments as first proposed by the lab of A.Steven (10) (unpublished data). We thank Fred Eiserling for the drawing of bacteriophage T4, Cathy Linsenmayer and Melissa McCoy for technical assistance, and Debu RayChaudhuri for criticism and encouragement. E.G. thanks Pat Durham for his excellent technical assistance at the start of this project. We especially thank Phil Harriman for his support of this work in its nascent phase. This work was supported in part by National Science Foundation Grant DBI 9834603 and Office of Naval Research Grant N00014–98–1–0784 (to E.G.). 1. Wood, W.B., Eiserling, F.A. & Crowther, R.A. (1994) in Molecular Biology of Bacteriophage T4, ed. Karam, J.D. (Am. Soc. Microbiol. Press, Washington, DC), pp. 282–290. 2. Henning, U. & Hashemolhosseini, S. (1994) in Molecular Biology of Bacteriophage T4, ed. Karam, J.D. (Am. Soc. Microbiol. Press, Washington, DC), pp. 291–298. 3. Wood, W.B. (1979) Harvey Lect. 73, 203–223. 4. Eiserling, F.A. & Black, L.W. (1994) in Molecular Biology of Bacteriophage T4, ed. Karam, J.D. (Am. Soc. Microbiol. Press, Washington, DC), pp. 209–212. 5. Beckendorf, S.K. (1973) J. Mol. Biol. 73, 37–53. 6. Arscott, P.G. & Goldberg, E.G. (1976) Virology 69, 15–22. 7. Crawford, J.T. & Goldberg, E.G. (1980) J. Mol. Biol 139, 679–690. 8. Earnshaw, W.C., Goldberg, E.G. & Crowther, R.A. (1979) J. Mol. Biol. 132, 101–131. 9. Ward, S. & Dickson, R.C. (1971) J. Mol. Biol. 62, 479–492. 10. Cerritelli, M.E., Wall, J.S., Simon, M.N., Conway, J.F. & Steven, A.C. (1996) J. Mol. Biol. 260, 767–780. 11. Fisher, K.M. & Bernstein, H. (1970) Molec. Gen. Genet. 106, 139–150. 12. McFall, E. & Stent, G.W. (1958) J. Gen. Microbiol. 18, 346–363.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGN OF PROTEIN STRUTS FOR SELF-ASSEMBLING NANOCONSTRUCTS
8493
13. Carlson, K. & Miller, E.S. (1994) in Molecular Biology of Bacteriophage T4, ed. Karam, J.D. (Am. Soc. Microbiol. Press, Washington, DC), pp. 421–441. 14. Casadaban, M.J. & Cohen, S.N. (1980) J. Mol. Biol. 138, 179–207. 15. Adams, M.H. (1959) Bacteriophages (Interscience, New York). 16. Sambrook, J. & Russel, D.W. (2001) Molecular Cloning (Cold Spring Harbor Lab. Press, Plainview, NY), 3rd Ed. 17. Sigal, I.S., Gibbs, J.B., D'Alonzo, J.S. & Scolnick, E.M. (1986) Proc. Natl. Acad. Sci. USA 83, 4725–4729. 18. Crawford, J.T. & Goldberg, E.G. (1977) J. Mol. Biol. 111, 305–313. 19. Crowther, R.A. (1980) J. Mol. Biol. 137, 159–174. 20. Riede, I., Drexler, K. & Eschbach, M.-L. (1985) Nucleic Acids Res. 13, 605–616. 21. Branden, C. & Tooze, J. (1999) Introduction to Protein Structure (Garland, New York), 2nd Ed. 22. Xu, G., Wang, W., Groves, J.T. & Hecht, M.H. (2001) Proc. Natl. Acad. Sci. USA 98, 3652–3657. 23. Regan, L. (1999) Curr. Opin. Struct. Biol. 9, 494–499. 24. Vitetta, E.S., Fulton, R.J., May, R.D., Till, M. & Uhr, J.W. (1987) Science 238, 1098–1104. 25. Byers, V.S. & Baldwin, R.W. (1988) Immunology 65, 329–335.
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGN OF PROTEIN STRUTS FOR SELF-ASSEMBLING NANOCONSTRUCTS 8494
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGN OF PROTEIN STRUTS FOR SELF-ASSEMBLING NANOCONSTRUCTS
Arthur M.Sackler COLLOQUIA OF THE NATIONAL ACADEMY OF SCIENCES NANOSCIENCE: UNDERLYING PHYSICAL CONCEPTS AND PHENOMENA May 18–20, 2001 National Academy of Sciences, Washington, DC Organized by John T.Yates, Jr., Phaedon Avouris, and George Whitesides Program Friday, May 18 Robert Greenler, University of Wisconsin, Milwaukee Milwaukee: Sunlight and Ice Crystals in the Skies of Antarctica
Saturday, May 19 Introductory Remarks John T.Yates Jr. Session I. Proximal Probes Moderator, John T.Yates, Jr. Karl H.Rieder, Free University of Berlin Atom and Molecule Manipulation with the STM Wilson Ho, University of California, Irvine Single Molecule Vibrational Spectroscopy and Microscopy Roland Wiesendanger, University of Hamburg Nano-scale Studies of Quantum Phenomena by Scanning Probe Spectroscopy Session II. Optical Techniques Moderator, Wilson Ho Sunney Xie, Harvard University Single Molecule Spectroscopy Hans-Joachim Freund, Fritz-Haber-lnstitut der Max-Planck-Gesellschaft Fluorescence of Deposited Nanoclusters Using the STM Session III. Quantum Dots & Clusters Moderator, Phaedon Avouris Paul Alivisatos, University of California, Berkeley Optical Properties of Nanoclusters Jzi Landman, Georgia Institute of Technology The Non-Scalable Regime: Electronic Structure, Fluid Flow, and Nanocatalysis
8495
About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.
DESIGN OF PROTEIN STRUTS FOR SELF-ASSEMBLING NANOCONSTRUCTS
Louis Brus, Columbia University Single Molecule Raman Spectra on Large Ag Nanocrystals Leo Kouwenhoven, Harvard University Experimental Studies of the Electronic Structure and Interactions in Quantum Dots Mostafa A.El-Sayed, Georgia Institute of Technology Interesting Photophysical and Photothermal Properties of Gold Nanorods Stephen Chou, Princeton University Quantum Dot Memory Devices Ellen Williams, University of Maryland Nanoscale Fluctuations on Solid Surfaces Keynote Presentation Mildred S.Dresselhaus, Massachusetts Institute of Technology Perspectives on Nanoscience Policy Issues
Sunday, May 20 Session III. Quantum Dots & Clusters (continued) Moderator, Phaedon Avouris Catherine Brechignac, CNRS Stability of Assembled Cluster Islands and Fractal Fragmentation Raphael Levine, Hebrew University of Jerusalem Towards Molecular Logic Machines Christopher Murray, IBM Synthesis and Magnetic Properties of Nanoclusters Session IV. Nanowires, Molecular Wires, and Devices Moderator, Charles Lieber Johannes Voit, Universitat Bayreuth One-Dimensional Electrons: From Centimeters to Nanometers Phaedon Avouris, IBM Carbon Nanotubes: Transport and Electronic Devices Mark Ratner, Northwestern University Modeling Molecular Electronic Devices Peter Eklund, Pennsylvania State University Carbon Nanotube Electronics Charles Lieber, Harvard University Nanowire and Nanotube Building Blocks for Nanoscale Science and Technology Mark Reed, Yale University Molecular Electronics James R.Heath, University of California, Los Angeles Molecular Devices and Computing Norton Lang, IBM First-Principles Calculations of the Conductance of Molecules Session V. Nanostructure Synthesis and Self-Assembly Moderator, Mark Ratner George Whitesides, Harvard University Lithography for Rapid Device Fabrication Nadrian Seeman, New York University Nanostructures and Devices Assembled from DNA Angela Belcher, University of Texas Evolving Biomolecular Control of Semiconductor and Magnetic Nanostructures
8496
