This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
i = — (r) =
Figure 23: A schematic picture of (a) undisturbed and (b) disturbed fragment of the double helix. The latter contains one open state.
5.7
DNA denaturation
The nonlinear model for thermal denaturation of the DNA molecule has been recently proposed by the authors of88'89. They assumed that the DNA molecule can be modeled by two discrete chains connected Morse potential Vn (n = 1, 2, 3, ...N) representing the hydrogen bonds (Fig. 24). For each base pair the model includes two degrees of freedom (un and vn) which correspond to the displacements of bases from their equilibrium positions along the direction of hydrogen bonds (Fig. 25). So, the Hamiltonian of the model has the form
with where m is a common mass for bases; k is a coupling constant; D and a are taken equal to 0.33 eV and 1.8 A
, respectively.
130
Figure 24: Two polynucleotide chains connected by hydrogen bonds.
Figure 25: Displacements (un, vn) of the bases in the model.88'89 Hydrogen bonds are shown by dotted lines.
To study denaturation process which includes the strand separation (Fig. 26) it is important to investigate the behaviour of the variables yn
These variables represent the out-of-phase displacements which stretch the hydrogen bonds. The dynamical equation for the variables can be found from Hamiltonian (Eq. 47) in the form
To solve the equation, Peyrard and Bishop restricted themselves to a con-
tinuum approximation and used a multiple-scale expansion which is valid when
131
the precursor phenomena are considered. As a result, they obtained a nonlinear Scrodinger equation whose soliton-like solutions were discussed in Sec. 3.6. This result permits one to suggest that an energy localization analogous to selffocusing phenomena well known in physics initiates the denaturation process in DNJ
Figure 26: A schematic picture of local strand separation.
5.8
Energy transport in alpha-helical proteins
The problem of energy transport along linear-chain molecules has long been posed. In the early 1970's it was already known that the energy (about 0.43 eV or 10 kcal/mol) released in the hydrolysis of adenosine triphosphate molecules (ATP) is an universal energy unit in many biological phenomena. However, the mechanisms of the energy trapping and transport were unknown. 90 In 1973 Davydov and Kislukha1 proposed a nonlinear mechanism for energy transport in one-dimensional chains. Later this mechanism was used to explain the energy transport in alpha-helical proteins.39'91 It was taken into account that proteins consist of chains of hydrogen-bonded peptide (H — N — C = 0) groups, and three such chains in a helical arrangement define the alpha-helix structure. 92 ' The energy released in the hydrolysis of ATP is not sufficient to excite the electronic states of alpha-helical molecule but it is sufficient to
excite the intramolecular C = 0 stretching mode (Amid-I vibration) with the excitation energy 0.21 eV. According to Davydov, the energy released in the hydrolysis of ATP could be trapped and transported in alpha-helical proteins as quanta of the intramolecular C = 0 stretching mode (Amid-I vibration). Indeed, Amide-I
132
vibrational energy might become self-trapped through an interaction with lowfrequency longitudinal acoustic phonons which arises because the Amid-I energy depends on the length of the hydrogen bond between the peptide groups.
The localized spatial region where the energy is trapped can be considered as a combination of intrapeptide and deformation excitations. It can propagate, and thus a soliton-like mechanism for energy transport is possible. Mathematical description of Davydov model was considered in Sec. 3.6. We can add here that many attempts to improve the model are known. The works 94 ' 40 ' 53 ' 95 ' 96 are among them. The model is well known in scientific community, but due to the absence of any reliable experimental basis, the validity of the model is still intensively discussed.97'98'99'100'101
5.9
Energy transfer along the protofilament in the tube of microtubulin
It is well known that neurons and other cells are comprised of protoplasm which in turn consists of membranes, organelles, nuclei, and the bulk interior medium of living cells: cytoplasm. The latter contains extensive arrays of linear polymers responsible for organization and integration of the cell behaviour. The most important of them are networks of microtubules, arrayed in parallel and interconnected by filamentous strands. The microtubules are hollow cylinders 25 nm in diameter whose lengths may span meters in some mammalian neurons 102 (Fig. 27a). The walls of the cylinders are assembles of 13 longitudinal protofilaments. Each of the protofilaments is a series of subunit protein known as tubulin. Each tubulin subunit is a dimer which consists of two slightly different classes of monomers known as a and /3 tubulin (Fig. 27b). Each dimer may be viewed as an electric dipole. The origin of the dipole character is in the fact that one binds 18 calcium ions ((7o2) per dimer. The tubulin dimer subunits are arranged in a hexagonal lattice which is slightly twisted. To explain the process of self-assembly of microtubules, Sataric et al proposed to use the nonlinear model. Taking into account that the process can occur when microtubules are in dynamical equilibrium in aqueous solutions of dissolved monomers, they assumed that in the course of self-assembly, the attachment of one dimer at the (+) end of a protofilament produces one torsion solitonic impulse (Fig. 28). This solitary wave propagates then through the corresponding protofilament. When the soliton reaches the ( —) end, its energy can be employed to detach the furthest dimer in protofilament. Fig. 29 shows that if the microtubule is free in the cytoplasm the process described above will lengthen the (+) end while the (—) end shortens, i.e. the microtubule will move towards the (+) end.
133
Figure 27: A scheme of microtubule (a) and a tubulin dimer subunit (b).
Figure 28: A schematic representation of solitonic role in the process of energy transfer from polymerized to depolymerized dimer.
134
Figure 29: A schematic representation of "treadmilling" assembly from tubulin dimers which could take place at the (+) end and with disassembly at the other extremity with liberation of tubulin dimers.
The dynamics of the model system can be described mathematically by the Hamiltonian
where Bn is the torsion angle of the n-th dimer; Ln is the corresponding an-
gular momentum; A,B,C,k,J are constants.
Hamiltonian Eq. 50 describes
protofilament as a chain of dimers which are capable of electric oscillations. The oscillating segments which form giant dipoles interact among themselves through the forces which are proportional to the inverse cube of the separation
distance between the dipoles. In the continuum limit, this model is reduced to one of the modifications of the model
135
5.10
Proton conductivity in biomembranes
According to the modern point of view, biological membranes are made of a mixture of lipids and proteins.104 The way in which these proteins and lipids are assembled in biological membranes is shown in Fig. 30. These lipids form an "oily" fluid bilayer in which the adherent proteins are free to float at will. Some of these proteins span the thickness of the bilayer. They are named "integral proteins". Other proteins are electrostatically attached to either the inner or outer surfaces of the bilayer; they are referred to as "peripheral proteins". According, to some integral proteins with a suitable fraction of hydrogen
bonding sidegroups, such as the hydroxyls of serine, threonine, tyrosine and the carboxyls of aspartic and glutamic acids, may fold in the hydrophobic lipid environment of the membrane in such a way as to form the chains of about 20 or more hydrogen bonds that span the membrane and the conduct protons across at. A fragment of the chain is shown schematically in Fig. 31. Since the anions always contain the oxygen atoms involved in a H bond with proton, we henceforth designate them as "oxygens" but their real rather complex structure must not be forgotten. The transport of protons in hydrogen-bonded chains is a long-standing problem. An important property of these chains is that the proton potentialenergy curve in each hydrogen bond has the form of a double well with two minima corresponding to the two equilibrium states of a proton between two neighboring oxygen atoms (Fig. 32). It is assumed that protons are transferred by jumps from one minimum to another along hydrogen bonds, i.e. by migration of ionic defects. A nonlinear model describing proton motion in hydrogen-bonded chains has been proposed by Antonchenko et <j/.loe They suggested that the remaining part of the chain, including the oxygen atoms does not behave as a fixed substrate and the lattice distortion around the ionic defects has to be considered. The basic idea of the model is that the coupling between oxygen atoms and protons can provide a mechanism which changes the potential barrier that protons have to overcome to jump from one minimum to another and makes their motion easier. So, the Hamiltonian of the model is the sum of three contributions
The first one is the proton part
136
here un is the displacement of the n-th proton with respect to the center of the oxygen pair; m is the mass of the proton; V(un) = e o (l —un/uo^ *s '^e double-well potential; £0 is the potential barrier; 2uo is the distance between the two minima; the last term represents the harmonic coupling between two neighboring protons with the characteristic frequency Wi. The second contribution is the oxygen part
where pn is the relative displacement between two oxygens in a pair; M is the mass of oxygen; fi0 is *ne frequency of the optical mode which is involved; the last term describes the harmonic coupling between neighboring oxygen pairs.
The third contribution describes the interaction between two subsystems
where x
ls
the strength of the coupling. The equations of motions resulting from the Hamiltonian have the form
In the case of strong coupling, a continuum approximation can be used, and Eq. 55 transforms to
where x = na is the continuum space variable; a is the lattice spacing; Co = au>i is the sound speed in the proton sublattice; Vo = afli. In the general case, solutions of Eq. 56 have not been found yet. However, in the particular case, when the coupling terms are neglected, soliton-like solutions of the equations can be easily found. These solutions could be interpreted as those describing the transport of protons.
137
Figure 30: A schematic representation of the model of biomembrane. It
consists of a lipid bilayer containing peripheral and integral proteins.
Figure 31: A fragment of a chain of hydrogen bonds.
138
Figure 32: A doubly periodic potential-energy curve.
6
Concluding remarks
In this review we presented different aspects of nonlinear dynamics of biopolyrners. First, we described the main theoretical results which led us to the conclusion that bjopolymers are an appropriate medium for excitation and propagation of nonlinear waves. We must note, however, that most of the models used are oversimplified. To construct a more realistic model, we should take into account: 1) a more detailed picture of the internal structure and mobility; 2) the effect of water and ion structure around biopolymers; 3) inhomogeneity of the internal structure; 4) statistical properties of nonlinear excitations (solitary waves) appearing in biopolymers. Second, we described different attempts to interpret experimental data in the framework of the nonlinear conception. We can conclude that this aspect of the problem is not so successful as the theoretical one: some of the results described above cannot be considered as a convincing evidence for the existence of nonlinear conformational waves in biopolymers because they admit alternative interpretations, and other results are required to be checked. Third, we described many possible applications of the nonlinear conception. This aspect of the problem could be considered as very perspective, because it might give an universal mechanism which could explain the role of dynamics in the functioning of biomolecules. However, these results are
considered now only as hypothetical, because they should be confirmed experimentally. 139
Acknowledgments I wish to thank prof. Alwyn Scott (University of Arizona, USA) for useful discussions and valuable comments on the manuscript.
References 1. A.S.Davydov and N.I. Kislukha, Phys. Status Solidi (b) 59, 465 (1973).
2. S.W. Englander et al, Proc. Nail. Acad. Sci. USA 77, 7222 (1980). 3. A. Fernander, Physica A 167, 338 (1990). 4. G. Carery et al, CRC Grit. Rev. Biochem. 3, 141 (1975).
5. P.G. Debruner and H. Frauenfelder, Annu. Rev. Phys. Chem 33, 283 (1982). 6. M. Karplus, Adv. Biophys. 18, 165 (1984). 7. H. Frauenfelder in Structure and Motion: Membranes, Nucleic Acids and Proteins, eds. E. dementi et al (Adenine Press, New York, 1985). 8. H. Fritzsche, Comm. Mol. Biophys. 1, 325 (1982). 9. J.W. Keepers and Th.L. James, J. Am. Chem. Soc. 104, 929 (1982). 10. W.R. McClure, Annu. Rev. Biochem. 54, 171 (1985). 11. L.V. Yakushevich, Molecular Biology, 23, 652 (1989) [in Russian]. 12. J.A. McCammon and S.C. Harvey, Dynamics of Proteins and Nucleic Acids (Cambridge Univ. Press, Cambridge, 1987). 13. J.A. Krumhansl and D.M. Alexander in Structure and Dynamics: Nucleic Acids and Proteins, eds. E. Clementi and R.H. Sarma (Adenine Press, New York, 1983). 14. A.S.Davydov, Uspekhi Fiz. Nauk 138, 603 (1982) [in Russian].
15. R.V. Polozov and L.V. Yakushevich, J. Theor. Biol 130, 423 (1988). 16. A.C.Scott et al, Proc. IEEE 61, 1443 (1973). 17. L.V. Yakushevich in Methods of Theoretical Physics and Their Applications to Biopolymer Science (Center Biol. Res. Acad. Sci. USSR, Pushchino, 1990).
18. A.C. Scott, Am. J. Phys. 37, 52 (1969). 19. L.V. Yakushevich, Studia Biophys. 121, 201, (1987). 20. S.Yomosa, Phys. Rev. A 30, 474 (1984). 21. S. Homma and S. Takeno, Progr. Theor. Phys. 72, 679 (1984).
22. Ch.-T. Zhang, Phys. Rev. A 35, 886 (1987). 23. L.V. Yakushevich, Phys. Lett. A 136, 413 (1989). 24. L.V. Yakushevich, Studia Biophys. 140, 163 (1991).
25. J. Gaeta, Phys. Lett. A 143, 227 (1990). 26. S.N. Volkov, Preprint ITP-84-52P (Inst. Theor. Phys., Kiev, 1984).
140
27. A. Khan et al, Bull Math. Biol. 47, 783 (1985). 28. L.A. Ostrovskii and A.M. Sutin, /. Appl. Math. Mech. 41, 543 (1977).
29. V. Muto et al, J. Biomol. Struct. Dyn. 5, 873 (1988). 30. M.Toda, Phys. Rep. C 18, 1 (1975). 31. P.L. Christiansen et al, Preprint (Tech. Univ. Denmark, 1990). 32. V. Muto, Nanobiology, 1, 325 (1992). 33. J.A. Krumhansl et al in Structure and Motion: Membranes, Nucleic Adda and Proteins, eds. E. dementi et al (Adenine Press, New York, 1985). 34. Z. Zhang and W. Olson in Proceedings, 6th Annual Conference on Nonlinearity of Condensing Matter, eds. A.R. Bishop et al (Springer, Berlin, 1987).
35. Ch.-T. Zhang, Phys. Rev. A 40, 2148 (1989). 36. J.-X. Xiao et al, J. Phys. A 20, 2425 (1987). 37. S.N.Volkov, J. Theor. Biol. 143, 485 (1990).
38. L.V. Yakushevich, Physica D 79, 77 (1994). 39. A.S.Davydov and A.D. Suprun, it Ukrain. Phys. J. 19, 44 (1974) [in Russian], 40. V.K.Fedyanin and L.V. Yakushevich, Int. J. Quant. Chem. 21, 1019 (1982). 41. M. Nakanishi et al, FEBBS Lett. 81, 61 (1977).
42. M. Nakanishi and M. Tsuboi, J. Mol. Biol. 124, 61 (1978). 43. C. Mandal et al, J. Mol. Biol. 135, 391 (1979). 44. V.K.Fedyanin et al, Studia Biophys. 116, 59 (1986). 45. V.K.Fedyanin and V. Lisy, Studia Biophys. 116, 65 (1986). 46. M.D. Frank-Kamenetskii, Molecular Biology 17, 639 (1983) [in Russian].
47. M.D. Frank-Kamenetskii, Nature 328, 17 (1987). 48. 49. 50. 51.
M. Gueron et al, Nature 328, 89 (1987). T. Weidlich et al, J. Biomol. Struct. Dyn. 8, 139 (1990). S.J.Webb, Phys. Rep. 60, 201 (1980). V.S.Bannikov et al, Dokl. Akad. Nauk USSR 253, 479 (1980) [in Russian] .
52. V.S.Bannikov and S.B. Rozhkov, Dokl.
Akad.
Nauk USSR 255, 746
(1980) [in Russian].
53. A.C. Scott, Phys. Lett. A 86, 60 (1981). 54. P.S. Lomdahl et al, Phys. Lett. A 92, 207 (1982). 55. S.P. Layne et al, Proc. Natl. Acad. Sci. USA 82, 7599 (1985). 56. K.G. Brown et al, Proc. Natl. Acad. Sci. USA 64, 1467 (1972).
57. T. Weidlich and S.M. Lindsay, /. Phys. Chem. 92, 6479 (1988). 58. T. Weidlich et al, J. Phys. Chem. 92, 3315 (1988). 141
59. H. Urabe and Y. Tominaga, J. Phys. Soc. Jpn. 50, 3543 (1981). 60. H. Urabe and Y. Tominaga, Biopolymers 21, 2477 (1982). 61. H. Urabe ei at, J. Chem. Phys. 78, 5937 (1983).
62. 63. 64. 65.
J.W. Powell ei al, Phys. Rev. A 35, 3929 (1987). L.V. Yakushevich, SPIE 1403, 507 (1990). S.J. Webb and A.D. Booth, Nature 222, 1199 (1969). M.L.Swicord and C.C. Davis, Biopolymers 21, 2453 (1982).
66. M.L.Swicord and C.C. Davis, Bioeleciromagneiics 4, 21 (1983).
67. G.S. Edwards ei al, Phys. Rev. Lett. 53, 1284 (1984). 68. V.Ya. Maleev et al, Biopolim. Kletka 2, 35 (1986). 69. C. Gabriel et al, Nature 328, 145 (1987). 70. K.R. Forster et al, Biophys. J. 52, 421 (1987).
71. R.A.Garibov and A.V. Ostrovskii, Uspekki Sovrem.
Biol.
110, 306
(1990) [in Russian].
72. L.L. Van Zandt, Phys. Rev. Lett 57, 2085 (1986). 73. L.L. Van Zandt and M.E. Davis, /. Biomol. Struct. Dyn. 3, 1045 (1986).
74. M.E. Davis and L.L. Van Zandt, Phys. Rev. A 37, 888 (1988). 75. V.K.Fedyanin and L.V. Yakushevich, Studia Biophys. 103, 171 (1984). 76. K.F. Baverstock and R.D. Cundall, J. Radial. Biol. 55, 151 (1989). 77. P.R. Selvin et al, Science 255, 82 (1992). 78. O. Edholm et al, Quart. Rev. Biophys. 17, 125 (1984).
79. P. Jensen ei al, Phys. Lett. A 95, 204 (1983). 80. H.M. Sobell in Biological Macro-molecules and Assemblies, eds. F.A.Jurnak and A. McPherson (John Wiley & Sons, New York, 1984).
81. 82. 83. 84.
L.V. Yakushevich, Nanobiology 1, 343 (1992). M.Salerno, Phys. Rev. A 44, 5292 (1991). M. Salerno, Phys. Lett. A 167, 49 (1992). M.Salerno and Yu.S. Kivshar, Phys. Lett. A 193, 263 (1994).
85. J.J. Ladik et al, Int. J. Quant. Chem. QBS Suppl. 5, 35 (1978). 86. J.J. Ladik in Structure and Motion: Membranes, Nucleic Acids and Proteins, eds. E. Clementi et al (Adenine Press, New York, 1985). 87. A.R. Fersht, TIBS 12, 301 (1987). 88. M. Peyrard and A.R. Bishop, Phys. Rev. Lett. 62, 2755 (1989). 89. M. Peyrard and A.R. Bishop in Nonlinear Coherent Structures, eds. M. Barthes and J. Leon (Springer-Verlag, Berlin, 1990). 90. D.E. Green, Science 181, 583 (1973). 91. A.S.Davydov, J. Theor. Biol. 96, 379 (1977). 92. L.Pauling and R.B. Corey, J. Am. Chem. Soc. 72, 5349 (1950). 93. L. Pauling ei al, Proc. Nail. Acad. Sci. USA 37, 205 (1951).
142
94. V.K.Fedyanin and L.V. Yakushevich, Int.
J. Quant.
Chem.
16, 215
(1979). 95. H. Motschmann et al, J. Phys. Condens. Mater. 1, 5083 (1989).
96. 97. 98. 99.
S.Ichinose, Phys. Status Soiidi (b) 155, 467 (1989). D.W. Brown et al, Phys. Rev. A 33, 4110 (1986). A.C. Scott, Physics of Many Particle Systems 12, 33 (1987) [in Russian]. M.A. Collins, /. Chem. Phys. 88, 399 (1988).
.00. L01. 102. LOS. 104.
L. Cruzeiro et al, Phys. Rev. A 37, 880 (1988). M.J.Skrinjar et al, Phys. Lett. A 133, 489 (1988). S. Rassmussen et al, Physica D 42, 428 (1990). M.Sataric et al, J. Mol. Electronics 6, 63 (1990). J.H. Byrne and S.C. Schultz in An Introduction to Membrane Transport and Bioelectricity (Raven Press, New York, 1988). LOS. J.F. Nagle and H.J. Morowitz, Proc. Nail. Acad. Sci. USA 75, 298
(1978). L06. V.Ya. Antonchenko et al, Phys. Status Soiidi (b) 115,631 (1983). L07. M. Peyrard et al, Phys. Rev. A 36, 903 (1987).
143
THE GROUND STATE OF FROHLICH ONE-DIMENSIONAL FERMI-POLARONS E.A. KOCHETOV Joint Institute JOT Nuclear Research Bogoliubov Laboratory of Theoretical Physics
141980 Dubna, Moscow region, Russia We study a one—dimensional system of N electrons in an ion crystal described by the Frohlich N—particle H a m i l t o n i a n . The formation of a system of two particles with opposite spins, i.e. Frohlich bipolarons is shown to be the most energetically preferable in the limit of strong electron—phonon coupling.
1
Introduction
Recent attempts at the microscopic interpretation of high— temperature superconductivity (HTSC) have aroused interest in the model of polarons of a large radius, i.e. the Frohlich polarons. A superconducting state in this model is realized as a Bose—condensate of bipolarons.1 Since a polaron arises only in the strong electron-phonon coupling limit, the properties of the polaron collective state resulting in superconductivity can be studied on the basis of data on the ground and weakly excited levels of a set of N electrons strongly correlated by a common phonon cloud. However, in view of obvious technical difficulties, one usually studies the properties of a single bipolaron, assuming its parameters to determine the properties of the collective motion such as the temperature of transition into a superconducting state.1'2 As for the mechanism of bipolarons formation and properties of the spectrum of elementary excitations in the vicinity of a bipolaron condensate, these problems still remain to be solved (waiting for their investigation). The ground state of a strongly coupled one—dimensional polaron system is shown in this note to be a condensate of bipolarons. We prove this statement within the variational approach typical of the problems with strong coupling in which the results are obtained by comparing upper estimations of energies. It should be pointed out that bipolaron clusterization of the wave function arises from strong effective repulsion caused by the Fermi-statistics of particles. Direct Coulomb repulsion is a weaker effect and does not influence the qualitative picture. For the Bose—statistics of polarons, it is just the Coulomb repulsion that provides the energy stability of multipolarons, the latter having a different number of particles. Note also that the determination of the ground state of an n dimensional 144
polaron system for n > 1 is technically a more complicated problem since the degeneration of eigenfunctions of the approximating Hamiltonian grows rapidly with n. The one-dimensional motion is considered as motion in a distinguished direction in a 3 dimensional crystal. Such an isolated direction can occur when a strong electromagnetic field is applied to a sample. 4'5
2
Hamiltonian of the system
The Hamiltonian of a system of charged particles in a quantized scalar field in one dimension is of the form
Here we introduce the following notation:
• V"(£) = V'(r( a! ) i s a quantized Schrodinger field with excitation quanta describing charged Fermi-particles; the z-component of spin a takes the values ±1. The particles are localized in one dimension:
• ak and a*, are quantized amplitudes of the phonon field corresponding to the optical branch of lattice vibrations with frequency uk and wave vector k:
• when linear dimensions L —v oo, we have
In what follows we will use dimensionless units, setting u>k = u> = m — 1. Then, in accordance with Ref. 6, we can put
where a is a dimensionless coupling constant.
145
To take account of the effect of polaron structure in the first approximation, we consider the Lee-Low-Pines transformation that dresses the operators i/> in a phonon cloud:
where
Here /j. and /£ are c-numbers to be determined from the condition of energy minimum. The {/-transformation separates the component representing an effective potential in the first approximation:
from the quantized amplitudes oj. and besides,
where «/ = ds/dx. Since the transformed operators V1 (£) anc' lACO describe the production and annihilation of structure excitations (polarons), it is natural to expand them into the complete set of square—integrable functions on the real axis:
where {>i(x)} is the set of functions complete on (—00, oo), and
We take the system {$i} to be a set of eigenfunctions of a harmonic oscillator with frequency Q that will be considered as a variational parameter:
Then operators a\a and a^a are operators of creation and annihilation of excitations from the vacuum state, i.e. the ground state of the Hamiltonian
146
As a trial ground state of the system Eq. 1, we choose the state
where \N > is the ground state of the Hamiltonian Eq. 3 for a fixed number of particles, N, and |{ajb} > is a coherent state of the phonon field; the functions ak will be determined from the condition of energy minimum. Thus, averaging Eq. 1 over the trial state Eq. 4 we obtain the ground-state energy as a function of fi, a/, and /£.
3
Formation of bipolaron Bose-gas
When particles obey the Bose-statistics, the state \N > is a condensate of N particles into the ground state <^ o (z). For Fermi-particles, the state \N > is a result of subsequent filling of states in accordance with the Pauli principle:
where N = 2k and the state |(Ti)i > is the i-th excitation of a one-dimensional oscillator with energy f^ filled with two particles of opposite spin projections.
(For structureless particles, states are filled with momenta starting from zero to the Fermi—momentum pp •) Averaging Eq. 1 over the states Eq.4 and minimizing the resulting expression with respect to o/t and fk, we obtain the upper estimation of the ground—state energy
where we have neglected direct two-particle repulsion and introduced the following notation:
Calculating the averages in Eq. 6 with Eq. 5 and minimizing the result with respect to fi, we obtain in the weak coupling limit
147
which signifies the independence of polarons in the first in a approximation, and in the strong coupling limit
where
and
is a generalized Laguerre polynomial. As an example, consider systems of two and four particles. Taking into account that
we find from Eq. 7
Since the ground—state energy of a one—dimensional polaron, according to Eq. 6, equals
it can be seen that
but
Thus, the state of two uncoupled bipolarons is energetically more preferable than the bound state of four polarons. In the case of interest for physics, when the number of particles is asymptotically large (N —<• oo), using the Trikomi expansion for Laguerre polynomials, we obtain
148
where J\(x) is the first-order Bessel function. Taking into account that
we find from Eqs.
8 and 7
i.e.,
Therefore, the bound state of the system of 2ifc Fermi—polarons is energetically instable with respect to the disintegration into k free bipolarons. Thus, the effect of strong electron-phonon interaction in the one-dimensional Fermi-system of polarons results, in the first approximation, in the formation of the Bose—gas of bipolarons. The spectrum of elementary excitations determining physical (macroscopic) properties of the system should now be determined with respect to a new vacuum, i.e. the condensate of bipolarons.
1. D. Emin, Phys. Rev. Lett. 62, 1544
(1989).
2. D. Emin and M.S. Hillery, Phys. Rev. B 39, 6575 (1989). 3. E.A. Kochetov and M.A. Smondyrev, Tear. Mat. Fiz. 85, 74, (1990) [in Russian].
4. E.A.
Kochetov E.A.
et al, Z. Phys. B 89, 177 (1992).
5. M.A. Smondyrev et al, Europhys. Lett. 19, 519 (1992). 6. F. Peeters and M.A. Smondyrev, Phys. Rev. B 43, 4920 (1991).
149
MEAN-FIELD THEORY OF THE SOLVATED ELECTRON AND DIELECTRON STATES G.N. CHUEV Department of Quantum-Mechanical Systems, Institute of Mathematical Problems of Biology, Russian Academy of Sciences, Pushchino, Moscow Region, 14S29S, Russia The self-trapped electron and dielectron states are considered in a disordered medium. The model is based on a variational mean-field estimation of path integrals. As a result, the problem is reduced to evaluation of the Shrodinger-like equation. The asymptotic behavior of the effective potential of this equation is investigated for the aolvated electron as well as for the dielectron (bipolaron) state. The statistical treatment of the environment allows us to consider the effect of environment pressure, temperature and added salts concentration on the electron states. These effects are estimated for the nonadiabatic electron transfer and for the shape of
the absorption spectrum, for its half-width and maximum. The model is compared with other theories of the self-consistent trapped electron states in disordered solidstate and liquid systems. The possibilities of the model are discussed for a wide range of problems of the solvated electron theory.
1
Introduction
The solvated electron problem has been studied for decades. At present, there are techniques allowing one to obtain sufficient experimental data on the solvated electron behavior in different media l. The solvated electron is also the object of intensive theoretical modeling (see reviews 2 ' 3 and references therein). The review is devoted to a solvated electron model, namely, a mean-field theory of the solvated electron and dielectron states in a disordered medium. Note that approaches based on the self- consistent state concept are actively investigated for an excess electron in ionic crystals and polar semiconductors (polaron and bipolaron models) 4 ' 5 , in heavily doped semiconductors 6 ' 7 > 8 ' 9 ) in dense gases 10, in nonpolar fluids ll . 12 > 13 . 14 ] m molten salts 1 5 > 1 6 ) and for the hydrated electron 17. We focus here on the original interpretation of the concept 18 and compare it with other methods. The developed solvated electron models are briefly surveyed in Sec. 2. The Sec. 3 deals with the statistical treatment of the electron state. The problem is formulated in Sec. 3.1 where the terms describing the solvated electron state are defined. The statistical description of the electron environment is examined in Sec. 3.2 where an effective action of the electron is derived by correlation functions over the medium density. In Sec. 4 a key mathematical method based on the reference action method is formulated. The general approach of the
150
method is described in Sec. 4.1. in Sec. 4.2 it is used to derive the Shrodinger equation for the solvated electron state in the molecular fluctuating medium. The physical nature of the effective interaction potential between the electron
and the medium is investigated in Sec. 4.3 where the semicontinual treatment of the potential is given. The mean-field consideration of the dielectron (bipolaron) state is studied in Sec. 5. In Sec.6 the model is used to estimate the electron transfer and the absorption spectrum. Particularly, the shape of the
absorption spectrum is studied in Sec. 6.1. The effect of environment temperature, pressure, and added salts concentration on the solvated electron state is estimated in Sec. 6.2 for the gaussian form of the absorption spectrum. The nonadiabatic electron transfer is examined in Sec. 6.3 on the basis of the mod-
el. In Sec. 7 our theory is compared with other models of the electron state in disordered solid-state systems (Sec. 7.1) and in liquids (Sec. 7.2). The conclusion is given in sec. 8 where the possibilities of the model are briefly viewed. The appendices with deducation of the main formulas are given at the end of the paper.
2
Models of the solvated electron
In terms of theoretical and computer modeling of solvated electron, the following approaches can be highlighted: a.) Polaron models. The main part of these models is based on the Pekar approach 4 . It should be noted, that this model was the first to be used in the russian scientific school 3 > 19 ' 20 . The model is based on the assumption that the self- localized electron state is due to the strong interaction between an excess electron and the polarization field induced by the electron. As a result, the self- consistent
electron state forms. This state is determined by macroscopical parameters of the medium, i.e. by the medium dielectric constants. The advantages of the model are its physical simplicity and clearness. However, its application
to real systems is very difficult, since the model does not consider the detailed molecular structure of the electron environment. The polaron models cannot also be used for the correct estimation of the thermodynamic effects (pressure, density of the medium, etc.) on the electron state. b.) Models based on the quantum- chemical calculations. The models used different modifications of quantum- chemical methods (semiempirical 21, ab initio 22, cluster 2 3 ) to calculate the localized electron state and its molecular environment. The models seem to be good for the detailed molecular structure effects. But the dependence of the electron state on the environment can only be obtained numerically, which does not allow
151
one to define the physical picture of the electron behavior. The calculation of
thermodynamic effects is likely to be impossible for these models. c.) Semicontinual self-consistent models. These models ' are the combination of the two previous types of approaches, they treat the discrete molecular structure of the environment (coordination number, size and charge of the environmental molecules) as well as the self-consistent polarization of the medium as a. whole, induced by the excess charge. Moreover, the short-ranged part of the electron- medium potential has the cavity form, while the long-ranged part is considered in the polaron approximation. As a rule, these models estimate the thermodynamic effects in accordance with the experiment 2 . The shortcoming of these models is a set of semiempirical parameters (radius and depth of the cavity), whose definition is rather arbitrary. d.) Models based on the path integrals treatment.
These models calculate high- dimensional integrals statistically and based on the Monte Carlo method. They are actively investigated at present 25 ' . Recently they have been applied to the bipolaron (dielectron) state 27 . There is a number of the methods allowing one to estimate correctly the expected values determining the state of a quantum particle in the classical environment. Since the environment is considered as a discrete molecular structure, the quantumchemical methods can be incorporated into the calculation scheme. The thermodynamic effects of the environment ( pressure, density) can also be also calculated within the method. But these models require modern high- performance computers and huge volume of calculations. Method does not enable to estimating the influence of numerical discrepancy in the input data on the results. e.) Mean-field models of the solvated electron. The basic idea of the mean-field models is to combine the merits of the two types of the approaches, i. e. the physical simplicity and clearness of the variational estimations for the polaron model and the possibility to consider statistically the detailed molecular structure by the path integrals. The idea was first stated for the solvated electron in 28 as the RISM-polaron theory. This theory was used as a base in numerous works 11'12'14'15'17'29'30'31. The liquid structure is treated by the reference interaction site model (RISM) in these
works. Another type of the mean-field models is connected with the solution of the effective Shrodinger equation 13.18.32'33 for the ground solvated electron state. We present here the original interpretation of the model 18 . The initial formulation of the problem is defined by the pair- wise inter-
action electron-isolated site potential allowing one to treat the environment microscopically. The isomorphism of the path integrals in quantum-mechanical
152
and statistical statement of the problem (see Fig.l) permits one to use the correlation function formalism and to include into consideration the quantum effects as well as the statistical influence of the medium. The simple mean-field variational estimations reduce the problem of the self-consistent free energy functional, which results in the Shrodinger-like equation for the electron wave functions. The application of the model may be wide and include the calculation of the absorption spectrum ' as well as the dynamics of electron mobility 12 . The comparison of the model and other theories of the self-localized electron in disordered medium is given in Sec. 7.
3 3.1
The statement of the problem Formalism of statistical treatment
The quantum particle state is defined statistically by the density matrix /j(r, r',/9) depending on temperature T = I//? and denoting the transfer probability from the state r to the state r' . The density matrix also depends on the time t for non equilibrium states. If we know the density matrix, we can calculate different averaged values
determining the electron state: characteristic size of the electron state, its averaged kinetic and potential energy, the influence of the weak external field, the electron mobility and magnetic susceptibility, the form of the absorption spectrum, etc.. One of the main quantities of the electron state is the partition function, which is a convolution integral of the density matrix
Below we derive a set of formulas for the partition function but not for the density matrix. It is due to awkwardness of the expressions for the density matrix. For quantum systems, the density matrix can be considered as an infinitydimensional path integral 34
153
here S is the action of electron, and symbols denote
In general, the problem of the path integral numerical calculation for a quantum particle can be reduced to the isomorphic many- particle problem of the ring polymer35 (Fig.l). There are effective computational methods for this problem, which perform high-dimensional integration by the Monte- Carlo or quantum molecular dynamic method
Figure 1: Isomorphism of problems of electron and ring polymer in solvent. Black circles denote electron density distribution and the corresponding ring polymer structure, while large circles present surrounding solvent molecules.
When the electron interacts with environment, the density matrix depends on the environment state. If we consider classical fluids as a medium, the partition function Z of the solvated electron in the classical fluid can be written as
154
the usual configurational integral over medium coordinates R n and the functional integral over the electron coordinate T(T). Note that in the general case the environment must be taken into account. We suppose that the degrees are connected with the displacements of the closed electron shells of the molecules. The displacement due to the shell polarization also offered for simplicity and does not restrict the method. Then, the partition function Z can be written
Here Se is the action of the electron in the medium, which can be calculated if we know the electron- medium potential, and M; is the dipole momentum of the i-th particle of the fluid, which is due to the displacement of the electron shell. We divide the interaction between fluid particles into the classical part U0, which is independent of the electron shells displacement, and the action Sjy| attributed to the quantum degrees of freedom
The first and the second terms are the kinetic and potential dipole energy respectively, while the last term is due to the interaction between dipoles. Here, we consider the simplest Drude model of the electron shell, which imply that dipoles oscillate with equal frequency o>o. We assume that the instantaneous part of the dipole equals zero, the inclusion of this part into consideration
does not change the result but complicates the derivations of formulas. The last term of (6) is proportional to the dipole- dipole operator T(r) = ®(T —
p f c )VV|r|
, where 0 is the step function and r^ is the hard core diameter of
a molecule. We also divide the electron action Sc into two parts, the first one Sj^c is independent of the electron shell displacements, while the second Sf^e is due to the interaction between the excess electron and closed electron shells. In general, the latter can be intricately connected with the dipole momentum of the fluid particle, but we assume that it is proportional to the dipole momentum
155
The action Sc of the electron in the medium can be calculated if we know the electron- medium potential. We present the potential as a sum of the pairwise electron-isolated site interactions. In Eq. (7) m is the electron mass, and u is the pair- wise potential, which can be estimated by scattering at an isolated atom. In some cases, approximations of the potential, such as the hard sphere potential can be used. In fact, u in Eq. (7) can depend on the additional indices denoting the sort of molecules or the type of sites for the same molecule. To simplify the formulas, we drop these indices. All the expressions can be easily generalized. We will be interested in the case when there is a gap between the energy of the excess electron and bound electrons. In other words, the state of the solvated electron is different from the states of closed electronic shells (the relation with the band structure in liquids will be discussed in Sec. 7.2). Thus, our problem is transformed to the calculation of the density matrix for the solvated electron. On the whole, this problem consists in the evaluation of the integral (5) having N3 classical and (JV + 1)3 quantum degrees of freedom, but the solvated electron state differs from the bound electrons. The dimension of (5) is too huge, thus the theoretical problem is to reduce the dimensionality of the integral, retaining all the interesting physical properties of the system. In some cases, it can be done by extraction of the degrees of freedom, which strongly interact with the solvated electron. In other cases, the system reduction is due to the choice of action form allowing to integrate (5) analytically for some degrees of freedom (for quadratic form of the action ). Sometimes, it can be possible due to low dimension of space (see 8 ). From the theoretical viewpoint, the successful employment of the path integrals method in statistical physics is due to the fact that the production functional of the system can be obtained in closed form by the path integrals37. By this, its functional derivatives determine a set of equations for the correlation functions of the system in the compact form. This treatment allows one to make a theoretical analysis of the statistical behavior of the system. 156
3.S
Description of medium by density correlations
The form of the electron action (6) is quadratic over dipole momentum. Therefore these degrees of freedom can be analytically excluded, and action (6) can be reduced by the effective potential (see Appendix 1)
-j
where «^, teraction
is the nonlocal effective potential due to the electron - dipole in-
Here A±j are the Fourier elements of matrix
Thus, the problem of the solvated electron in the quantum polarizable fluid can be treated as the problem of the electron in some effective fluid with potential (9). It should be noted that effective action (9) is similar to the electron- phonon interaction. (If the medium interaction potential Ug is harmonica!, phonon coordinates can be introduced. Assuming the linear
dependence of the electron- particle potential on the phonon coordinates, we can obtain the Pekar-Frohlich hamiltonian 4'38 from (9).) The next stage of the reduction is a short-cut description of the medium by a set of correlation functions. Let us introduce the equilibrium correlation functions of the density medium Xn defined as 39
Here Xi (R.) >8 tne average medium density, while Xz(H-i R-') denotes the densitydensity correlation function connected with the structure factor. The correlation function can be obtained from the analytical statistical theory of liquids or 157
by molecular modeling . Some information about this function can be found experimentally for simple liquids. We can determine Xs(K-i H/> R.") in a similar way. Figure 2 plots schematically the distance dependence of density-density
correlation functions and potential u(r~). This potential includes short-range repulsive part u, and long-range attractive part u\. We introduce the generalized Mayer functions / of the first order and / of the second order for the electron,
Then, the partition function can be transformed (see Appendix 2) by effective action S e , which only depends on the electron path and the density correlation functions of all the orders. In general, we may consider such higher correlation functions over medium density as we need. But in many practical cases, one restricts the molecular structure treatment by the second order cor-
relation functions, and assumes that xs = • • • = 0. This is common practice in statistical physics 39>4° to reduce the dimension of the studied system. As a result, we obtain
In expression (14), the symbol * denotes the convolution integration
158
Figure 2: Distance dependence of density-density correlation functions (a) (1 for gas, 2 for liquid, and 3 for solid state) and potential u(r) (b).
Formulas (9)- (14) determine the effective interaction between an excess electron and the medium by the correlation functions of the medium density, which are the statistical quantities while the electron- particle potential
is determined by the quantum-chemical calculations.
In some problems of
disordered systems 8 , the electron - medium potential is the statistical quantity, and its different approximation are used . From our viewpoint, relation
(14) is more convenient, since it uses independent characteristics of the medium and electron-particle interaction. Therefore, they can be defined within independent approaches. The density correlation functions can be obtained
by statistical theory, while the pair- wise electron- particle potential may be gained from quantum- chemical calculations. Thus, if we ignore nongaussian correlations, we may reduce the dimension of the configurational integral to the two-dimensional integrals. As a whole, our problem is to calculate the path integral (14) where effective action is 159
nonlocal in time and space.
4 4-1
The effective potential for the solvated electron Formalism of the reference action
In fact, the integral of the type (14) is analytically calculated only for a special form of the action . There are a few mathematical methods allowing one to calculate the path integrals. These are the perturbation methods and the saddle point method for functional integrals. Both the methods are based on the formalism of the reference action 34 . Using a reference action So, we can estimate the partition function as
where the symbols denote
If the partition function Zg(Tn) is known, we can include the high order corrections using the perturbation methods and thereby ascertain the inequality. The variational estimation of (15) is another approach analogies to the saddle point method. Relation (15) is valid for any real 5 o (r n ), however, the estimation of the partition function by (15) will be optimal, if we fit the parameters Fn of So to maximize the right side of (15). Hence, we can replace the order of integration by an electron paths and medium coordinates and reduce the problem to calculation of the averaged Mayer function
We can define < /(R)/(R') > and < f(R,R') > in a similar way. Usually, the quadratic form of the reference action is used 34 > 36 ) it enables ones to calculate the integrals (16) analytically. In the case when the electron strongly interacts with the medium and forms a localized state, the electron state can be calculated in the mean- field approximation by the reference action which is determined in self-consistent manner, (see Fig.3).
160
Figure 3: Instantaneous configuration of an excess electron in polar liquid obtained by the Monte-Carlo method: delocalized (a) and localized (b) states. Dark circles denote electron distribution, light circles are the positions of surrounding molecules.
4-2
Mean- field approximation
It is possible to obtain the effective Shrodinger equation determining the electron state in the molecular medium by the reference action method 13 ' 18 . Let us introduce the reference action So, which is defined by the onepartition Green function G(r,r') as
where >n denotes a set of the electron wave functions <j>n with energy En. In the mean-field approximation, the averaged values can be expressed by the 161
one-partition Green function (see Appendix 3). Consequently
where G =< G(r, r) >= / G(i, r)dr, and u,(r), UI(T) are the short- and longrange parts of the electron-particle potential. For the Coulomb polarizable systems, where the term proportional to the first power of uj(r) vanishes, we
should include into considerations the terms proportional to U|(r)wj(r ). If the ground electron state with ^o is not degenerate, that is /3(E0 — Ei) » 1, it follows from (19) that
To compute the averaged Mayer function / for nonlocal time-dependent potential, it is necessary to know the dynamic behavior of the system, and the frequency-dependent correlation functions of the medium. There are a few
approaches to avoid these difficulties. They are self-consistent (SCA) and adiabatic approximations (AA). The adiabatic treatment yields the low-frequency limit, when uo/E —> 0, and the electron coordinate can be regarded as an adiabatic variable. The other limit <*Jo/E ~* °° corresponds to the self-consistent approximation, when the environment can be treated as a classical polarizable fluid. As a result, ones can express the second order Mayer function of the electron as
In the self-consistent treatment we get UQ = oo.
The obtained relations are the consequence of the mean- field approximation and the definition of the Mayer function. Thus, the free energy functional can be expressed by a set of the electron wave functions determined by the
reference action S0. solvated electron
We may also estimate the entropy change r,.nt for the
162
The optimal choice of So giving the best estimation of (14) is obtained from the extremum of the functional ^j = 0, witch yields the effective Shrodinger equation for the ground and excited electron wave functions
We may obtain the continual limit of this equation, if we postulate that Xz — 0The Shrodinger equation has a simple physical nature. The effective electron- medium potential is denned by the sum of the electron interaction
with all the particles Vfj (r) ~ / dRp(R,
equations were obtained both in solid-state and liquid disordered systems (see Sec.7). It should be noted, that although in these cases the steady electron states are described by the same equations, the dynamical behavior of the electron localization may be different. In the solids there is a distribution of
localized electron states and the main contribution to the free energy functional is due to the electron state described by (23), moreover, the medium density changes insufficiently. While in liquids the electron induces large fluctuations in the medium density, and it is self-localized on these local fluctuations. If the interaction between electron and medium particle is week and
ft < u >«
1, Eq.( 23) has the instanton form 4 1
This form is well known in the field theory, for ordered systems it corre-
sponds to the polaron problem 4'5 (the excited states of polaron were investigated i n 4 2 ) , while in disordered medium it is obtained in the problem of the electron localization in gaussian random field (see Sec. 7.1).
Thus, we have reduced the calculation of path integral (9) to the investigation of the Shrodinger equation. The solution to the problem can be obtained by direct numerical calculations or by variational estimations. Both 163
the methods have conventional mathematical supplement and do not require huge computing resources. As a result, it can be found how the electron state
depends on the environment. The examples of the estimations will be given in Sec.6. 4-3
Semicontinual treatment of the effective potential
Eq.(23) determines the self- consistent effective potential of the electron being in the equilibrium in the molecular medium. The long-ranged and shortranged parts of the potential coincide with the results of the Semicontinual theories2. Particularly, it can be found that the shot- ranged part has a cavity form V,j (k —» oo) = Vo where Vo is the cavity depth. On the other hand, we can find for Ve/ (k -* 0) that
where u,(k) denotes the Fourier transform of the Coulomb part of the electronparticle interaction. The expression in square brackets is connected with static dielectric constant e of fluids 40 .
while the second term of (24) can be expressed by the high-frequency dielectric constant too for our model
In the self-consistent limit, we should use e^ = 1. Therefore, the effective potential of the solvated electron in the polar disordered medium has the following asymptotics
Considering the Debye screening for an electrolyte, we can obtain the result similar to 2O. The difference between the self-consistent field and adiabatic treatments of the solvated electron in a polar fluid was discussed in 43. In the simplest case, this difference can be taken into account by replacing c = 1 — - with the Pekar factor cc} = — - -. J
«oo
€
164
Simple joining of asymptotics results in the semicontinual effective potential 2 , which includes cavity and polaron tails
In essence, the semicontinual treatment is the approximation of our potential with additional phenomelogical parameter, i.e. cavity radius (see Fig.4). Using electron distribution, we can calculate also the electron-solvent correlation function ge,(r), which mean the probability to find a solvent molecule at distance T from the center of electron localization. Figure 5 presents the comparison of the simulation data with the mean-field calculation for ^.(r).
In our statistical formulation, this parameter is not necessary and the form of the potential can be calculated from the microscopic consideration of the medi-
um, i.e., from the electron- particle interaction u(r — R) and the correlation functions of the medium density XiiXl-
Figure 4: Effective electron-medium potential. Solid line corresponds to the mean-field model, the dashed one to the semicontinual treatment, triangles present data of quantum molecular dynamic simulations.
165
Figure 5: Comparison of the simulation data (line 1) with the mean-field
calculation (line 2) for electrron-positive ion correlation function ge+(r) in molten salts.
5
Mean-field treatment of the dielectron states
The dielectron state in liquids has also been the subject of experimental and theoretical study (see references to earlier works in 4 4 ). Although there are experimental data indicating that spin pairing occurs in metal- alkali halide and metal-ammonia solutions 45'46, the theory is not yet as comprehensive as experiment. Recently, the success combination of the path simulation tech-
nique and the density functional approach 2r has stimulated the interest in the problem. We sketch here the employment of our model for the dielectron problem. All the above consideration can be extended to the dielectron problem. In this case we should include into consideration the electron-electron interaction.
166
Then, the action of the dielectron problem can be written as
Here the second term represents the electron-electron interaction due to the
Coulomb repulsion. Repeating the above derivation, we express the effective action of the dielectron as
where fdi, fdi are respectively the generalized first order and second order
Mayer functions for the dielectron
To reveal the physical nature, we consider only the ground state of the localized dielectron. Then using the mean-field approximation, we can present the averaged Mayer functions of the dielectron as
where $(r, r') is the dielectron wave function of the ground state. Here we take into account the physical equivalence of both electrons. As in the one-electron case, the complete physical nature of the dielectron is determined by the frequency-dependent correlation functions. The second
167
order Mayer function fn of the dielectron can be evaluated by the wave function $(r, r') only in self-consistent and adiabatic limits. As a result, we can obtain
Here the first term in the exponent of (35) represents the one-electron correlation, while the second is due to the two-electron correlations. The free energy for the dielectron formation can be expressed as
As with the electron problem, we can reduce the functional to the biinstanton form for the week electron-particle interaction
where r?(r, r') is the potential due to the charge of the bi-instanton, while Wj corresponds to the nonlocal self-interaction
The continual limit of the functional yields the standard bipolaron
functional 48
168
In the self-consistent limit, (.&> is equal to unit. In principle, the investigation of the functional (36) and the related Shrodinger-like equation for bipolaron state requires the relation between oneelectron and dielectron wave functions. Different approaches can be used for this purpose. Direct variational method uses ia special variational form of the wave functions 48 . Another method considers the dielectron correlations by the Kohn-Sham formalism and to use the density functional model for the spin pairing problem. The semicontinual form of the functional was studied in 44 , where the shortrange cavity form and bipolaronic tails of the effective potential were assumed. In our model it is a natural consequence of the microscopical consideration. Recently, the analogous form of the effective potential for the bipolarons has been obtained by path simulations 27 , where the peanut form of the dielectron distribution was found.
The adiabatic limit of functional (39) is well known 4 and is investigated for different types of the bipolaron wave function by the variational method. The problem of the bipolaron stability in the presence of short-range and longrange interactions was studied in 4 9 ' 5 0 . Without considering the problem as a whole, we point out that the presence of the sharp cavity-like short-range potential should increase the stability of the bipolaron formation.
6 6.1
Application of the model The shape of the absorption spectrum
In general, the absorption of light is defined by the dynamics of the electron behavior as well as the medium state. However, we suppose that the time of the electron relaxation exceeds greatly the relaxation time of the environment, and the medium adiabatically follows the electron. The probability of the light absorption W(ui) can be presented as integral over complex variable 9 = it/h
Here u is the frequency of light, while dij is the matrix electron-photon element for the initial -final states transition, and F{, F] are the initial and final timedependent free energy functionals, respectively.
Note that the electron and medium states are strongly correlated, they are connected by common relations determining the free energy surfaces of the 169
initial and final states. The problem of the medium influence on the absorption spectrum reduces to computation of integral (40), with taking into account these correlations between the electron and environment states. The difference between the light absorption problem and the effective potential potential is the complexity of the variable 9. All consideration of Sec. 4 can be repeated, but the analytical continuation is needed for Ff(0), Fj(6) 31. As a result, we can obtain
We will show below, that 2?, are connected with the reorganization energy. Although the general expressions are are easily calculated, we present here the continual limit (x? = 0) °f these expressions to save room
where e = 1 + /, e = 1 + /, and subscripts in averaging brackets denote the averaging over initial and final states, respectively. For a weak electron-particle interaction, the above equations are transformed to the conventional expression of the reorganization energy
where E° and E* is the short-range and long-range parts of the reorganization energy. There are two limiting cases, determined by the value S = ET(3 1. S » 1 (Strong coupling)
170
Integral (40) can be estimated by the saddle point method The main contribution is due to the vicinity of the saddle point and the spectrum has a gaussian form
Here Wf:max,Aj are the maximum frequency and half-width of /-th line.
2. 5 «
1 (Weak coupling).
In this case it is necessary to consider the range of the integration where |7m0| is large. But therein, 8F(9) can be presented as a linear over 0 and the
spectrum has the Lorentz form
6.S
The environment effect on the maximum and width of the absorption line
In general, the absorption spectrum should be calculated numerically by the Shrodinger equation (23). To demonstrate the possibilities of our method, we estimate how the medium state (its temperature, pressure and concentration of added salts) affects the maximum and with of the gaussian absorption spectrum of the solvated electron 18. Note that for the Coulomb potential
Here Uei is the potential electron energy. The latter follows from the virial theorem. For the polaron potential one can obtain that h<jjmax = |(7|f/e||, where
C ~ 1 is the numerical coefficient, which is of the order of unity. The shortranged interaction can slightly reduce the coefficient, C ~ 0.5-=-l. Despite this fact we can put ( 7 = 1 for the estimation. a.) The temperature effect. It should be pointed out that the dielectric constant weakly depends on the temperature. Neglecting this dependence, we can get
171
In this estimation, we also neglect the dependencies of the cavity radius and density on the temperature. Substituting Vo ~ 1 -=- 2 ev, we obtain
which agrees well with the experimental data 1 . For the half-width of the absorption line, we can find the known squared dependence on the temperature A ~ V T. b.) The pressure influence. From these reasons, we can find the pressure dependence of the absorption maximum.
For polar liquids with £ > 10 the last term is less then 10 % and pressure dependence is due to the density effect. The correlation between the absorption maximum and density dependencies was experimentally found in c.) The added salts concentration influence. The presence of added ions causes the Debye screening of the polarization potential. For the dilute electrolyte it can be obtained that
where TV = (e2nf))~1'2 is the Debye radius, and n is the concentration of added salts. It follows
The estimation obtained in 2 0 , correctly determines the sign of the maximum shift due to the added salts concentration. This shift is widely experimentally observed, (see for example 52 ). For the width of the absorption line, it is necessary to consider the next terms of the expansion, which is proportional to n. As a result, we have
here r2p is the size of the first excited electron state. This dependence explains the asymmetry of broadening absorption band. Since the broadening is proportional to the size of the excited state, the broadening is larger for the high-
172
energy transition la —> 3p, which broadens the spectrum in the short-length range. This asymmetry of the broadening was experimentally observed in 5 2 . For dipolar liquids Ufi ~ , where g is the Kirkwood factor. It follows from our consideration that the absorption maximum is proportional to this factor. This correlation has been observed in experiment (see reviews 53 ). More detailed analysis 54 shows that this correlation is sufficiently approximated, it may be due to some effects from our viewpoint. For example, the short- ranged part of electron- particle potential may cause deviations in this dependence.
6.3
Nonadiabatic electron transfer
The reasons of the previous Sec. is valid for the nonadiabatic electron transfer. We should treat an integral similar to (40) and add donor and accepter - electron potential for the electron states localized in donor and accepter respectively. Considering the ground electron states only, we can find in twolevel approximation that the transfer probability is described by the following relation
and saddle point 6* is determined by the extremum of the functional SF(0"). In general, it yields a nonlinear algebraic equation for 6*. It differs from the Marcus electron transfer theory 55 where SF(6*) is a quadratic function over 0. The solution of this algebraic equation should be found numerically. However, we present some interesting considerations following from (52). First, the temperature dependence of the electron transfer rate constant is not generally presented by Arrehnius law. The rate constant does not exponentially increase with temperature but it can decrease for some parameters of the problem. This possibility was pointed out in 5 6 . Second, the derived relations of the electron transfer parameters may be useful when the calculation of the reorganization energy via the conventional electron transfer theory 55 is rather difficult. This is the case for the substances with essential contributions of short- range interactions, when the calculation or measurement of the dielectric constants is very difficult. Third, various crude approximations of the results obtained can estimate the dependence of the transfer probability on the medium parameters. For example, for molecular nonpolar liquids the electron pseudopotential can be approximated as
173
where d is the scattering length. Then, in classic approximation we may get
here re denotes the electron radius, S(k) is the structure factor and L ~
/ dr<£? n (r)<£? (r) is the exchange integral. This relation can be used for the estimation of the temperature, density and scattering length effects on the activation barrier for the nonpolar molecular medium. Fourth, it should be noted that all the results are also valid for inhomogeneous fluids. Particularly, they can be used for the electron transfer near the phase boundary or near electrodes or in the presence of an external field. There are several models of inhomogeneous fluids for which the correlation functions were calculated ; these data can be used to estimate the effect of
the medium heterogeneity on the electron transfer. 7
7.1
Comparison •with other models of the electron state in disordered medium Solid-state systems
The problem of an excess electron in disordered lattice is the simplest example of the electron problem in disordered systems 57. On the other hand, these studies have been initiated by the development of the electron band theory in ordered lattices. However, in the disordered systems the band structure has more complicated character related to new physical phenomena. A more illustrative example is the Anderson localization problem 58. A wide set of other illustrations can be found in 5 9 . The key question studied intensely in this field (see reviews ' ' 8 ' 8 ), is the
calculation of the density of the electron states (DOS) D(E) related to the oneelectron partition Green function averaged over all the impurities configuration
Thus, the problem is reduced to the calculation of the imaginary part of the Green function, and its dependence on the concentration and type of the impurities and on the electron energy. In this sense, this problem for the so-called deep tails region is analogous to our treatment. Leaving aside huge volume of investigations, reviewed i n 6 ' ' 8 ' 9 , we will compare our model with basic theoretical approaches in this field. a.)The optimal- fluctuation method
174
If the inter-impurity distance is less than the electron radius, the (DOS) can be evaluated by the averaged concentration of the impurities p(r). cording to 8
Ac-
Here F(/J) is the entropy of the medium, depending on the impurity concentration p. The variational estimations of this equation yield the extremum of the concentration Po(r)
where 7 is the Lagrange multiplier. It is determined by the fact that E,
If the entropy is proportional to the logarithm of the impurities concentration F oc lnp(r), then from (57-58) we can get equation (23) in the continual limit. This case corresponds to the Poisson range of spectrum. When the deviations from the averaged concentration po = const are rather small, the
gaussian statistics can be used and F oc p(r) — p0. For this case, the electron wave function is determined by the Shrodinger equation of the instanton type (24), which coincides with the equation obtained by Halperin and Lax 80 ' 81 . Note that the results of the approach differ from our model in continual limit only in the Lagrange multiplier 7 (in our model 7 = /3). b.) Path-integral approach. It seems that the path-integral approach was first used in e2 to calculate of the DOS in disordered systems. The method was actively studied in Sayakanit's works 63'84. It is based on the Feynman idea 65 to estimate the action by the reference quadratic action. The statement of the problem is the following. The one-particle Green function is determined by the continual integral
175
Then, the effective action is evaluated according to (15) by the quadratic
reference action
The parameter 7 is fitted to get the optimal estimation of (15). As a result, the nonlinear algebraic equation relating the electron energy E to the parameter 7(xi,ti(r)) is obtained. The analytical calculations were made in (60) where the Mayer function was expanded over the interaction potential till second order. Sa-yakanit used the method to evaluate the D(E) for the screened Coulomb potential. The effective action depending on the two parameters TI , 72 was investigated in 63 . The estimation of D(E) corrected by the higher cumulants was obtained in 6 8 . On the whole for the deep tails region, the method gets the estimations of the DOS close to hose obtained by the optimal-fluctuation theory. The only difference is in the higher energy region where the DOS is overestimated by the path-integral model. It should be pointed out that equation (60) is the continual limit of effective action (14), when Xi = const, Xi = 0> i > 1 and /3 is replaced by it/h. Thus, all evaluations obtained by the path-integral method are similar to our consideration and only differ in the reference action form. The quadratic form of the reference action is chosen in the path-integral model to make analytical evaluation. c.) Replica method. Another approach connected with our treatment is the replica method 6 7 . This approach was first introduced by Edwards and Anderson .In ' the method was used to evaluate the DOS in heavily doped semiconductors. The starting point of the method is the representation of the Green function by the gaussian integrals
here H is the matrix determining the Hamiltonian of the system. The problem is too complicated, since the Hamiltonian depending on the impurities configuration is involved both in the numerator and denominator. However, using the replica trick identity, this equation can be expressed as a multi-dimensional integral. For any functions g(x) and f(x) we can write
176
where x = (z • • • zn) is the replica vector. As a result, the order of averaging and integration over inner degrees of freedom can be changed, and the Green function can be expressed as
To average the Green function, the data on impurity distribution are re-
quired.
These data can be obtained from the density correlation functions
formalism as in our approach. The final result is the expression of the Green function by the density correlation functions and distribution function over replica. For the gaussian statistics, the averaging and evaluation of the multidimensional integral can be made analytically to obtain the instanton type (24) of the equation for the electron wave function. The numerical calculation of (24) was obtained in 71 for different distributions of the correlations. The polaronic effect is neglected in these methods, and the medium density is assumed to be unchanged. The influence of both the polaron effect and the medium disorder was first examined in 2. and bipolaron effect was investigated in 73 The polaron and bipolaron were shown to be localized and more stable due to disorder. The polaron state in disordered medium was investigated 74 by the path-integral method. The polaron effect was discussed in detail in 75 where it was shown to result in an additional peak of the DOS in the deep
tails region. In principle, the consideration of 74 ' 75 coincide with our model. The evaluation of the absorption spectrum presented there for the acoustic polaron is similar to our estimation (40), the difference is only in the quadratic form of the reference action, and simple approximation of potential wells by gaussian-like distribution.
7.2
Liquid systems
a.)Semicontinual models. It has long been known that the disorder of liquid can affect the solvated electron. The earlier attempts to include the disorder into the consideration were based on the semicontinual models, where several statistical quantities of the effective electron- medium interaction were treated as a parameter of the averaging. For example, the number of molecules in the first coordination shell was used as a fluctuating value in 7 6 , while in 77 this role was played by the polarization of the medium, the depth of the cavity was used as the fluctuating quantity in 78 . The modern version of these models includes different types of
177
disorder: the radius of the cavity and the energy of hydrogen bounds in 79, the interaction potential between electron and point charges in 80. All these approaches are the limiting cases of the general formula (23) for the effective potential. b.) Density functional models The continual limit of our model also coincides with the self-trapped electron theory for dense gases and liquids10. It is shown in13 that both the models have an identical starting formulation. The key of the model is that the free energy includes the kinetic energy of the electron, the term proportional to
the chemical potential of the medium, and the term determined by the change in energy of the environment particles due to the excess field induced by the electron
Here fm is the Helmholtz energy of the medium. The variational estimations of the equation yield the extremum of the density po(r) according to (57) and the electron wave function determined by (58).
In these models the pair correlations between the medium molecules are ignored or treated as small corrections to the free energy functional F and to the effective potential Vej , while the pair-wise interaction is represented by
pseudopotential (53), which is generalized to the case of Van der Waals fluid in l0'81. Another generalization includes surface interaction effects. c.) RISM and other related models It was pointed out that the RISM- polaron theory was a starting point of the mean-field models for the solvated electron. This theory uses the direct correlation function cc(r) of the electron instead of the Mayer function. These functions have the same long-range asymptotics, but differ significantly in the short-range region. As was emphasized in , the difference weakly affects the solvated electron state, if the state is rather diffusive. The direct correlation function is connected with the electron-isolated site potential u by some additional relations. This brings up a necessity to compute the additional RISMlike integral equation
where he = gf — 1, gc is the pair electron-solvent distribution function, and w is the response function of the electron
178
To solve the equation, the additional closing approximation of type c(r,<) = exp[—j0u(r)]/(/i s (r) is required. The RISM-polaron model uses the meanspherical approximation. The resulting effective action of the electron is evaluated by reference action of type (60) with a set of fitting parameters 7i, • • • , 7 n The variational estimation leads to the set of the nonlinear algebraic equations connecting fn and the electron state parameters. The problem of the electron state is reduced to the calculation of a nonlinear algebraic set of equations for the parameters -yn (force constants). These equations are calculated by some iterative procedure. Note that the use of the quadratic reference action allows one to estimate the localized electron state as well as quasi- free electron states. The theory of the RISM-polaron model was considered in 28 . Numerical calculations for hard sphere liquids were presented in . The theory results were compared with the simulations in 11 . I n 2 9 , the model was generalized to polarizable fluids, and in 31 to the real time electron behavior. The method was used in 12 to calculate the electron mobility. The RISM- polaron theory was used for the hydrated electron in 17 , for calculation of the electron state 15 and its absorption spectrum 18 in molten salts. The calculation of the effective electron mass was obtained in 14 for nonpolar fluids. The connection between the mean-field treatment and the RISM-polaron theory was investigated in 13 for classical fluids. Another treatment, which interlies between our and the RISM-polaron consideration is the work of Zhu and Cukier 32 ' 33 . The Shrodinger-like equation was also obtained for electron state in nonpolar 32 and polar fluids 33 in the mean-field treatment. In contrast to our model, the effective potential in this equation was expressed by the electron-solvent distribution function. Thus, an additional relation connecting this function and pair-wise potential is required. The Ornstein-Zernike-like integral equation was used as this connection. The calculations based on this model were compared with simulation. d.) Theories of band structure in liquid In contrast to our case, the problem is reduced to the investigation of collective excitations (band structure) for topologically disordered systems, when the excess electron energy is rather close to the energy of bound electrons. This problem was repeatedly considered (see reviews82 and references therein). The result is the calculation of the self-consistent equation for the one-particle Green function connecting the distribution function of the electron density and binary 83 or many-particle 84 function of the spatial distribution of the fluid particles. As is mentioned in , the problem of band structure in liquids relates to the calculation of the energy spectrum for the quantum polarizable fluid 179
8
Conclusion
The mean-field model was developed to treat the localized solvated electron and dielectron states in fluctuative environment. The model is based on the variational estimations of path integrals and was applied to calculate the solvated electron state in fluids, but it can be extended to the self-trapped electron state in solid-state systems for the deep tails region. The developed method allows us to calculate different static and kinetic electron parameters: the electron energy and radius, absorption spectrum, and activation energy of the electron transfer etc... Our treatment makes it possible to consider the phase transition medium effect on the solvated electron. The idea of such a consideration was first given in 86 and was based on the combination of the autolocalization phenomena and renormalization theory. Similar to the estimation of the absorption spectrum and electron transfer,
we can consider how weak external fields affect the electron, and try to calculate electron effective mass, mobility, susceptibility and EPR- signal. We suppose
these calculations can be performed by the Feynman method 87 developed in calculating similar quantities for the polaron in lattice. We believe that the model has only one limitation, i. e. the mean- field approximation. Our model can not be used in systems where the solvated electron forms chemical bounds and where quantum- chemical calculations are necessary. But our model may be extended to mesoscopic systems by the inclusion "ab initio' quantum- calculations in a way similar t o 2 2 . The presented model can be generalized to the wide class of disordered media with different type of molecular disorder, such as polymer solutions, glasses, melting systems and so on.
Appendix 1. Exclusion of quantum degrees of freedom for environment Let us introduce the frequency Fourier components for the dipole momentum and pair-wise potential
180
We can find the extremum of the partition function Z over the Fourier components of the dipole M;(u> n )
As a result, we can obtain the set of linear equations for the Fourier components of the dipole momentum
The above equation describes the equilibrium dipoles induced by the external filed UM(um,Jli). I*8 formal solution is the following
Here I is the unit matrix and symbol denotes the inverse matrix. Using this relation, the quantum internal degrees of freedom can be excluded from the effective action
where A\j are the Fourier elements of matrix
Appendix 2. Expression of the effective electron action by correlation functions The electron- medium interaction can be expressed as a product of the generalized Mayer function of the electron
181
The averaging of electron- medium interaction over fluids configuration yields
where symbol * denotes convolution integration determined by (14), and p1, p2 • • • are the one, two and more irreducible particle distribution functions of the fluid 3 9 . Substituting the definition of the correlation density functions XiX* • ' ' in tni8 expression we can obtain the final form of the electron- medium interaction
where symbols [n] denote the integer part of number n. The latter is the mathematical consequence of the Mayer function definitions (11,13). In spite of the fact that several terms of the expression have long been used in physics of disordered systems (for example, if we keep the first term only, we obtain continual approximation ), the general formula (14) is less studied, the derivation of Eq. (14) was given in 88 , and in 13 was applied to the solvated electron problem. Appendix 3. Estimation of the generalized Mayer function for the
electron Let's us derive formulas for the expressions < /<(r(t)) >, < /j,/y >, (below we dropped the dependence of the Mayer function on the medium particle coordinate in all the formulas ). To simplify our derivations, we take into account only the short-range repulsive part of the potential u, i.e., assume that u = u,. Giving the Taylor expansion of the Mayer function we write (11) in the following form
182
where p(r; r), p(r, r'; r', r) are one, two and more partial density matrices, defined by action So. In general, they are connected with n— partial Green
functions,
here Gn is the n— partial and G(r, r) is the one partial Green function determined by the equation
In the mean- field approximation the n- partial density matrix of the electron can be presented as a one-partial density matrix in n power, that is P(T, r', r', r) = p(r, r)p(r', r') and so on. As a result we obtain
We may derive an expression for < /;, /;- > in the similar way. If U{ < EQ , then we can estimate
Thus, neglecting the off- diagonal elements (77) is the condition of the mean-
field approximation. It is correct if u< < Eo, that is the energy of the ground electron state should be much greater than the interaction energy of the elec-
tron and one particle of the medium. But the electron state can be rather localized if X i f f » 1Taking into account the long-range part ut of potential u, we should add terms proportional to tij, if thy vanishes such in Coulomb and polar liquids, we should also add terms proportional to u*. As a result, we obtain Eq. 19.
References
1. E.J.Hart and M.Anbar, The Hydrated Electron (Wiley, New York, 1970); J.Jortner and N.R.Kestner, Electrons in Fluids (Springer, Berlin, 1973); J.C.Thompson, Electrons in Liquid Ammonia ( Clarendon, Oxford, London, 1976); L.Kevanand B.Webster B., Electron solvent and anion interactions (Amsterdam, Elseiver, 1976).
183
2. L.Kevan and D.-F.Feng, Chem. Rev. 80,1 (1980). 3. E.M.Itskovitch, A.M.Kuznetsov, J.Ulstrup, The Chemical physics of solvatation , p.C., eds. R.R.Dogonadze, E.M.Itskovitch, E.Kalman E., A.A.Kornyshev, J.Ulstrup (Elseiver, Amsterdam, 1988). 4. S.I.Pekar, Research on Electron Theory in Crystals (US AEG, Washington, 1963). 5. J.T.Devreese and F.M.Peeters, Polaron and excitons in Polar Semiconductors and Ionic Crystals (Plenum, New York, 1982). 6. Shklovskiii, B.I., Efros, A.L., Electronic Properties of Doped Semiconductors (Springer, Berlin, 1984). 7. V.L.Bonch-Bruevich, I.P.Zvyagin, P.Kaiper, A.G.Mironov, P.Enderlain, B.Esser, The electron theory of disordered semiconductors (Nauka, Moscow, 1981). 8. I.M.Livshitz, S.A.Gradeskul, L.A.Pastur, Introduction in theory of disordered system (Nauka, Moscow, 1982). 9. P.Van Mieghem, Rev. Mod. Phys. 64, 755 (1992). 10. A.G.Khrapak and I.T.Yakubov, Usp. Fiz. Nauk 129, 45 (1979); J.P.Hernandez, Rev. Mod. Phys. 63, 675 (1991). 11. D.Laria and D.Chandler, /. Chem. Phys. 87, 4088 (1987). 12. D.Hsu and D.Chandler, /. Chem. Phys. 93, 5075 (1990). 13. M.R.Shaw and D.Thirumalai, /. Chem. Phys. 93, 3450 (1990). 14. Y.Singh, Indian J. Chem. A 31, 539 (1992); A.Sethia and Y.Singh, Phys. Rev. B 46, 9959 (1992). 15. G.Malescio and M.Parrinello, Phys. Rev. A 35, 897 (1987); G.Malescio, Nuovo dm. D 13, 1031 (1990). 16. G.Malescio, Phys. Rev. A 36, 5847 (1987); G.Malescio, Mol. Phys. 69, 895 (1990). 17. D.Laria, D.Wu, D.Chandler, /. Chem. Phys. 95, 4444 (1991). 18. G.N.Chuev, JETP 78, 334 (1994). 19. M.F.Deigen, Zh. Eks. Tear. Fiz. 26, 300 (1954); R.R.Dogonadze, Ber. Bunsenges. Physik. Chem. 75, 628 (1971); M.Ya.Ovchinnikova, Theor. Exper. Chem. 17, 507 (1981); A.A.Kornyshev, /. Chem. Soc. Faraday. Trans. II 79, 651 (1983). 20. V.D.Lakhno and O.V.Vasil'ev, Chem. Phys. 153, 147 (1991). 21. F.Fueki, /. Chem. Phys. 44, 3140 (1966); G.Howat and B.C.Webster, /. Phys. Chem. 76, 3714 (1972); M.Weissman and N.V.Cohen, Chem. Phys. Lett. 7, 445 (1970); J.M.Moscowitz, M.Boring, J.H.Wood, J.Chem. Phys. 62, 2254 (1975). 22. M.D.Newton, J. Phys. Chem. 79, 2795 (1975); T.Clark and G.Tiling, J. Amer. Chem. Soc. 113, 1109 (1987); N.R.Kestner and J.Jortner,
184
J. Phys.
Chem.
88,
3818 (1984);
H.F.Hameka, G.W.Robinson,
C.J.Marsden, J. Phys. Chem. 91, 3150 (1987).
23. 24. 25. 26.
L.Kevan, J. Phys. Chem. 85, 1628 (1981). D.A.Copeland, N.R.Kestner, J.Jortner, J. Chem. Phys. 53, 1189 (1970). B.J.Berne and D.Thirumalai, Ann. Rev. Phys. Chem. 37, 401 (1986). A.Wallquist, D.Thirumalai, B.J.Berne, /. Chem. Phys. 85, 1583 (1986); J.Schnitker and P.J.Rossky, /. Chem. Phys. 86, 3462 (1987); M.Parrinello and A.Rahman, /. Chem. Phys. 80, 860 (1984); M.Marchi,
M.Sprik, M.L.Klein, Faraday Discuss.
Chem.
Soc. 85, 373 (1988);
R.N.Barnett, U.Landman, C.L.Clevelend, J.Jortner, J. Chem. Phys. 89,
382 (1988). 27. A.Seloni, M.Parrinello, R.Car, P.Carevali, /. Phys. Chem. 91, 4947 (1987); E.S.Fois, A.Seloni, M.Parrinello, R.Car, /. Phys. Chem. 92, 3268
(1988); E.S.Fois, A.Seloni, M.Parrinello, Phys. Rev. B 39, 4812 (1989); G.J.Martyna, Z.Deng, M.L.Klein, J. Chem. Phys. 98, 555 (1993). 28. D.Chandler, Y.Singh, D.M.Richardson, J. Chem. Phys. 81, 1975 (1984).
29. A.L.Nichols III and D.Chandler, /. Chem. Phys. 84, 398 (1986). 30. A.L.Nichols III and D.Chandler, rtJ. Chem. Phys. 81, 5109 (1984).
31. A.L.Nichols III and D.Chandler, J. Chem. Phys. 87, 6671 (1987). 32. J.Zhu and R.Cukier, J. Chem. Phys. 99, 1288 (1993). 33. 34. 35. 36.
J.Zhu and R.Cukier, J. Chem. Phys. 99, 5384 (1993). R.P.Feynman Statistical Mechanics ( Benjamin, London, 1972). D.Chandler and P.G.Wolynes, J. Chem. Phys. 74, 4078 (1981). L.S.Schulman, Techniques and Applications of Path Integration (NewYork, Wiley, 1981). 37. N.N.Bogolyubov The dynamical theory problems in statistical physics(
Moscow, OGIZ, 1946). 38. H.Frohlich, Advances in Physics 3, 325, (1954). 39. R.Balescu, Equilibrium and Non equilibrium Statistical Mechanics (Wi-
ley, New-York, 1975). 40. J.P.Hansen and I.R.McDonald Theory of the simple liquids (Academic, London, 1986).
41. A.A.Belavin, A.M.Polyakov, Zh.
Eks. Tear.
Fiz. 22, 245 (1975);
R.Rajaraman Solitons ans Instantons ( North-Holland, Amsterdam, 1982).
42. G.N.Chuev, V.D.Lakhno, Phys. Rev. B 48, 12655 (1993). 43. J.Jortner, Mol. Phys. 5, 257 (1962). 44. D.-F.Feng, K.Feuki, L.Kevan, J. Chem. Phys. 58, 3281 (1973). 45. N.Nicoloso, W.Freyland, J. Phys. Chem. 87, 1997 (1987).
46. Y.Nakamura, /. de Phys. IV 1, C5-61 (1991). 185
47. S.L.Lundquist, N.H.March Theory of the Inhomogeneou3 Electron Gas (Plenum, New York, 1983). 48. V.L.Vinetskii, M.S. Giterman, Sov. Phyi. JETP 6, 560 (1958). 49. H.Hiramoto, Y.Toyozawa, J. Phys. Soc. Jpn. 54, 245 (1985). 50. D.Emin, M.S.Hilery, Phys. Rev. B 39, 6575 (1989). 51. F.-Y.Jou, G.R.Freeman, J. Phys. Chem. 81, 909 (1977). 52. I.V.Kreitus, /. Phys. Chem. 89, 1987 (1985); S.Golden, T.R.Tuttle, S.M.Lwenje, J. Phys. Chem. 91, 1360 (1987). 53. A.V.Vannikov, Uspekhi Khimii 44, 1931 (1975); G.R.Freeman, Ann. Rev. Phys. Chem. 34, 463 (1983). 54. J.P.Jay-Gerin, C.Ferradini, Radial. Phys. Chem. 36, 317 (1990). 55. R.A.Marcus, Annu. Rev. Phys. Chem. 15, 155 (1964); R.A.Marcus, N.Sutin, Biochim. Biophys. Ada 811, 265 (1985). 56. A.I.Karasevskii, D.V.Matjushov, A.V.Garodskii, Chem. Phys. 142,1 (1990). 57. N.F.Mott, E.A.Davis Electronic Processes in Non-Crystalline Materials (Clarendon, Oxford, 1979). 58. P.W.Anderson.PfcjM- Rev. 109, 1492 (1958). 59. J.M.Ziman, Models of Disorder (Cambridge University, Cambridge, 1979). 60. B.Halperin, M.Lax, Phys. Rev. 148, 722 (1966). 61. I.M.Livshitz, Zh. Eks. Tear. Fiz. 53, 743 (1967). 62. R.A.Abram, S.F.Edwards, J. Phys. C 5, 1183 (1972). 63. V.Samathiyakanit, J. Phys. C 7, 2849 (1974). 64. V.Samathiyakanit, Phys. Rev. B 19, 2266 (1979); V.Sa-yakanit, H.R.Clyde, Phys. Rev. B 22, 6222 (1980); V.Sa-yakanit, W.Stritrakool, H.R.Clyde, Phys. Rev. B 25, 2776 (1982). 65. R.P.Feynman, Phys. Rev. 97 (1955). 66. E.P.Gross, /. Slat. Phys. 33, 107 (1983). 67. G.Parisi Statistical Field Theory. Frontiers in Physics (Addison-Wesley, Reading, MA, 1988). 68. S.F.Edwards, P.W.Anderson, J. Phys. F 5, 965 (1975). 69. J.L.Cardy, /. Phys. C 11, L321 (1978). 70. E.Brezin, G.Parisi, J. Phys. C 13, L307 (1980). 71. S.John, M.Y.Choy, M.H.Cohen, C.M.Soukoulis, Phys. Rev. B 37, 6963 (1988). 72. M.H.Cohen, E.N.Economou, C.M.Soukoulis, Phys. Rev. Lett. 51, 1202 (1983). 73. M.H.Cohen, E.N.Economou, C.M.Soukoulis, Phys. Rev. 29, 4496 (1984). 186
74. C.H.Grein, S.John, Phys. Rev. B 36, 7457 (1987); C.H.Grein, S.John, Pkys. Rev. B 41, 7641 (1990). 75. S.John, C.H.Grein, Rev. Solid State Sci. 4, 1 (1990). 76. A.Gaathon, J.Jortner Electrons in Fluids eds. J.Jortner, N.R.Kestner ( New-York, Springer-Verlag, 1973). 77. M.Tachiya, Y.Tabata, K.Oshima, /. Phys. Chem. 77, 263 (1973), 78. M.Tachiya, A.Mozumder, /. Chem. Phys. 60, 3037 (1974); K.Funabashi, Y.Maruyama, J. Chem. Phys. 55, 4494 (1971). 79. W.M.Bartczak, M.Hilczer, J.Kroh, J. Phys. Chem. 91, 3894 (1987). 80. M.Hilczer, W.M.Bartczak, /. Phys. Chem. 91, 3834 (1987); M.Hilczer, W.M.Bartczak, J. Phys. Chem. 97, 508 (1993). 81. Miller B.N., Reese T. Phys. Rev. A 39, 4735, (1989). 82. R.M.Stratt, Annu. Rev. Phys. Chem. 41, 175 (1990); R.M.Stratt, Adv. Chem. Phys. 78,1 (1990). 83. B.-C.Xu, R.M.Stratt, J. Chem. Phys. 91, 5613 (1989). 84. D.E.Logan, M.D.Winn, /. Phys. C 21, 5773 (1988). 85. D.Chandler, K.S.Schweizer, P.G.Wolynes, Phys. Rev. Lett. 49, 1100 (1982); J.S.Hoy, K.Olaussen, J. Chem. Phys. 77, 2583 (1982); K.S.Schweizer, D.Chandler, J. Chem. Phys. 78, 4118 (1983). 86. M.A.Krivoglaz, Usp. Fiz. Nauk 111, 619 (1973). 87. R.P.Feynman, R.W.Hellwarth, C.K.Iddings, P.M.Platzman, Phys. Rev. 127, 1104 (1962). 88. R.Frieberg, J.M.Luttinger, Phys. Rev. B 12, 4460 (1975). 89. J.M.Luttinger, Phys. Rev. B 13, 2596 (1976).
187
RESONANCE, LOCALIZED, AND POLARON-TYPE ELECTRON STATES IN ELASTIC MATERIALS WITH TOPOLOGICAL DEFECTS V.A.OSIPOV Joint Institute for fiuclear Research Bogoliubov Laboratory of Theoretical Physics 141980 Dubna, Moscow region, Russia Recent results concerning the problem of an electron localization near linear defects in elastic materials are presented. It is shown that the localized in the xy plane
electronic states may appear for dislocations and topologically stable disclinations. In this case, the motion of an electron along the line of a defect becomes essentially one-dimensional a n d , even in nonpolar crystals, the interaction of the electron with the lattice will result in a localized polaron-type state. In the case of partial disclinations the resonance electronic states are f o u n d to exist. The problem of an electron localization near the charged linear defects is discussed.
1
Introduction
The theoretical and experimental investigations of dislocated crystals show clearly that the presence of dislocations can essentially modify the electronic properties of semiconductors and metals (see, e.g., reviews '2 and the references therein). As is known, the dislocation-induced strains in a crystal result in an additional deformation potential, which affects the properties of free carriers. 3 In particular, there is a dislocation-induced contribution to the resistivity as well as an electron contribution to the braking of dislocations. An important effect, which was confirmed experimentally, is the appearance of two-dimensional electron bound states near the dislocation line. In this case, the electron spectrum turns out to be essentially modified since discrete levels appear. An attractive effect based on the deformation interaction is the self-localization of an electron moving along the dislocation line. 5'8 Namely, it was shown that, instead of a freely moving electron, a polaron-type object with the renormalized mass can arise. It is interesting to note in this connection that the term polaron is often used in reference to any auto-localized electron states in the modern condensed matter physics. Whereas the standard polaron exists in ionic crystals due to the interaction of an electron with the induced polarization of the lattice it is known that, different physical reasons may result in a self-localized electron state. One of them is realized in quasione-dimensional systems where the deformation interaction of an electron with 188
the lattice always results in the localized electron state. Thus, when the motion of an electron in a crystal becomes essentially one-dimensional, one can expect the appearance of a localized polaron-type state. A possible reason for a high anisotropy in a crystal is the strong magnetic field. The formation of the polaron-type states in this case has been studied in 4 . Another typical example of the strong anisotropy in a crystal is the presence of linear defects like dislocations and disclinations which are of interest here. In the present paper we consider three problems related to polaron-type states in materials with linear defects. In Sec. 2 we analyze the problem of an electron localization near the defect line due to the deformation interaction. We restrict consideration to defects which produce strains only in the plane normal to the defect line. In particular, this is true for straight edge dislocations and wedge disclinations. In Sec. 3 we present the self-consistent method for description of polaron-type states near the dislocation line. To illustrate the method, both the edge and screw dislocations are considered. The problem of electron states near charged linear defects is discussed in Sec. 4. Sec. 5 is devoted to concluding comments.
2
Resonance and localized electronic states
Let us study the problem of an electron localization in the plane xy which is taken to be normal to the defect line directed along the z-axis. The interaction of electrons with elastic media is considered in the framework of the deformation-potential theory. The validity of the effective mass approximation is assumed. When the defect fields are considered as the external ones, the two-dimensional stationary Schrodinger equation takes the following general form:
Here m^B is the effective electronic mass tensor, Udef (r) is the deformation potential, r = (x, y), A and B take their values from the set {1, 2}. The electron energy E is measured relative to the bottom of the conduction band in the undeformed crystal. As is seen, the most important deviation from the standard model is the replacement of the conventional derivative 8A = Q jQiA by the covariant one DA • Such a replacement follows from basic assumptions of the gauge model of dislocations and disclinations:8 a) the invariance under the inhomogeneous action of the gauge translational and/or rotational groups and b) the Yang-Mills minimal coupling arguments. 189
Let us restrict our consideration to acoustic deformations. As is known (see, e.g.,9, in general, there are six independent components of the deformation-potential tensor. For cubic crystals they can be reduced to two components, so that the deformation potential takes the form
where EAB
ls
the strain tensor, I\AB — EAB — \f>ABSpEjiBi and G* and „
are the deformation potential constants. In what follows, we will restrict our consideration to isotropic materials where carriers belong to a singlet band
centered at k = 0, k is the wavevector. The only strain component that can affect the energy of such a band is the dilatation, i.e. only the first term in the r.h.s. of Eq. 2 has to be taken into account. Accordingly, the effective mass tensor reduces to a scalar m*. Let us consider first the dislocations. In the presence of dislocations the translational symmetry of elastic media is broken, so that the gauge group is T(3). In this case, 7 the covariant derivative coincides with the conventional
one, so that we arrive at the standard model known in the dislocation theory. 5 For a straight edge dislocation the deformation potential takes the known form
where b is the Burgers vector, cr is the Poisson constant, and > is the polar angle in the plane xy. The analysis of Eq. 1 with the potential Eq. 3 was presented elsewhere (see, e.g.,l |11>8). It was shown that there appear discrete electron levels due to the potential Eq. 3 when Gj > 0. These electron states were experimentally observed in different semiconductors.
In the case of a single screw dislocation the deformation potential is written as
where b = (0, 0, b). As the second order in the levels were found to be For straight wedge
is seen, this potential is weaker than Eq. 3 since it is of Burgers vector. As a result of Eq. 4, discrete electron very close to the bottom of the conduction band. 5 disclinations only rotations in the plane normal to the
defect line are of importance, so that the gauge group is G = SO(2). The covariant derivative takes then the form DA9 = (8A - iWA)9, WA are the gauge fields associated with rotational defects. 7 Thus, for rotational defects a 190
perturbation appears not only in the potential energy due to the deformation potential, but in the kinetic energy as well. Let us consider the linear disclination oriented along the z axis. In this
case, the exact solution of the problem was found to have a vortex-like form. Particularly, the gauge fields WA were found to be: r'12
where v is the Frank index. The exact solution of the problem has been obtained for two cases: a) small i/, this corresponds to the partial topologically unstable disclinations, 13 and b) v = 1, a complete topologically stable disclination. 12 It is important to note that in either case the dilatation was found to depend only on the radial vector r. Then, the stationary Schrodinger Eq. 1 is rewritten in the following form:
Here we have used the ansatz '^(r) = £/• u'E(r) 'j
, j = 0, ±1, ±2, . . .; and
k* = 2m*E/ft . The effective potential in Eq. 6 consists of two parts, the deformation-potential energy and the so-called centrifugal energy. Let us note once more that the most important distinction from the case of dislocations and point impurities is that the interaction of an electron with the gauge field due to disclinations modifies the kinetic term as well. It is clear that this modification will result in a principally new situation for the localization process. As indicated earlier, in dislocated materials the localized electron states always exist (at least with j = 0) provided that the deformation potential is the attractive one. This is not the case for topologically unstable disclinations with the fractional Frank index where even for j = 0 there exists a compensating positive term in Eq. 6 caused by the centrifugal energy. Hence the problem of the electron localization in this case requires an additional analysis. Let us consider first a disclination with a small Frank index in a cylinder with an inner radius Rc and an external radius R, Taking into account the explicit form of the strain tensor (see, e.g. 14 ), the effective potential in Eq. 6 is written as
191
A simple qualitative analysis shows that the potential Eq. 7 is repulsive for i> < 0 (it corresponds to the positive disclination) and, obviously, the discrete electronic states do not appear in this case. On the other hand, for v > 0 (negative disclination) the effective potential Eq. 7 may be attractive. In particular, there is a minimum of Eq. 7 at r^in = (j — i/) 2 A /2m* Dv where D = G
One can see that the depth of the potential well depends on three main parameters: the Frank index j/, constant D (which depends on GjJ), and the effective radius R. The depth of the well increases with any of these parameters. In fact, there are two topical problems. First, one can consider a single small crystallite involving a wedge disclination. In this case electrons are locked in the region R where R characterizes the size of the crystallite. Let us note that the physically interesting region is R ~ 10~ 6 — 10~5 cm which corresponds to the mesoscopic structural level of the plastic deformation. As is known (see, e.g. 14 ), in crystallites of this size the creation of the small-angle (partial) disclinations becomes energetically preferable in comparison with dislocations of the same geometry. Notice that in this case R/RC ~ 102, so that the last term in the square brackets of Eq. 7 is negligible. The boundary condition for Eq. 6 is then u}E(r) \r=R= 0. With this condition, Eq. 6 was studied numerically in 15 . The wavefunction was found to have a resonance-like behaviour. Namely, there exists a distinctive pick in the vicinity of the disclination line. The amplitude of the wavefunction as well as the depth of the lowest electron levels depend essentially on the model parameters. In particular, they decrease rapidly as v decrease. The second problem appears in an investigation of granular materials with intragrain disclinations (e.g., in polycrystals). The distance R can be considered now as the effective radius of elastic strains due to disclinations. It is clear that R coincides with the size of a grain. The deformation potential takes the form Eq. 7 for r < R whereas it tends rapidly to zero at r —> oo. As a result, the effective potential has both a well at Uj(r) < 0 and a barrier at Uj(r) > 0. Accordingly, one of two possibilities for an electron can be realized depending on the model parameters, either a localized electron state with E < 0 or a resonance state with E > 0.
An exact solution for topologically stable linear disclination which corresponds to the complete straight wedge disclination has been found i n 1 2 . It is important to note that this solution contains the information about the core
192
region of the disclination. Namely, we found the explicit form of the deformation potential in a wide space region including the core of a defect. The effective potential was written as follows: 7
where D = 4(A + /i)/3(A + 2/j,), A and fj. are the Lame constants, K(j) = (j — j/) 2 ft /2m*, v can have any integer value. The point r = ro in Eq. 9 turns out to be the boundary between two regions: the core region where deformations are large, and the region beyond the core where deformations decrease slowly and tend to a constant value at r —> oo. Let us analyze the case I = 0. One can see that for j = 1 the potential Eq. 9 becomes attractive, and, therefore, the discrete levels exist at any set of the model parameters. This conclusion was confirmed by numerical calculations in 15. For the state with j = 0 the localization takes place as well but with the lower amplitude. As 7-0 increases, the depth of the potential well rapidly increases and the first electron level becomes remarkably deeper. Conversely, for small ro the well is shallow and the lowest level lies close to the edge of the continuum electron spectrum.
3
Polaron-type electronic states
As is shown in the previous section, the deformation potential arising from the long-range strain field in the presence of dislocations may result in localized electronic states with energies close to the conduction band. In this case, 5 ' 6 the motion of an electron along the dislocation line becomes essentially one-dimensional and, even in nonpolar crystals, the interaction of the electron with the lattice will result in a localized polaron-type state. In this section, we consider this problem in the framework of the gauge theory of dislocations 8 extended in le by including electronic fields. In accordance with the basic assumption of the gauge approach, 8 dislocations can be described by the Lagrangian of elasticity theory that is invariant under the inhomogeneous action of the gauge translational group T(3). The gauge theory is strongly nonlinear in its origin and, which is important, includes the interaction of an electron with the lattice in a self-consistent way. The model Lagrangian takes the following form: le
193
where
describes the elastic properties of the material,
describes the dislocations,
describes the long-wave electronic states within the effective mass approximation, and describes the deformation interaction between electrons and the lattice. The strain tensor is determined to be E^g = B^S^j B3g — 8AB and the distortion tensor is B\ = d^x* +
and
Here Zf = pi, ZA = —ai where pi is the momentum, G = 2G,j, and <Ti is the stress tensor which takes the form
It should be noted that Eqs. 15-17 comprise the self-consistent system of equations for- dislocation dynamics in the presence of electron fields. In general, these equations are strongly nonlinear. The data in the classical theory
194
of elasticity solutions for screw dislocations were found by the linearization procedure. IT It will be shown, however, that the important results concerning the polaron-type states can be obtained directly from Eqs. 15-17 without any
approximations of the linear theory of elasticity. Namely, let us study the electron states in the presence of the static dislocations oriented along the z axis. According to Eq. 17, an electron interacts with the lattice via the deformation potential. As is shown in the previous section, this interaction may result in the states localized in the xy plane, and an infinite number of discrete levels with E < 0 condensed to the point E = 0 may appear. For screw dislocations this effect is shown to take place only in the second order approximation (in elastic displacements) and for j = 0 where
as in the case of edge dislocations the electron localization appears in the first order approximation. As a rule, in the studies of the problem of an electron localization in dislocated crystals, the deformation potential in Eq. 17 is considered as the external one. The electron terms in Eqs. 15 and 16 are ignored and strains EAB caused by a dislocation are only taken into account in Eq. 17. One can
see, however, that in accordance with Eqs. 15-17, the electron fields in turn will affect the strain field. As a result, the local deformation of a lattice due to the electron-phonon interaction will take place. To clarify this point, let us write the strain tensor in the form EAB — EAB "*" ^^AB where SE^B characterizes additional deformations due to the electron-phonon interaction. In the static case, Eq. 15 can be rewritten in the form
Assuming that E^B satisfies Eqs. 15 and 16 when electron fields are absent, we easily find that
where K = \ + -p. The stationary Schrodinger equation takes the form
The electron energy E is measured relative to the bottom of the conduction band in the undeformed crystal. The symmetry of the problem requires the following form for the wave function if>: i/> — iji°(r, 6)f (z) where V"° satisfies the equation
195
In this case, one can obtain from Eqs. 21 and 22 the equation for /(z)
where a = J^^ So \ i/>° |4 rdrdO and e = E° — E. As is seen, € defines the difference between electron energies when the electron-phonon interaction is taken into account. It should be noted that E and E° are negative for localized states. The normalized solution of Eq. 23 reads
Here £ = 2 m * G 2 a / 1 K h 2 , f = A 2 £ 2 /2m*, and z0 is a constant. Note that £ -1 characterizes the region of an electron localization in the z-th direction. Since t > 0 one gets E < E°, i.e. the electron energy decreases in the presence of the electron-phonon interaction. Let us calculate the polaron energy Ep which is the sum of the electron energy E and elastic energy of a deformed crystal
One can see that, as usual, the gain in the electron energy in the presence of the electron-phonon interaction is accompanied by the loss in the lattice energy. Thus, the total gain in the energy due to the formation of the polaron-type state (the binding energy) is E° - Ep = E° - E - |e = |e. Let us emphasize that this result is obtained without any information about the concrete form of a dislocation. However, this information becomes essential if we want to calculate the physical characteristics of the polaron-type state. In addition, two important approximations, the continuum limit and the adiabatic approximation, should be satisfied. One should also be sure that the discrete levels at E < 0 in the electronic spectrum really exist. As an example, let us consider a screw dislocation. Eq. 22 for screw dislocations was studied i n 5 . The results are as follows: in the ground state the wave function i/>° has the form i/>° — AK^T (qr) where K^ (qr) is the Macdonald's function, and q~ is the radius of an electron localization in the xy plane. The lowest energy level E° is found to be \E°\ = h3q*/2m*. The characteristic parameters are approximated as
196
where r0 is the core radius of the dislocation, C ~ ir is a constant, b is the third component of the Burgers vector, and g0 is a constant which characterizes the quadratic in deformations part of the deformation potential. In general, go depends on the elastic moduli up to the fourth order. If we take into account only the second order elastic moduli, we get (see 17 ) go = —G(l — 2
4
Charged defects
In this section we consider charged linear defects. As is known, dislocations in semiconductors and metals turn out to be charged according to the ShockleyRead mechanism. 1 ' Namely, in the presence of an edge dislocation the unpaired bonds lying at the edge of the extra half plane of atoms can accept additional electrons or holes. As a result, the dislocation line will be charged. Ideally, the charge q of the defect could be equal to (L/a)e where L is the length of a defect line. In fact, however, the Coulomb interaction between additional electrons located on dislocation line results in a smaller charge q = fe, where / = a/c is the occupation coefficient and c is the average distance between electrons on the defect. The value of / is different for semiconductors and metals. Typically, / < 0.3. Notice that for two-dimensional systems, dislocations and disclinations are in fact point defects, so that / = 1 (or / = 0 for an empty state). It is clear that the idea of dangling bonds is appropriate for wedge disclinations as well. For example, in the case of 60° wedge disclination, the atom belonging to the disclination line has seven or five bonds instead of six ones in an ideal hexagonal lattice. It is clear that an additional Coulomb interaction between free carriers and charged defects influences the localization process. Besides, the transport properties of a crystal will be modified in the presence of charged linear defects. As is known, any additional charge will be screened by free carriers. In metals the radius of the screening is compatible with the lattice constant. Thus, the charge located on the defect is of little consequence. This is not the case for 197
semiconductors. For example, let us consider an n-type semiconductor. In general, there are two types of the screening: a) by free carriers and b) by ionized impurities. The Poisson equation takes then the following form:
where
consider two limiting cases. At high temperatures (e
AKo(r/rD), K0 is the McDonald's function. In the region 0 < r < TO one can use an approximation
where a = e2//ccT is the ratio of the Coulomb interaction energy of the nearest electrons captured by a defect to their thermal energy, and In 7 = 0.577 is the
Euler constant. At low temperatures (e<(>/T) » 1 at 0 < r < R'. When R* > rD one can neglect the Debye screening and use the approximation ne <; nd. In this case, Eq. 27 takes the form
By using the boundary conditions (d
where •7rjZ*2n(j = //a. As is seen, R* is the distance where the electric field of charged defects is compensated for by the field of ionized impurities. Note that (R*/rD)* = 4e2f/aKT.
198
Let us write the whole effective potential for Eq. 6 at low temperatures
where Uc = etp(r) is the Coulomb potential, Uc — J 2 A 2 /2m*r 2 is the centrifugal energy, J = j for dislocations, and J = j — v for disclinations.
The problem of an electron localization near charged dislocations has been recently reviewed in 2 2 . For shallow electron levels Eq. 6 with Eq. 33 was solved
in 23 within the one-band approximation. When the effective potential well goes deeper, the region of an electron localization becomes smaller reaching a few lattice constants. In the case of deep electron levels it is possible to
use a rough approximation assuming that the Coulomb interaction results in a shift of discrete levels by a middle value £/c(0) ~ ef(c)- M Notice that the Coulomb potential depends on r logarithmically whereas the deformation potential behaves like 1/r.
Let us consider the wedge partial disclinations. It is interesting to note that the deformation potential due to wedge disclinations has a logarithmic
dependence on r, just as the Coulomb potential does. At low temperatures the effective potential is written as
Here D is determined in Eq. 8, and C = e2f/Ka. As is shown in Sec. 2, for
C = 0 and v > 0 the localized and/or resonance electron states can appear in the vicinity of the disclination line. It stands to reason that the Coulomb part
in Eq. 34 will modify this picture. In particular, the depth of the potential well decreases, so that the resonance electron states become preferable. This
fact is of importance for the scattering problem in defect semiconductors at low temperatures. Indeed, as was found in 1 3 , the relaxation time related with disclination vortices consists of two parts: the pure topological one and that
coming from the deformation-potential scattering
Here n&i, denotes the density of disclinations. As a result, the residual resistivity due to disclinations is
199
For deep electron levels the phase SQ is small, so that the deformation-potential scattering is ineffective. By contrast, for resonance electron states sin S0 ~ 1 thus giving the essential increase in resistivity. It is also interesting to note that the effective potential Eq. 34 depends now on the temperature as well as on the density of impurity via rD and R* ( R* depends on T through /). As a result, the discrete electron spectrum can be varied by changing T and rip. In particular, one can change the number of
discrete localized or resonance levels and even forbid their appearance. 5
Conclusion
In conclusion, we presented some results related to electronic states in materials with topological defects. As is shown, the presence of dislocations and disclinations can essentially modify the electron spectrum. This finding is of importance in the studies of different phenomena in defect crystals. It is expected that the presence of discrete electron levels due to defects will essentially affect the thermodynamic and kinetic properties of materials. In particular, the shallow electron levels in semiconductors influence the dislocation-induced resistivity at low temperatures. The deep electron levels serve as the centers of recombination for excess carriers. The polaron-type states would be of importance in dielectrics. Let us note that the existence of discrete electron levels
in dislocated semiconductors is well established whereas the results concerning the polaron-type states still remain to be confirmed experimentally. Disclinations are of importance in highly deformed materials that involve large mutual rotations of neighbouring elements. The experimental study of electrical properties of such materials has much potential for yielding information about disclinations. Notice also that in modern theoretical models disclinations are considered as important constituents of different disordered materials, like amorphous bodies, metal glasses, polymers, etc. However, up to now there is no direct evidence in favor of disclination structure of these materials. That is why any theoretical predictions concerning the electronic properties, especially specific transport properties, of disclinated materials are of considerable interest. The careful measurements of transport properties of these materials would be a good test to clarify both the availability of disclinations and their density.
Acknowledgements Part of this work has been financially supported by the grant from Russian Fund of Fundamental Research No 94-02-05867. The research described in this 200
publication was made possible in part by grant N RFR300 from the International Science Foundation.
References 1. R. Labusch and W. Schroter in Dislocations in Sohds, ed. F.R.N. Nabarro (North-Holland, Amsterdam, 5, 1980). 2. Yu.A. Osip'yan et al, Adv. in Phys. 35, 115 (1986). 3. R. Landauer, Phys. Rev. 94, 1386 (1954). 4. L.S. Kukushkin, Sov. Phys.- JETP Lett. 7, 251 (1968). 5. A.M. Kosevich, Sov. J. Low Temp. Phys. 4, 902 (1978). 6. V.P. Voronov and A.M. Kosevich, Sov. J. Low Temp. Phys. 6, 371 (1980). 7. V.A.Osipov, J. Phys. A 26, 1375 (1993). 8. A. Kadic and D.G.B. Edelen in Lecture Notes in Physics, ed. H. Araki, et al (Springer, Berlin, 174, 1983). 9. B.K. Ridley, Quantum Processes in Semiconductors (Clarendon, Oxford, 1982). 10. V. Celli et al, Phys. Rev. Lett. 8, 96 (1962). 11. P.R. Emtage, Phys. Rev. 163, 865 (1967). 12. V.A.Osipov, Phys. Lett. A 164, 327 (1992). 13. V.A.Osipov, Phys. Lett. A 193, 97 (1994). 14. V.I. Vladimirov and A.E. Romanov, Disclinations in Crystals (Nauka, Leningrad, 1986) [in Russian]. 15. V.A.Osipov and S.E.Krasavin, /. Phys.: Cond. Matt. 7, L95 (1995). 16. V.A.Osipov, Physica A 175, 369 (1991). 17. V.A.Osipov, /. Phys. A 24, 3237 (1991). 18. V.M. Nabutovsky and B.Ya. Shapiro, Sov. Phys.- JETP 75, 948 (1978). 19. W. Shockley, Phys. Rev. 91, 228 (1953). 20. W.T. Read, Phil. Mag. 45, 775 (1954). 21. R.A. Vardanyan, Sov. Phys.- JETP 73, 2313 (1977). 22. V.B.Shikin and Yu. V.Shikina, Usp. Fiz. Nauk 165, 887 (1995). 23. V.L. Bonch-Bruevich and V.B. Glasko, Sov.Phys.-Solid State 3, 36 (1961).
201
BIPOLARONS IN ANISOTROPIC CRYSTALS N.I. KASHIRINA, E.V. MOZDOR, E.A. PASHITSKIJ, and V.I. SHEKA Institute of Physics of Semiconductors of National Ukraine Academy of Science, Kiev 25202, Ukraine We consider a simple bipolaron treatment for anisptropic crystals in the strongcoupling limit. We take into account anisotropy of effective band masses and dielectric constants of the crystals.
The bipolaron theory was first studied in1 by the adiabatic approximation for the case of strong coupling of electron-phonon interaction and isotropic continuum medium. Living aside the history of the bipolaron study presented in 2 ' 3 , we set forth the problem by the traditional Frohlich Hamiltonian which is generalized to the case of anisotropic two-atomic crystals. ' 5 For the bipolaron problem the corresponding Hamiltonian has the form
where
is the kinetic energy of the electron with coordinates r^ and T y ,
is the potential energy of electrons. The phonon frequencies with normal coordinates die and a£ polarization vectors BJ (k) satisfy the dispersion equation
202
where
while the polarization vectors follow the condition
It should be noted that Hamiltonian Eq. 1 describes the bipolaron states in crystals with the uniaxial anisotropy. In this case, if we reduce the tensors of effective mass and dielectric permittivity to the main axes, notations e0 and £00, mxx — myv = mi, mzz = m\\, txx = evy = £j_, f-zz — f\\ can be introduced. In contrast to works, 6 ' 7 where the limiting cases of low-dimensional systems
are considered, we study crystals with arbitrary anisotropy of effective band mass and dielectric permittivity (in the general case we also take into account their dispersion in time and space). In the weak-coupling limit there are no localized bipolaron states in the
continuum medium, for which is used Hamiltonian Eq. 1 holds. Therefore, the bipolaron formation is more probable in the strong-coupling limit. In the latter case, the phonon operators in Eq. 1 can be replaced by c-numbers /jy, so that the average two-participle energy takes on the form
The minimization of Eq. 6 over fu. and the summation over the polarization vectors (which is similar to minimization of the polaron energy, developed in 5 ) result in the adiabatic Pekar bipolaron functional
where
203
Good results were obtained in8 by the minimization of functional Eq. 7 for isotropic systems, where the trial wave functions took on the form
where r12 = |rj — r 2 |, and r t ,r2 are the coordinates of the first and second electron, respectively. The account of the term with multiplier /3r12 enables us to consider the electron correlation effects. But in the case of anisotropic crystals, the use of these trial wave functions leads to cumbersome mathematical problems in the evaluation of integrals for potential energy V\ 2 and for the term corresponding to the electron-phonon interaction. However, using Gaussian trial wave functions, we make the analytical calculations in the explicit form. In this paper, we use the trial wave function, which is axially symmetric and has the form
In this relation, the parameters are chosen to account the symmetry or antisymmetry with respect to exchange of the electron coordinates. The interference terms 2o2;.zi.z2 and 26 2 jp 1 /9 2 in the exponential index take into account the correlation effects. To calculate the minimum of functional Eq. 7, it is convenient to make the scale transformation r =>• Ar, where A is the scale multiplier, which can be found by the virial theorem, A = -(V"12 + Vef)/2T, and Ex = -("^12 + V'cf)/^TN, where N is the normalization multiplier. All averaged quantities in the functional Eq. 7 can be easily calculated by the wave functions Eq. 9. The minimization of E^ with respect to the parameters Cj,mj,bij (i = 1,2,3; j = 1,N) was made by the method of fast descent and it was applied to calculations of the energy of para- and ortohelium ground states. The calculations for N = 7 yield the following results
E0 = -2.8722374 for the parahelium, and El = -2.1183365 for the ortohelium, which coincide with the best calculations of corresponding quantities Eq. 9 in all significant digits. This application shows that our approximation of trial functions by Eq. 7 is rather convenient to minimize the bipolaron functional. We note that the choice of the wave functions in form Eq. 7 enables us to perform easily the symmetrization procedure (or antisymmetrization) over the electron coordinates. For instance, for the triplet states the coefficients Cj 204
must differ in sign for the terms different in the exchanging electron coordinates TJ and r 2 , while for the singlet state they coincide in sign. The following syrnmetrization (the choice of equivalent variational parameters corresponding to the terms, which are different in exchanging electron coordinates) can be made by the minimization of the functional. The possibility of calculation of singlet-triplet splitting for two-electron system by the wave functions chosen as a linear combination of gaussian orbitales enables us to estimate the dependence of the exchange energy on the inter-ion distance without ignoring the electron correlation effects which can play a key role. Figure 1. shows the dependence of the bipolaron ground state energy (line 1) and the polaron one (line 2) on the ratio of effective masses ( in logarithmic scale) for the cases of anisotropy "light axis" ("IX/TOII > 1) and "light plane" ( m _L/ m i| < 1> when EOO /CQ =>• 0- The corresponding bipolaron bound energy is plotted in Fig. 2. The bipolaron and polaron energies of the ground state are shown to decrease if the anisotropy rises, while the polaron bound energy increases (especially for crystals with quasi-one-dimensional anisotropy). In the isotropic case, we result in the lower energy of the bipolaron ground state and the larger bound energy AW (AW/2/ p = 0.253 as t^ /e 0 —» 0) than the best results calculated for the strong-coupling case, with the use of the conventional trial wave functions Eq. 8. The minimization of the polaron functional by trial function involving seven gaussian orbitales exactly leads to the energy evaluated in10 in the strong-coupling limit. We use this value to evaluate the bipolaron bound energy. The situation is similar to the case of isotropic crystals. In comparison with isotropic case, the regions of parameter 77 = £oo/ £ 0i where bipolaron exists, are slightly extended. For example, if m±/mii = 32 the bipolaron is stable as 0 < 7; < r)c = 0.150, and if rn±/m\\ = 1/64 it yields TJC = 0.156. In the isotropic case T)c = 0.140. We note that the calculation of the bipolaron energy in the ground state for anisotropic crystals is similar to the calculations for the crystals with anisotropic effective masses, which we study. Moreover, in the limit too/Co —* 0) using the scale transformation of electron coordinates (2^2 —h V e J_/£||Zi,2 we can reduce the problem to the case under consideration, where effective masses are anisotropic, while the dielectric constants are isotropic. The new effective masses are connected with the initial dielectric constants by the relation
To discuss the bipolaron formation in the crystals where high-temperature
conductivity is observed, we consider a concrete example La2CuOt, in which there is a great anisotropy of effective masses and dielectric permittivities. 205
This crystal is representative to illustrate the possibility for calculation of the bipolaron bound energy in the crystals where the anisotropy of effective masses and dielectric constants occur coincidentally. In our assumption, the data on static and high-frequency dielectric constants are only required, these constants are well-known from experiments 11 , eoo = 4, e0 = 50 in the plane of layers CuO?, and e0 = 23 in the perpendicular direction.
Figure 1: Energy of the polaron ground state (line 1), and bipolaron ground sate (line 2). m* = min{m^y, mg }
Figure 2: Bipolaron bound energy for the aniaotropic cases: "light axis" (m* = "iz)i snd "light plane" (m* = m ^ y ) In the case of isotropical effective masses, the bipolaron bound energy 206
is 15.6% of the polaron energy, but in the limiting case of the maximum anisotropy of effective masses mxy "^ rnz > this quantity is 25.2% of the polaron energy. Therefore the bipolaron formation should be quite possible in crystals with high-temperature conductivity, especially, if we take into account that these crystals are the systems with "light plane" anisotropy. Figure 3 shows the bipolaron bound energies expressed in the units of the
double polaron energy at various values of the ratio mXY/mz. The lines 1, 2, 3, 4, 5 correspond to this ratio at 1, 2 ~ 2 , 2 ~ 4 , 2~ 1 5 , 2~ 2 0 , respectively.
Figure 3. Dependence of the bipolaron bound energy on parameter v = oo/eo- The lines 1, 2 , 3 , 4 , 5 , correspond to the parameter equal
f
to 1, 2 ~ 2 , 2 ~ 4 , 2~ 1 5 , 2~ 2 0 , respectively, mXY/Tnz = 1-
It is seen that the region of bipolaron existence extends, and the bound energy increases as the anisotropy of crystals grows. We point out that we can 207
consider the two-dimensional systems by this method using scale transformation Ip(2D)/Ip(3D) = 2/3(3ir/4) 2 (where Ip(3D) and / P (2Z>) are the polaron energies in the two- and three-dimensional systems, respectively). Similar relations can be derived for the bipolaron ground state energy and, consequently, for bipolaron bound energy. The possibility of this scale transformation to the two-dimensional systems is revealed in1 , and numerically tested by us for the case of the maximum anisotropy of effective masses. We note that we consider electron-optical-phonon interaction. Using the continuum approximation, we can easily express the bound energy for electronacoustic-phonon interactions. But the absence of reliable data on the tensor of the deformation potential makes it impossible to calculate the bound bipolaron energy with account of the condenson effect. We note that this effect can play a key role due to the increasing bound energy in this case. The bound energy of autolocalized state may change significantly as a result of the interaction with plasma vibrations of charge carriers in the conduction and valence bands, or in the systems of movable ions for crystals with ion conductivity. But in addition to the effect caused by screening Coulomb interactions of electrons and ion vacancies, for bounded systems, such as polarons, bipolarons, F, F', F2 -centers there is a screening of electron interactions with optical phonons. Therefore, without detailed calculations of the bound energy for the polaron and bipolaron surrounded by plasmons we cannot estimate the plasmon effect on these energies.
References 1. S.I. Pekar, Studies in the Electronic Theory of Crystls (Gostekhizdat, Moscow, 1951) [in Russian]. 2. G. Verbist et al, Phys. Scripta 39, 66 (1991). 3. V.I. Vinitskij et al, Ukr. Fiz. Zh. 37, 76 (1992) ( U k r . Phys. J.). 4. E. Bergman, J. Phys. and Chem. Sol. 35, 1433 (1974).
5. V.I. Orbukh V.I. and V.I. Sheka, Ukr. Fiz. Zh 30, 771 (1985) (Ukr. Phys. J.). 6. J.T. Devreese, Phys. Scripta 39, 309 (1989). 7. E.P. Pokatilov et al, Phys. Status Solidi (b) 171, 437 (1992). 8. S.G. Suprun and B.Ya. Moizhes, Fiz. Tverd. Tela 24 (5), 1571 (1982) (Sov. Phys. Solid State Phys.). 9. P. Gombash, Many Particle Problem in Quantum Mechanics (Inost. Lit., Moscow, 1953) [in Russian]. 208
10. 11. 12. 13.
S.J. Miyake, J. Phys. Soc. Jpn 38, 187 (1972). D. Reagor at al, Phys. Rev. Lett. 62, 2048 (1989). G. Verbist et at, Phys. Rev. B 43, 2712 (1991). V.D.Lakhno, Phys. Rev. B 46, 7519 (1992).
209
SU(2) PATH INTEGRAL FOR THE HEISENBERG FERROMAGNETIC E.A.KOCHETOV Joint Institute for Nuclear Research Bogoliubov Laboratory of Theoretical Physics 141980 Dubna, Moscow region, Russia We have derived the expression for the partition f u n c t i o n in terms of the SU(2) (spin) path integrals for a q u a n t u m Heisenberg ferromagnetic. We evaluate the excitation s p e c t r u m in the proximity of the mean field, and calculate corrections of effective interactions to the temperature Green functions, reproducing the results of the spin diagram technique at finite temperatures.
1
Introduction
One of the interesting problems of solid-state physics is the problem of the quantum Heisenberg ferromagnetic at finite temperatures. The partition function of the quantum Heisenberg ferromagnetic is expressed as
where Sj are the SU(2) generators of dimensionality 2S + 1, and Jij is the symmetrical matrix of exchange interaction (Ju = 0), which is positively determined and can be written as1
where T stands for a chronologically ordered product and the summation over all repeating indices is assumed. In Eq. 1 we imply the summation over all repeating indices. Deriving Eq. 1, we used the Hubbard-Stratonovich equality. The partition function of noninteracting spines in the external fluctuating field
210
V"(*) can be written as the SU(2) path integral
where we put •iji = tl>x + iipv, V1 — V"i ~ *V"y > and V1^ = *l>x • The functional measure in Eq. 2 is formally defined as a product of infinite number of SU(2) invariant point measures
According to 2 , integral Eq. 2 is evaluated by the substitution
where
As a result, we obtain
where the function z = u/v depends on if> via the Riccatti-type equation
Thus, we take the partition function of the form
211
Relations Eqs. 5-6 serve the starting point for expansion in the vicinity of the mean field point.
2
Mean-field theory
We choose a steady mean field in the z direction as «&j = (0, 0, *j). equation of the stationary phase
The
(subscript |0 denotes value at a stationary point) yields
where b(x) = SBs(Sx), and BS(X) is the Brillouin function. Deducing Eq. 7, we assume that JQ = S • J\j and $i = $ due to spatial homogeneity of the system. Expanding S(T/J) at the point $ up to the second order terms in the deviation 77 = 1/1 — ijj |Q, we get
where the inverse effective longitudinal and transverse interactions are
212
Taking variational derivatives of Eq. 4, we find
that results in the temperature Green function of noninteracting spines with the Hamiltonian H = H0 = $£s' Z ) ; n$ = (e?* - I)" 1 .
Turning to Eq. 10, we obtain
which can be rewritten in the energy-momentum space as
Due to Eq. 12, the Larkin equation for the total propagator
yields
where (G0(un) = b/(iu>n + $),
un = 2irn//3), and
is the energy of elementary spin-wave excitations. We can easily calculate the path integral Eq. 8 contributing to the partition function due to Gaussian fluctuations in the vicinity of the mean field. The result coincides with estimations given in 1 . We note that our approach does not use the transformation from the spin representation to the Bose (Fermi) oscillatory representation of generators of SU(1) group, which essentially changes the initial Hamiltonian and complicates the calculations. Besides the proposed method enables us to find the expansions of the Green function in the vicinity of the mean field. 213
3
Green functions
For clarity we consider the transverse Green function
where G^. (T, c r \ i p ( t ) ) is the Green function of noninteracting spines in the external fluctuating field ip(t), which is calculated as a functional derivative of Eq. 3
where we introduce the notation
To calculate corrections to the Green function, caused by the Gaussian
fluctuations over the mean field, we should expand S(ifi) in Eq. 16 up to the
second order terms as in Eq. 8. Then, expanding the functional derivative Eq. 17 in powers of 77 and evaluating new integrals, we can easily solve the problem. As is seen from Eq. 8, the corrections increase as powers of the effective interactions Jefj-tin H Jejj.tr. At the stationary point, the required functional derivatives of zt(s) with respect to V'j(t) are calculated with the use of Eq. 4. The zero approximation yields
214
as it must. We extract the first order correction caused by the account of the forth term in Eq. 17,
Adding up these two terms, we find Eq. 14, and, consequently, we obtain the spectrum of spin-wave excitations. We should account Eq. 18 separately, since it has the zero order with respect to inverse effective volume v of the interaction, while the other first order terms with respect to Jcj j present one-loop corrections and have the first order with respect to v. Since the parameter v is small in the mean-field theory, the Green function Eq. 14 should be treated as the zero approximation of the total Green function calculated in the mean-field theory. It is clear, that relation Eq. 17 includes all diagrams of conventional spin diagram techniques at finite temperatures, where, however, interaction lines are replaced by effective lines.
4
Conclusion
Here we have developed a new method based on the spin path integration to calculate thermodynamic functions in the Heisenberg model. The method can be immediately generalized to more complicated groups describing dynamic
symmetries of complex systems. For instance, the supergroup J7(2|l) is a group of dynamic symmetry for the Hubbard model (t —- J model), 4 and the superpath integral constructed by this group can be used to study the model.
References 1. S. Leibler and H. Orland, Ann. Phys. 132, 277 (1981).
2. E.A. Kochetov, Phys. Lett. A 180, 383 (1993). 3. E. Manousakis, Rev. Mod. Phys. 63, 1 (1991). 4. P.B. Wiegmann, Phys. Rev. Lett. 60, 821 (1988).
215
LOW-TEMPERATURE ELECTRON MOBILITY OF ACOUSTICAL POLARON B.A.KOTIYA", and V.F.LOS b a
Georgia Technical University, Institute of Fundamental Research,
Tbilisi 380075, Republic Georgia Institute of Metal Physics of National Ukraine Academy of Science, Kiev S5214&, Ukraine Abstract
We derive a new exact evolution equations for correlation functions of a subsystem interacting with the Bose field. The electroconductivity for an electron in the acoustic polaron model is treated by the the linear Green-Kubo theory. For this model we evaluate the expressions for electroconductivity and low-temperature electron mobility, and calculate the correction to electron mobility caused by the account of electron-phonon correlations at the initial moment. 1
Introduction
In recent years, an interest to the problems of noneqiulibrium statistical mechanics has increased, for instance, one of the intensely studied fields
is kinetics of dynamic subsystem interacting with a thermostat (Bose field). 1 ' 2 We note that one of the important problems of condensed physics and quantum field theory is the problem of interaction between a particle and a quantum Bose field. The electron-phonon system is the simplest example of this electron-phonon interaction in a crystal. ~~ At present, a large number of problems concerning the behavior of slow carriers in polar semiconductors and ionic crystals are treated by the polaron model. The main attention has been focused on the theories of linear electron transfer processes in polar crystals. The investigation of polaron (electron) transfer and evaluation of polaron kinetic characteristics is one of pressing problems of the modern theory of large polarons. ~~ In this paper, we derive an exact quantum evolution equation (generalized quantum kinetic equation) for the two-time equilibrium correlation function of an (electron) subsystem interacting with the Bose (phonon) field. Our study is based on the Liouville superoperator formalism and the method of projective operators. Here we consider the acoustical polaron model in the continual limit. Using the second order perturbation theory for the case of weak electron-phonon interaction and low temperatures, we derive the equations with except phonon amplitudes and the expressions for the velocity-velocity correlation functions of the electron.
216
We also deduce the relations for the time and frequency of relaxation velocity and momentum of the electron. Using the above approach for the acoustic polaron at low temperatures, we evaluate the kinetic transfer coefficients (mobility, electroconductivity) for the electron. 2
Equation for correlation functions
We consider the dynamic subsystem S interacting with the Bose (phonon) field (thermostat) S. We write the Hamiltonian of the whole system (5 + E) in the form
where H,,Hz,Hi are the Hamiltonians of the subsystem, the (Bose) phonon field, and electron-phonon interaction, respectively.
where u(k) > 0 is the frequency of the quantum Bose (phonon) field, 6^, &J are the creation and annihilation Bose operators for the state, which is described by quantum numbers k, and Cfc(s), C^(s) are the operators related to subsystem 5. For the model of the continual acoustic large polaron, 10 we have
where p, k, and r are the electron momentum, wave vector, and coordinate, respectively, and m is the electron effective mass, V, is the sound velocity, a is the dimensionless electron-phonon coupling constant, V is the volume of a crystal, D is the constant of the deformation potential, and p is the density of the crystal. The main goal of our study is to calculate the low-temperature electron mobility (conductivity) for the acoustic polaron, which is described in the weak-coupling case (a < 1) by the Hamiltonian of Frohlich type. 1 " 3 Using the superLiuville formalism and the method of projective operators, we can derive an exact equation of quantum evolution for the two-time-dependent equilibrium correlation function (A,B,(t)), (here A,, and B, are the arbitrary operators related to the subsystem 5),
217
which describes the subsystem interacting with the Bose (phonon) field (see, for instance 11 ). This equation is written as
In this relation, as is follows from the definition, (A,B,(t)) = Z~1Sp(eKp[-/3H]A,ex.f[iLt]B,), L is the Liouville superoperator acting on the arbitrary operator C according to the rule
where L = L, + LS + Li, and L,, LE , Li correspond to the Hamiltonians of the subsystem and phonon field, determined by Eq. 1, /3 = 1/fcsT is the inverse temperature, Z = 5pexp[—flH] is the partition function of the whole system (S + E); P • • • = Sj>£(p% • • •) is the projection operator for averaging over the thermostat states.
In the last equation Z is the partition function of thermostat S, it can be presented as
In Eq. 4 we introduce the mass and integral superoperators, which are determined by
where H0 = H, + H-E • We note that Eq. 4 for the correlation function (A,B,(t)) includes the operators (amplitudes) of thermostat (Bose field). We suppose that the characteristic times are rather different in the system (S + E) due to weak interaction of Hi(Li), i.e. Trci ^
from Eq. 4, we obtain the Markovian kinetic equation for the correlation function (A,B,(t)). This equation with account of the second order perturbations for electron-phonon interactions (with respect to V^ for the considered acoustic polaron) can be written at rather long time (t ^> t 0 )
where a < 1, J\f^(fl) = (exp[/3fi.w(k)] — 1)-1 is the average number of
completing phonons in the state k, r(z) = exp [i/hT(P)Z] r = exp[-i/h]T(P)Z,
and the relation [E, C"] ± ^ (k) =
EC - exp[±/3hu>(k)]CE holds for operators E and C'. In accordance with Eq. 5, the equation for two-time dependent equilibrium function, which is the z component of the velocity operator of the electron (v^v^^)), takes the form
where a < 1, T >• ta , and VM = dl(p)/dpf
— p^/m.
We can derive the evolution equation for correlation function (v,(i)v,) in a similar way. Actually, using the relation
we can easily obtain the equation for correlation function (7) by the substitution t —» — t into Eq. 6. We note that the evolution of the initial
electron-phonon correlations is described by the last terms of Eqs. 5 and 6.
219
3
Calculation of low-temperature electron mobility
To estimate the low-temperature electron mobility (conductivity) for the case of the acoustic polaron, we use the Green-Kubo theory of linear response. 1 -1 According to the Green-Kubo theory, we express the dissipative part of the electroconductivity tensor as
where w is the frequency of the external field, while *n(l) is the symmetrized equilibrium time-correlation function of z, which is the component of the current operator
Let us consider the one-band approximation for the electron in the isotropic case when a^v = crxf =
where a < 1,7 3> 1. In this relation P is the dimensionless electron momentum, P = p/V,m. For the relaxation frequency (characteristic time) of the electron velocity we have
where dQ^ = sin &d®dip, P cos * = sin © cos ipP^ + sin & sin tpPy + cos ©P., and 0, ip are the spherical angles of the wave vector k, while $ is the angle between the vectors k and P.
220
We consider a slowly moving electron at P ^ 1 (p <; V,m, v ^ •»,). According to Eq. 11 we evaluate the relaxation frequency of the electron velocity
Here we consider the weak coupling case a < 1. Therefore, for a slowly moving electron (when the electron velocity is much less than the sound velocity in the crystal v
where a < l ; 7 ^> 1, t > Trel . Substituting these asymptotic expressions Eq. 13 into Eqs. 8 and 9, OW»(:tO) = ne3(v*vm(±t)), we arrive at the expression for low-temperature electric conductivity
where a < 1; 7 ^ 1, w <; mV^/hf, and n is the concentration of electrons. For low-temperature dc electron mobility of the acoustic polaron we have
where
A/i =
32a(^v
exp[47] sin 2 (32c*7 exp[-47]) ,
a < 1,
•y > 1 .
This correction A/i to the low-temperature electron mobility ;*o is caused by the fact that the model takes account of the electron-phonon correlations at the initial moment. References 1. N.N. Bogolubov, Preprint JINR E 17-1822 (JINR, Dubna, 19T8). 2. N . N . Bogolubov and N . N . J r . Bogolubov, Tear. Mat. Fiz. 43 (1), 3 (1980). 3. N.N. Jr. Bogolubov et al, Tear. Mat. Fiz. 67 (1), 115 (1985).
221
4. V.F.Los and A.G. Maitinenko, Physisca A 138, 518 (1986). 5. K.Rodiiges and V.K. Fedyanin, Element. Chast. i Yadra 15 (40), 871 (1984) (Sov. J. Part, and Nucl.). 6. J. Appel in Polarons, ed. Ya.A. Firsov (Nauka, Moscow, 1975) [in
Russian]. 7. F.M.Peetres and J.T. Devieese, Solid State Physics 38, 81 (1984).
8. R. Zwanzig, Phyaica 30 (6), 1109 (1964). 9. V.F.Los, Doklady AN SSSR 240 (5), 1078 (1978) (Sov. Phys. Dokl.). 10. F.M.Peettes and J.T. Devieese, Phys. Rev. B 32 (6), 3515 (1985). 11. B.A. Kotiya and V.F.Los in Polarons and Applications, ed.
V.D. Lakhno. (Wiley, Chishestei, 1994). 12. R. Kubo, /. Phys. Soc. Jpn. 19 (6), 570 (1957). 13. R. Balescu, Equilibrium and Nonequilibrium Statistical Mechanics
(Wiley, New-York, 1975). 14. A. Isihara, Statistical Physics (Academic Press, New-York, 1971).
222
POLARON EFFECTS ON MAGNITORESISTANCE IN QUASI-TWO-DIMENSIONAL CONDUCTORS L.S.
KUKUSHKIN
Institute for Low Temerature Physics and Engineering of National Ukraine Academy of Science, Kharkov 310164, Ukraine We evaluate a low-temperature magnitoresistance of a quasi-two-dimensional conductor in the strong transverse magnetic field, where the scattering of carriers is mainly caused by phonons. The m u l t i p h o n o n effects play a key role, since the electron-phonon interaction is supposed to be strong. We estimate the t e m p e r a t u r e and field dependence of the conductivity.
The discovery of the quantum Hall effect and high-temperature conductivity have aroused interest in the study of properties of quasi-two-dimensional conductors. As a result, a large number of works devoted to the study of two-dimensional conductors have come out. In the present communication, we calculate the conductivity of a two-dimensional electron gas in strong magnetic field H, which is perpendicular to the plane of the electron motion. Our calculations are based on the method developed in . We consider the case when the scattering of carriers is primarily caused by electron-phonon interaction. We note that the works concerned with the problem (see Refs. 2 and 3) are usually based on the second order perturbation theory and assume the diffusion mechanism for the centers of the cyclotron motion of orbits. The direct consideration of multiphonon effects resulting from strong electron-phonon interactions is a complicated problem, because there is no gradient invariance in this approximation, since the centers of the cyclotron orbits are numbered by a quantum number depending on calibration of the vector potential. But at low temperatures T
223
We suppose that a carrier moves in the plane (z, y), and a magnetic field is directed along the axis Z, then the one-electron Hamiltonian of the system can be presented in terms of the Landau calibration as
where po
=
(e^0~
*8 the magnetic length (A = c = 1); n^ = <** a^ is
the number operator of phonons with frequency u>(k). Here, to simplify the consideration, we assume that the effective mass is isotropic, and phonons are three-dimensional (as it takes place in heterostructures and bicrystals) or twodimensional (as in thin films and for the electrons above the helium surface). The method developed in is used in an unitary transformation of Hamiltonian Eq. 1. After this transformation Hamiltonian Eq. 1 takes the form
where c* and c are the creation and annihilation operators of the electron oscillator c* = 2~l/2(popy + ip^y). We note that the unitary operators
may be reduced to unit ones by some general assumptions. However, in the calculation of low-temperature conductivity with due regard for the multiphonon effect the difference between these operators is essential. The behavior of the system is determined by the operators
where a^ = 2~1'2 po(kx+iky ). The first multiplier in Eq. 4 restricts the phonon contribution for wave vectors k > kjj = Ip^ 1, and corresponds to an increased electron-phonon coupling as the magnetic field increases, especially for the interaction of deformation type. Therefore, there are no operators px and z in the transformed Hamiltonian. Moreover, they do not arise in the expression for conductivity tensor derived by 224
the Kubo relation, if using the unitary operator U, we transform the expression for the spur in the Kubo relation. Then, the diagonal components {o\ f c } of nondegenerate electron gas are presented as
where i/e = v — it(e —> +0); p@ = Z~l exp(—/3H), Z is the partition function, /0 = T~l, C±(T) = C* (T)±C(T) are the operators in the Heisenberg presentation, and n is the number of electrons. In Eq. 5, index (2) denotes that pp and c-k(-r) are determined by Hamiltonian Eq. 2. For degenerate gas, the conductivity (2^
is described by expression Eq. 5, with np\
(2^
replaced by Fermi function pp
depending on H^3'. To derive the Hall conductivity we should change one of
the multipliers c^ 2 ) (<) into »c(_2)(<). The direct calculation of Eq. 5 is still a complicated problem, but the evaluation of conductivity by Eq. 5 is possible, if we use the adiabatic approximation to separate the electron and phonon subsystems. The criterion of this approximation is Qc ^> uj, where LJ is the characteristic phonon frequency, for example,
for acoustic phonons it is skjj
, then the adiabatic approximation
results in the inequality T
The method proposed in5 allows to calculate (ryy(Q,T) taking into account the two-phonon effects in the low-temperature limit, when ky < kjf • In this case, operators F^ can be expanded in power series of y and py, including quadratic 225
terms. As a result, the adiabatic functions V'JvCj/i Q) (where N is the number of Landau level), calculated at fixed medium coordinates Q = {Q-^}, are the wave functions of oscillator with frequency and equilibrium position depending on Q. The power dependence of cryv on temperature is related to the intraband processes, and leads only to the dependence of matrix elements on Q
In optics this dependence results from deviations on the Frank-Condon principle. If the number of carriers is less than the multiplicity of degenerated
Landau levels, (n
It should be mentioned that the type of electron-phonon interaction and the form of phonon dispersion affect strongly the temperature and field dependence of tryy- For acoustic phonons with ai(ifc) = sk and deformation type of their interaction with electrons, when Vk = iD(hk/psl0SY , (where /0 is the localization radius along the magnetic field, p is the medium density, and D is the constant of the deformation potential), the conductivity is expressed as
Here fl lo = 5/66 is the Bernoulli number. Expression Eq. 9 differs from the corresponding expression in6 by multiplier ~ (T/fi,.) 2 , which is related to the adiabatic parameter. It seems to result from the fact that the processes of phonon entertainment are not adequately taken into account in . (The total entertainment is easily seen to 226
lead to (7j fy (0,T) = 0, and crxy = ne/H). For three-dimensional phonons at deformation interaction crvv ~ D*T12H~8. The latter also differs from the corresponding expression in6 by the same multiplier. At piezoelectric interaction,
when Vk ~ k~3'2 and phonons are acoustic,
itable, moreover, the state extends as the magnetic field rises. Account of this phenomenon should have effect on the behavior of
IB n
°t fulfilled, then we cannot expand operators
{Fjj.} in power series of y and py, because of the effect of the cut-off multiplier exp( — fc 2 />o/4) in Ffc. Hence, the phonons with k w ku play a key role in the conductivity instead of thermal phonons, that results in the distinct (more weak) dependence of conductivity on temperature and magnetic field. At kf > KH, to find the adiabatic wave function, it is necessary to solve the differencedifferential equation, since {F^} are the translation operators
We point out that, for the second order of {Vj.} this effect results in the expres-
sion of
227
case
Expression Eq. 10 can be used as a notation of the effective time of carrier free path. This fact justifies our calculations of the effects of other contributions in the longitudinal conductivity, which are not the main ones. The similar situation takes place for the diagonal components of the electroresistance tensor, since in the twc-dime,nsional case
and Hall conductivity at Slcr* ^ 1 is not dissipative and greatly exceeds the diagonal conductivity.
References 1. L.S. Kukushkin, Zh.
Eks. Tear.
Fiz.
78, 1020 (1980) (Sov.
Phys.
JETP). 2. 3. 4. 5. 6.
R. Kubo et al, Solid State Phys. 17, 269 (1965). F. Stern et al, Rev. Mod. Phys. 2, 54 (1982). R. Kubo and Y. Toyozawa, Progr. Theor. Phys. 13, 160 (1955). L.S. Kukushkin et al, Fiz. Niz. Tern. 20 (5), (1994) (Low. Tern. Phys.). Yu.A. Bychkov et al, Pis. Zh. Eks. Tear. Fiz. 34, 496 (1981) (Sov.
Phys. JETP Lett.). 7. V.B.Shikin and Yu.P.Monarkha, Two-dimensional Charged Systems in Helium (Nauka, Moscow, 1989) [in Russian].
8. Yu.P.Monarkha, Fiz. Niz. Tern. 19, 235 (1993) (Low. Tern. Phys.). 9. E.D.Vol and L.S. Kukushkin, Fiz. Niz. Tern. 9, 97 (1983) (Low. Tern. Phys.).
228
NUMERICAL INVESTIGATION OF A QUANTUMFIELD MODEL FOR STRONG-COUPLED BINUCLEON I.V. AMIRKHANOV, I.V.PUZYNIN, T.P. PUZYNINA, T.A. STRIZH, E.V. ZEMLYANAYA Joint Institute for Nuclear Research, Laboratory of Computing Techniques and Automation 141980 Dubna, Moscow Region, Russia V.D.LAKHNO Department of Quantum-mechanical Systems, Institute of Mathematical Problems of Biology, RAS Pushchino, Moscow Region, 14SS9S, Russia We consider a quantumfield model for binucleon in the case of nucleon spot interaction with scalar and pseudoscalar meson fields. The nonrelativistic problem of two nucleons is shown to reduce to the one-particle problem. We derive in the strong coupling limit the nonlinear equations describing two nucleons in the meson field. We also discuss the applicability of the model to obtain the deuteron and bineuteron characteristics.
1
Introduction
The calculation of binucleon states has been in focus of many investigations.1'2 This problem is of interest, since its solution can yield direct data on nuclear forces. In this paper we try to calculate a bound state of two nucleons in a meson field by using the consistent translation-invariant theory. It is well known that in the strong coupling limit the field can be considered as classical in zero approximation.3 Therefore the interaction $(PI, r 2 ) of two particles with the classical component of the field should be written as
On the other hand, since the medium is supposed to be isotropic and homogeneous, two quantum particles should interact as
This contradiction is absent in the quantumfield theory of weak and intermediate coupling which describes the particles interacting by exchange of field quanta. The use of the perturbation theory in this limit leads to a translation - invariant expression Eq. 2 for the interaction between two particles. 229
The aim of this paper is to consider the translation-invariant quantumfield model for the interaction between nucleons and pion field in the strong coupling limit and to study nonlinear differential equations for nucleons, arising in this limit.
2
Translation-invariant theory in the strong coupling limit
According to current concepts, the main contribution to the long - range nucleon-nucleon interaction is caused by ir-mesons. In the nonrelativistic approximation the pion-nucleon interaction is determined by the pseudoscalar coupling Hamiltonian:
where fj, is the pion mass, / is the interaction constant, a and r are spin and isospin operators, respectively. Expanding the pion field
where ro is an arbitrary reference point, we can present Eq. 3 as
K
where
The total Hamiltonian for two nucleons in a meson field is expressed as
where
Jip corresponds to two free nucleons with coordinates r t and r 2 , m is the nucleon mass,
230
where i = 1,2 corresponds to the first and the second nucleons, respectively;
is a Hamiltonian of a free meson field. We can use the results of strong coupling theory ' for Hamiltonian Eqs. 9 and 10. According to this theory, coordinates TO and r divide, and the Schrodinger
equation in zero approximation is written for the binucleon wave function i/>o as
where r = ri — r 2 is relative coordinate, and
where
where a is the unit vector. The Bogolubov - Tyablikov method was used in5 for the case of scalar interaction. According to 5 , the potential of scalar interaction He is written as
where g is the constant of the interaction with the scalar field. We take into account both contributions into the interaction, putting in Eq. 13
231
3
Differential equations for binucleon.
The Schrodinger equation for binucleon with potential Eqs. 16—18 can be presented in the variational form as the functional:
where
Taking into account the functional Eq. 20 to be minimum for functions V"( r ) with the normalizing conditions
we get the set of equations
Using the notation
and assuming A = 1, we can present Eq. 23 as
Energy levels e and wave functions i/> of the binucleon can be determined by solving the system Eq. 24 for functions V>(r), which satisfy the normalizing 232
condition Eq. 22 and are limited within 0 < r < oo, with boundary conditions
(Here r = V z2 + J/2 + z2.) Then for the binucleon radius R and the quadrupole moment Q the following formulas can be used
Using in Eqs. 24 the dimensionless variables
we get the following system of equations (bar mark for vector f was omitted)
where
Using the last equation we can determine -y, for example, supposing ro = 10- 13 m. Taking the change of variables Eq. 28 into account, we rewrite
Eqs. 22, 26, 27 and 19 as
233
Then we use the relations between the variables:
Here the expressions in the brackets are dimensionless variables. The set of differential equations for bipolaron Eq. 29 is the nonlinear threeparameter eigenvalue problem. Numerical integration of Eq. 29 was performed in5 in the limit kfic —* <x>. Thus, the solution of Eq. 29 for finite hue is an actual problem.
4
Statement of the boundary value problem for the set of equations.
We discuss some alternatives for the statement of the boundary problem for the system Eq. 29 in the special case of axial symmetry, i.e. $(p,z,
we use the finite interval: —ZM <
z
<
z
u,
^ < p < PM for cylindrical
coordinates and 0 < r < rM for the spherical coordinates.
4-1
Alternative I.
We are supposed to solve the system Eq. 29 in cylindrical coordinates using the boundary conditions determined on the rectangle —ZM^KfGD (see Fig. 1), 234
that is
on the interval — ZM < z < ZM and
for 0 < p < PM • Here
where r = v p2 + z2 , and
235
Thus, the determination of the physical parameters of the deuteron is to solve the spectral problem for the set of partial differential Eq. 29 with the boundary conditions (Eqs. 39 and 40) and the normalizing conditions Eq. 31. Taking into account the symmetry of the problem, the boundary conditions can be determined on rectangle QzMCpM (the hatched region in Fig. 1).
Figure 1.
Then on the rectangle border OzM, zMC and ptfC *ne boundary conditions are as in Eqs. 39, 40 and for the border OpM they take the form
4-S.
Alternative II.
We can formulate the problem in terms of the ordinary differential equations.
236
The solution of the system Eq. 29 in cylindrical coordinates we present as
where function {nn>$n} is determined as the solution of a spectral problem
Here AfM
is determined from the following relations:
The function (p^ ig replaced by V\A or VZA , an i n } or {MS™, *3n}- In special case we can use only one set of functions {$„}. Substituting the expansion Eq. 42 into the system Eq. 29, we have:
where
237
Functions yn, Vj,-, V3j satisfy the following boundary conditions
where
The coefficients A( — ZM ), -<4.i(—ZM) and A$(—ZM ) are determined in the same way.
Using the problem symmetry, as in the first alternative, the set Eq. 44 can be solved on the interval 0 < 0 < ZM • The boundary conditions are as in
Eqs.
49 and 50 is replaced by:
238
The normalizing conditions for Eq. 44 take the form:
if we use boundary conditions Eqs. 49 and 50 and the form
for the boundary conditions Eq. 49 and 51. Expressions Eqs. 19, 26 and 27 for the expansion Eq. 42 can be written as:
where
4-3
Alternative HI.
Let us present the solution of the system Eq. 29 as an expansion in spherical coordinates:
239
where
pi are the Legendre polynoms. Substituting the expansion Eq. 55 into the system Eq. 29, we obtain
where
240
Taking into account the asymptotic behavior of solutions
the system of Eq. 57 will be solved with the following boundary conditions:
Thus, physical characteristics of binucleon can be found by the solution to the spectral problem Eq. 57 with boundary conditions Eq. 58 and normalizing conditions
Taking into account Eq. 55, the expressions Eqs. 19, 26 and 27 used for calculating the physical parameters take the form
where
241
5
Numerical methods and numerical results.
We investigated numerically the set of Eq. 44. Calculations of the basic functions {$„} = {*i n } = {*3n}, n = l,L, L = 1,2,3,4,5, for a given pM were performed. The code SLIPl 6 was used to solve numerically the problem
Eq. 43 for the base functions {$ n } with normalizing conditions Eq. 43. The algorithm suggested in was used to solve the problem Eq. 44. This algorithm is based on the combination of continuous analog of the Newton method (CANM) and the continuation method and allows us to reduce the solution to simple linear problems. The following iterative scheme was used: • The problem Eq. 43 was solved with the boundary conditions
for the set of basic functions.
• Using the initial approximation of yn(z) and a set of basic functions calculated at the first step, we evaluate the right side of the second and the third equations of system Eq. 44 J\j and J3j, (j = n) using Eqs. 4648. Then we determine functions V\j and V3j (j = n) as the solution of a linear boundary problem by the sweep method. • Substituting the approximations of Vjy and V$j into the first equations of system Eq. 62, we have an eigenvalue problem for the set of linear differential equations, which can be solved by CANM in the modified code based on the program START.9 The function exp(— \z\) was used as the initial approximation for j/ n (*) together with the solutions obtained for the other set of parameters. The number of Newtonian iterations do not usually exceed 10-13. • Solutions yn from the previous step are used to calculate Jij and J3j in the next iteration and so on. The process of iterations stops when the
solutions obtained after two sequential iterations would coincide with each other with a given accuracy e. Actually 5-8 iterations are enough to reach the accuracy e = 10~3. 242
• Parameter Ap is calculated with the help of the solutions yn. Then the problem Eq. 43 is solved again with the corresponding boundary conditions, and a new set of basic functions is determined. The solution to the problem Eq. 44 are calculated again for this set by the above algorithm.
The process will continue until stabilization, i.e. until the newly-obtained solutions and the value APM coincide with the preceding ones with the accuracy given a priori. Actually 3-4 iteration steps are required to reach the stabilization. The discrete approximation of the problem was made on a h-step uniform
grid {zi} with the help of the difference schemes of the second order convergence O(A 2 ). The calculations on a sequence of the twice compressible grids
were shown in Table 1 and confirm the 2nd order of the convergence( see Table 1). Table 1:
Table 2 lists the results obtained on a sequence of expanding intervals [—ZM,ZM] showing convergence on parameter ZM• Table 2:
It should be noted that in practice we use a finite number L of functions
On the other hand, the big value of PM 243
requires a large number of basic functions. For the optimal solving, the number of basic functions L and pM can be determined by the numerical experiments. These results are given in Tables 3 and 4. In the top raw of each column the value A, normalization Nn and amplitude <j>n(z) for n = 1, L, obtained for the boundary conditions Eq. 63 are given. The same values after the solution stabilization for parameter AfM are given in the bottom row. Table 3:
Table 4:
It can be seen from Tables 3 and 4 that pM = 5 — 7 can be considered as optimal for this set of parameters, since in this case the solution does not depend on PM • In addition, the initial norms and the expansion amplitudes do not differ much from the limiting values. With increasing or decreasing PM the dependence of the results on this parameter makes stronger. Let us note that the contribution on the norm of each next expansion component decreases. This allows us to assume that when increasing a number L of functions $ n , one can obtain a solution with a required accuracy.
Functions yn(n = 1, ..., 5), V\ n (n = 1, ..., 5), V-^n(n = 1, ..., 5) for this set of parameters are presented in Fig. 2. The function f>(p, z) (see Eq. 42) is 245
presented in Fig.3.
Figure 2.
246
Figure 3.
247
Figure 4.
248
Now let us turn our attention to experimental data. According to10 — radius R = 1.963 • 10~13 m, — quadrupole momentum Q = 2.86 • 10~
cm ,
— binding energy 2.246eV. It follows from Eqs. 52-54 that at r0 = 10~13 the calculated dimensionless values correspond to the experimental ones if the following relations are fulfilled:
According to Eq. 30, the parameters of the model must be as follows
and the B and Nc may be arbitrary. However, we failed to get satisfactory results for these parameters. Considering ki as arbitrary parameter and fitting its value by the mentioned relations Eq. 64, we obtained the best results at
Hence, we have
Turning to dimension values with the help of relations Eqs. 30-36 we find
The same calculations were done for the problem statement by alternative
III. For the parameters presented in Tables 1-4 at / = 0, • • • , 5 the achieved eigenvalue equals A = 0.67 and is in satisfactory agreement with the results presented in Tables 3 and 4 for alternative II. For more precise calculations the increasing number of basic functions are needed in both statements.
6
Conclusion
Calculating numerically the nonlinear Eq. 41 for two nucleons in a scalar and pseudoscalar field, we evaluate binding energy, effective radius, and deuteron quadrupole momentum, which fit the experimental data Eq. 82, if we assume the effective mass of scalar meson to be near the mass of pseudoscalar meson
fj,a w 140 Mev and k* = k% = 1. At the same time, this fitting yields the coupling constant g and / one order less than the experimental data determined 249
by the local linear model of nucleon-meson interactions.
But our model is
nonlocal and nonlinear.
Note that we investigated only the simplest case of spot interaction. The nucleon formfactor included in the equations can change significantly the final results. The other important trend is to generalize the theory to the case of intermediate coupling constants and take into account relativistic effects.
Authors are grateful to the Russian Foundation for Basic Research (Grant N 94-01-01119) for financial support.
References 1. T. Ericson and W. Weise in Fiona and Nuclei (Claredon Press, Oxford, 1988). 2. G.E. Brown and A.D. Jackson in The Nucleon - Nucleon Interaction (North - Holland, Amsterdam, 1976).
3. 4. 5. 6.
E.P. Gross, Ann. Phys. 19, 219 (1962). N . N . Bogolubov, Ukr. Mat. Zh. 2, 3(1950). V.D.Lakhno, Tear. Mat. Fiz. 100, 219 (1994). I.V. Puzynin and T.P. Puzynina in Algorithms and programs for solving
some problems of physics (KFKI-74-34, Budapest (93-111), 1974). 7. I.V. Amirkhanov et al, in Polaron and Applications, ed. V.D. Lakhno,
(Willey, Chishester, 1994). 8. A.P. Zhidkov et al, PEPAN 4 127, (1973); I.V. Puzynin et al, Zh. Vych. Math. Fiz. 32, 846 (1992).
9. Yu.S. Smirnov, JINR Russian]. 10. K.N. Mukhin,
Communications (Ell-88-912, Dubna, 1988) [in
Experimental
Nuclear Physics (Energoizdat,
1983) [in Riussian]. 11. V.D. Lakhno Chapter 3 in this book.
250
Moscow,