Exciton condensation in quantum wells

The theory of exciton condensation is given in two-dimensional systems under suggestion that condensation occurs in really space and condensed phase arises as a result of an attractive interaction between excitons. Due to the finite value of exciton lifetime the sizes of exciton condensed phase r...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Физика низких температур
Datum:2006
1. Verfasser: Sugakov, V.I.
Format: Artikel
Sprache:English
Veröffentlicht: Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України 2006
Schlagworte:
Online Zugang:https://nasplib.isofts.kiev.ua/handle/123456789/120895
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Назва журналу:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Zitieren:Exciton condensation in quantum wells / V.I. Sugakov // Физика низких температур. — 2006. — Т. 32, № 11. — С. 1449–1457. — Бібліогр.: 30 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id nasplib_isofts_kiev_ua-123456789-120895
record_format dspace
spelling Sugakov, V.I.
2017-06-13T09:23:36Z
2017-06-13T09:23:36Z
2006
Exciton condensation in quantum wells / V.I. Sugakov // Физика низких температур. — 2006. — Т. 32, № 11. — С. 1449–1457. — Бібліогр.: 30 назв. — англ.
0132-6414
PACS: 73.21.Fg, 78.67.De
https://nasplib.isofts.kiev.ua/handle/123456789/120895
The theory of exciton condensation is given in two-dimensional systems under suggestion that condensation occurs in really space and condensed phase arises as a result of an attractive interaction between excitons. Due to the finite value of exciton lifetime the sizes of exciton condensed phase regions are restricted and the condensed phase appears in a form of system of islands amid exciton gas. The joint solution of kinetic equations for island size and exciton diffusion equation in the space between islands has been obtained. The theory is applied to explanation of experimental manifestation of condensed phase in quantum wells and also to explanation of the periodical fragmentation, which was observed in luminescence spectrum from a ring around a laser spot in a crystal with double quantum wells. For such explanations the theory does not require the exciton Bose–Einstein condensation.
en
Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
Физика низких температур
Quantum Effects in Semiconductors and Insulating Materials
Exciton condensation in quantum wells
Article
published earlier
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
title Exciton condensation in quantum wells
spellingShingle Exciton condensation in quantum wells
Sugakov, V.I.
Quantum Effects in Semiconductors and Insulating Materials
title_short Exciton condensation in quantum wells
title_full Exciton condensation in quantum wells
title_fullStr Exciton condensation in quantum wells
title_full_unstemmed Exciton condensation in quantum wells
title_sort exciton condensation in quantum wells
author Sugakov, V.I.
author_facet Sugakov, V.I.
topic Quantum Effects in Semiconductors and Insulating Materials
topic_facet Quantum Effects in Semiconductors and Insulating Materials
publishDate 2006
language English
container_title Физика низких температур
publisher Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
format Article
description The theory of exciton condensation is given in two-dimensional systems under suggestion that condensation occurs in really space and condensed phase arises as a result of an attractive interaction between excitons. Due to the finite value of exciton lifetime the sizes of exciton condensed phase regions are restricted and the condensed phase appears in a form of system of islands amid exciton gas. The joint solution of kinetic equations for island size and exciton diffusion equation in the space between islands has been obtained. The theory is applied to explanation of experimental manifestation of condensed phase in quantum wells and also to explanation of the periodical fragmentation, which was observed in luminescence spectrum from a ring around a laser spot in a crystal with double quantum wells. For such explanations the theory does not require the exciton Bose–Einstein condensation.
issn 0132-6414
url https://nasplib.isofts.kiev.ua/handle/123456789/120895
citation_txt Exciton condensation in quantum wells / V.I. Sugakov // Физика низких температур. — 2006. — Т. 32, № 11. — С. 1449–1457. — Бібліогр.: 30 назв. — англ.
work_keys_str_mv AT sugakovvi excitoncondensationinquantumwells
first_indexed 2025-11-26T15:19:29Z
last_indexed 2025-11-26T15:19:29Z
_version_ 1850626180998234112
fulltext Fizika Nizkikh Temperatur, 2006, v. 32, No. 11, p. 1449–1457 Exciton condensation in quantum wells V.I. Sugakov Institute for Nuclear Research, 47 Nauki Ave., Kiev, 03680, Ukraine E-mail: sugakov@i.com.ua Received May 24, 2006, revised July 7, 2006 The theory of exciton condensation is given in two-dimensional systems under suggestion that condensation occurs in really space and condensed phase arises as a result of an attractive interac- tion between excitons. Due to the finite value of exciton lifetime the sizes of exciton condensed phase regions are restricted and the condensed phase appears in a form of system of islands amid exciton gas. The joint solution of kinetic equations for island size and exciton diffusion equation in the space between islands has been obtained. The theory is applied to explanation of experimental manifestation of condensed phase in quantum wells and also to explanation of the periodical frag- mentation, which was observed in luminescence spectrum from a ring around a laser spot in a crys- tal with double quantum wells. For such explanations the theory does not require the exciton Bose–Einstein condensation. PACS: 73.21.Fg, 78.67.De Keywords: two-dimensional systems, exciton condensation theory, luminescence spectrum periodical fragmentation. 1. Introduction There were many attempts to observe exciton con- densed phase. Stirring problem is the problem of Bose–Einstain condensation of excitons [1,2] with its interesting physical consequences: superfluidity, nar- rowing of optical bands and others. But usually elec- tron-electron correlation energies in condensed phase are larger than energy of interaction between excitons and consequently excitons lose their individuality and well known and well good studied state of elec- tron-hole liquid is created [3]. The creation of elec- tron-hole liquid phase was observed and studied in ger- manium, silicon and other materials [4–7]. The crystal Cu2O was considered for some time as per- spective system for exciton Bose–Einstein condensation because the exciton in Cu2O has small radius and large value of binding energy. But due to Auger decay pro- cesses it is difficult to create in this crystal the exciton density that satisfies Bose–Einstein condencation crite- rion [8]. So the observation of Bose–Einstain condensa- tion is not conformed in bulk materials. Recently intense investigations of excitons under in- tensive irradiation were carried out for indirect excitons in quantum wells (QWs) [9–16]. Indirect excitons in double QW structures with electrons and holes driven to the separate wells by an electric field are long living par- ticles. The enhanced lifetime makes it possible to create large concentrations of excitons and study effects of the exciton-exciton interaction and a possibility of an exci- ton condensation, in particular. The authors [9] ob- served an appearance of narrow line in luminescence spectra in the crystal with double quantum well at some threshold value of pumping. In AlGaAs and InGaAs based structures, a nontrivial feature in the photo- luminescence spectra of indirect excitons was observed [11–16]. A spot of the laser excitation was reported to be surrounded by a concentric bright ring separated from the laser spot by an annular dark intermediate re- gion. The distance between the ring and the spot grew with increasing the intensity of the pumping and rea- ched up to hundreds �m exceeding by much the exciton diffusion path. At low temperatures, the external ring fragmented into a structure with a strongly evaluated periodicity on a macroscopic scale [11,12,15]. 2. Model of system. Basic equations In the paper a model of exciton condensation is de- veloped which is based on the traditional phase transi- tion theory, generalized for the case of a nonequilib- © V.I. Sugakov, 2006 rium system, in which particles (excitons) have finite lifetime. The condensed exciton phase is assumed to exist, and its density is higher than that of the gaseous phase. In this case, the condensation takes place in real space, not in momentum space at k � 0, i.e., the condensed phase is not the Bose–Einstein condensate. The reasons for the application of the traditional phase transition theory for the description of the exciton system are following: Real QWs comprise defects of different types. Even at perfect boundaries in semiconductor alloys like Ga Al As1�x x there are always composition fluctuations either in the well or in the barrier, or both. The authors of [17] showed that the disorder sufficiently toughens the criteria for the appearance of quantum coherence ef- fects. Disorder leads to broadening of luminescence line. For the line, which the authors [9,10,15] connect with the condensed phase, the width is Q � 0.1–1 meV. Ac- cording to the calculation [17], forQ � 0 5. meV the con- densed phase can appear only for the densities more than 2 5 1011 2. � �cm . In the case of the smallest disorder ob- served in experiment (Q � 01. meV) at the exciton den- sity 5 1010 2� �cm , the Kosterlitz–Thouless transition happens at T � 0 4. K, which is significantly lower than the critical temperature in the absence of disorder. The understanding of the toughening of the quantum criteria is possible from the following argumentation. The pres- ence of disorder gives rise to the exciton scattering with wave vector changing in the well plane, and this causes broadening of the exciton emission line width Q. The free path time � can be estimated from the relation �Q � �. Hence, for the line width Q � 0.1–1 meV the time of the free path equals to 0.7–7 ps, and for the aver- age exciton velocity corresponding to T � 1 K the free pass length v� � 7–70 nm. Therefore, for the exciton density of 1010–1011 cm–2, the wave function loses its coherence due to the scattering on defects at the dis- tances smaller than the distance between excitons. This makes doubtful the explanation by means of cooperative quantum effects the phenomena observed at T � 1 K [11,15]. So, on the one hand, the condition for exciton collective quantum effects is facilitated in comparison with the condition for the system of alkaline metal atoms, in which Bose condensation was observed [18], because the exciton mass is sufficiently smaller than atom mass. On the other hand, these conditions become tougher, because of the presence of random forces due to such external to the system factors as the spatial disor- der, the interaction with phonons, etc. For the formation of the condensed state in real space, the attractive interaction between excitons is required. At the same time, owing to the presence of a dipole moment, directed perpendicularly to the well plane, repulsive interaction is realized for indirect excitons at large distances between them. However, at close distances, the exchange interaction between electrons and between holes has the essential contribu- tion. It is known that in a single QW bases on AlGaAs, a biexciton exists with a relatively large binding energy (23%) of the exciton binding energy [19]. The exchange interaction between the electrons and between holes of two excitons, located in the same well, takes place in the double QW, too. Therefore, one has to expect an attraction between excitons lo- cated at close distances (several exciton radii) at not very large distances between QWs, when the dipole moments of excitons are not very large. Thus, the cal- culations [20,21] of multiexciton system of indirect excitons taking into consideration exchange energy, Van-der-Waals energy and correlation energy showed that the energy of the system as the function of the density has a minimum at nonzero exciton density for not very separated quantum wells (d a� 11. , where d is the distance between the wells, a is the exciton radius in the well), when the dipole—dipole repulsion is not too large (see Fig. 2 in the [21]). In this case, the ap- pearance of excitonic liquid state become possible in a double QW. The authors also showed that in these conditions, the excitonic liquid is more advantageous energy-wise than the formation of biexcitons. Using the parameters of GaAs crystal and the results of pa- per [21], it is can be shown that the binding energy per exciton is negative (i.e., there is an attractive interaction) at the exciton density of order of (1.1–2.5)�1011 cm–2. This value is the exciton density in the condensed phase. In the model considered here, the system consists of both condensed and gas phase regions. At critical parameters of system the fraction of condensed phase islands is small. So, the critical density is determined by the mean value of the den- sity, which is of several times smaller than the con- densed phase density. Therefore, the theory predicts the realistic value of the density of several units of 1010 cm–2, at which the condensation can occur. Besides the phase of a dielectric liquid phases with different polarization can be realized in double QWs. The phase diagrams of these phases are presented in [22]. The sizes of condensed phase regions are restricted both from above and from below. The radius of new phase nucleus must be larger some threshold which is determined by a surface energy. A presence of the sur- face energy determines the shape of condensed phase, in two-dimensional case the condensed phase has disk-like shape. A restriction of the size from above caused by nonequilibrium system which is connected with exciton creation and exciton decay. In some range of pumping a number of excitons created by 1450 Fizika Nizkikh Temperatur, 2006, v. 32, No. 11 V.I. Sugakov light in unit time inside of condensed phase region is less than the number of excitons which disappear. In this case the steady state of the system is possible if losses of excitons inside of condensed phase compen- sates by an exciton inflow from surrounding region of exciton gas phase. Such compensation is not possible at large radius of exciton condensed phase. As the re- sult in the two-dimensional case the condensed phase must exist as a system of islands similar to elec- tron-hole drops in bulk semiconductors. The condition of free energy minimum is satisfied if the islands have the form of disks. In studied system the islands should be placed on the ring where the exciton density has maximum. Since islands of condensed phases takes excitons from the same source, two islands can not be situated close one to another. Moreover, the distance between islands can not be large because in this case the exciton density between them becomes greater than a critical value and as the result a new island may appear. So, there is specific interaction between condensed phases through the exciton concentration fields and as a result there is optimal distance between islands. To study the creation and structure of the exciton condensed phase we used the kinetic equations which describe a nucleation and growth of a new phase. Such theory was suggested for a description of a liquid phase formation from supersaturated gas, spinodal de- composition dynamics [23] and electron-hole drops in semiconductors [24–27]. There are several original pe- culiarities of the presented here paper, namely: 1) the consideration of two-dimensional problem, 2) taking into account the mutual influence of condensed phase islands, 3) the investigation of nonuniform particle distribution, 4) taking into account the finite value lifetime and the pumping of excitons. The distribution of exciton density at a steady state irradiation will be determined from a joint solution of the system of equations for the size distribution func- tion of islands and the exciton diffusion outside is- lands. The size of the islands are determined by four processes: the creation of excitons by the pumping, the capture of the excitons from environment, an escape of excitons from the island, and the exciton decay. Let us introduce the distribution function fn , which deter- mines the probability of the island to have n excitons. The distribution function satisfies the following ki- netic equation: � � � � f t j jn n n1 , (1) where jn is the probability current for transitions be- tween island states with n and (n 1) excitons, while jn�1 stands for transitions between the states with (n 1) and n excitons, j W n f W n fn n n� � � � ( ) ( )( ) ( )1 1, (2) W ( )� is the probability of transition in unit time with increasing (decreasing) number of excitons on unity. The values will be determined later. We shall study the steady-state solutions of Eq. (1). There are two types of stationary solutions. In the first case j j jn n� � ��1 const. In the theory of phase tran- sitions this solution is used for a determination of a rate of phase nucleation. In the second case j jn n� ��1 0. (3) The solution under the condition (3) describes the dy- namic equilibrium state which is formed after some time of pumping action. Such type solution we will investigate in future. Setting � � �f / tn 0 and jn � 0 in (2) we obtain the following steady-state solution: f f W m W m n m m n � � � � � � � � � � � �1 1 1 exp ln ( ) ( ) ( ) ( ) . (4) As we shall see the distribution function has sharp maximum at some value of n, which is much larger than unity (n �� 1). Islands with a large number of excitons and with a shape of disks are described by radiuses R n/n /� ( )� 1 2 instead of n. In this case the probabilities of W n( )( )� è W n( )( )� are satisfy the following conditions: W R Rc R W R R Gfi n ( )( ) ( ) ( )� � 2 2� � , (5) W R Rc W R R c /i if i ( )( ) ( )� � 2 2� � �ex , (6) where W R W n( ) ( )( ) ( )� �� , Wfi and Wif are the probabilities for the exciton to be captured by the disk and to escape from the disk per unit length of the circle in unit time and per one exciton of disc circle, respectively, G G r / Rn n n� � ( )dS � 2 is the mean value of the exciton pumping G r( ) over the nth island area, c R( ) and ci are the exciton densities on the circle of the disk and inside the disk. The relationship between the transition probabili- tiesWfi andWif may be obtain using the detailed ba- lance principle. In the case of infinite exciton lifetime and absence of pumping the equilibrium state between condensed phases is formed due to processes of exciton exchange between the island and an environment. Then in an equilibrium state the following condition should take place W R c W cfi if i( ) eq � . (7) Exciton condensation in quantum wells Fizika Nizkikh Temperatur, 2006, v. 32, No. 11 1451 Where ceq is the density of exciton gas which is in equilibrium state with the island. The Eq. (7) can be rewritten in the form W R W R W W c RkT if fi if fi i ( ) ( ) ( ) ( ) exp� � � � � � � �� �0 , (8) where Wfi ( )� and Wif ( )� are the transition probabil- ities in the case of a straight line boundary between condensed and gas phases, W /W c /cfi if i( ) ( )� � � � , c� is the equilibrium concentration of excitons on the straight division line c c /kT� � 10 exp( )� , (9) � is the energy condensation per one exciton, c10 � � � �( )*m kT/2 2 � , m* is the effective exciton mass, � is the degeneracy of the exciton state, �0 is the surface strain. The calculation of these probabilities needs an ap- plication of quantum—mechanical calculations. For an estimation of these values we shall calculate the exciton flux from gas phase in direction of the con- densed phase. To penetrate in an island exciton should overcome barrier caused by a dipole—dipole repulsion between excitons. Hereinafter we introduce the radius distribution function f R( ) instead of the particle distribution func- tion fn by the condition f R f dn/dR Rc fn i n( ) � � 2� . Let us introduce the dimensionless variables: ~R � � �R c c R c R /ci / i( ) , ~( ~) ( ) ,1 2 � �fi fi i /W c� ex ( ) ,1 2 ~G � �G /ci�ex . After some transformations the radius dis- tribution function (4) may be presented in the form f R f s Rb( ~) exp( ( ~))� , (10) where s R c R RG c /R /T fi fi ( ~) ln ~( ~) ~ ~ exp( ( ~) ) ~� 2 2 2 10 � � � � � R RdR R R s � � � � � �� ~ ~, ~ ~ (11) f f W m W m b m ns � � � � � � � � � � �1 1 1 exp ln ( ) ( ) ( ) ( ) , (12) where n s is the value of order of several unit (n s � 10), n R cs s i� � 2 . The factor fb marked out spe- cially since the sum in exponent of it can not be re- placed by integral. In the stationary case, the equation for the exciton density c R( ~) in the region outside the islands satisfies the diffusion equation D c r c r Gex ex � 2 ( ) ( ) � � , (13) where Dex is the exciton diffusion coefficient. The solution of the equation (13) should satisfy the following boundary conditions: the current of excitons to every island should be equal to the difference be- tween the number of excitons captured by the island and the number of excitons which escape from the island. So, the boundary conditions on ith disk has the form 2 2� �R D c R W c W ci i i fi i iex ( ( ) )| ( ( ( ) )| )� � � r n r if , (14) where Ri is the radius of the ith island, n is the out- ward and unitary normal vector to the circular boundary of the ith island, the notation �i means that the boundary conditions of Eq. (14) are applied at the surface of an island. The solution of equation (12) may be written in the following form: � �c n r A K /l r mim m i mi i( ) ( ) ( ) cos( ( ) ),r r r� � � ��ex 0 � (15) where � �n r D K /l G dSex ex ( ) ( ) ( )� � � �� 1 2 0� r r r , (16) K xm( ) is the modified Bessel function, ri is the posi- tion of the center of the ith island, � i r( ) is the polar angle of the point r in the coordinate system with the origin at the center of the ith island, l D /� ( )ex ex� 1 2 is the exciton diffusion length. The first term in the expression (15) is the particular solution of the inho- mogeneous equation (13). The second term is the solution of the homogeneous equation (13). So, the ex- pression (15) satisfies the equation (13). The unknown coefficients Aim are to be determined from the bound- ary conditions (14). The solution (15) has distinct meaning: the first term describes the exciton density created by pumping in the case of absence of condensa- tion phases, the second term describes the perturbation of the exciton density caused by presence of condensed phase islands which capture excitons. 3. Uniform pumping In this section we shall investigate an uniform cre- ation of excitons in double quantum well [28], i.e., the pumping does not depend on space coordinates. In this case the first term in (15) equals n Gex ex� � . (17) Firstly we consider the behavior of the system in a vicinity of threshold of condensed phase formation. In this case the mutual influence between islands can be neglected and single island may be studied. The solu- 1452 Fizika Nizkikh Temperatur, 2006, v. 32, No. 11 V.I. Sugakov tion (15) gives together with boundary condition (14) the following dependence of exciton density outside of island with a center in the origin of coordinates (in di- mension units) � �~(~) ~ ( ~) (~ ~ )c G A R K /lsr r� 00 0 , (18) where A R G c /R /T K R/l l K s 00 10 1 ( ~) [~ ~ exp( ( ~) )] ( ~ ~ ) ~� � � � � 0( ~ ~ )R/l . (19) The radius distribution function obtained from (10) for different pumping is presented in Fig. 1. The function has sharp maximum for exception of the pumping very closed to the threshold value. The maximum determines the most probable meaning of radius of the island. The threshold value of pumping can be obtained from two conditions: � � � � � �f R / R f R / R( ~) ~ ( ~) ~2 2 0 . (20) We calculated the phase diagram «critical pumping —temperature» using the formulae (10), (20). Param- eters (m D*, ,ex ex� ) were chosen for GaAs, �, �, Td were fitting parameters. Results are presented in Fig. 2, where points are experimental data from [9]. It is seen that theory describes the experiments without suggestion about Bose–Einstein condensation. Many islands arise with increasing pumping. There is specific interaction between islands through exciton concentration fields. In studied system the islands should be localized on the ring where the exciton den- sity has maximum. Since islands of condensed phases takes excitons from the same source, two islands can not be situated close one to another. Moreover, the distance between islands can not be large because in this case the exciton density between them becomes greater than a critical value and as the result a new is- land may appear. Thus, there is optimal distance be- tween islands. The connection between islands occurs due to the dependence of the exciton density c R( ~) in (11) at the surface of the considered island versus the presence of other islands. We shall make two approximations in further calculations. We shall suggest that the distance between islands is larger than the island radius. Under such condition the exciton density perturbation caused by the presence of some island slowly changes in limits of a size of other island and this perturbation may be considered as constant in the limits of island size. It means that only terms with m � 0 gives the contribu- tion in the sum of the expression (15). So, Aim � 0 at m � 0. Also we suggest that this perturbation is formed by many islands and as result it may be studied in a meanfield approximation. Let us consider the exciton density in vicinity of some island, for example, with i � 0. In mean field approximation all coefficients Ai0 in the sum of (15) can be replaced by their average value Ai0 in all terms for exception of the term with i � 0, the sum over i can be changed by integral in a suggestion that islands are distributed in space with density c N/Sisl � , where N is the number of islands, S is the area of quantum well. Such approximation al- lows to simplify the boundary condition (14). After some transformations from boundary conditions (14) we obtain the following expressions for exciton density on the surface of the island with the radius ~R: ~( ~) ~( ~,~ ) ~ ( ~, ~,~ ) ( ~ ~ )c R c R c G A R R c K R/l� � isl isl00 0 2 00 2�A R c l c( ~,~ ) ~ ~ isl isl , (21) where Exciton condensation in quantum wells Fizika Nizkikh Temperatur, 2006, v. 32, No. 11 1453 5 10 15 20 25 30 35 1 2 3 Radius, m D is tr ib u tio n fu n ct io n 0.5 0.4 0.3 0.2 0.1 0 0.6 Fig. 1. Radius distribution function for single condensed phase island in quantum well at different pumping ~ G: 0.00715 (1), 0.01 (2), 0.015 (3). � � 10 K, a � 4 K, T � 2 K, ~ ,l ci� � �320 1011 2cm . 0.5 0.4 0.3 0.2 0.1 0.5 1.0 1.5 2.0 2.5 3.0 3.5 T, K Gc 0 Fig. 2. The dependence of threshold value of pumping on temperature (solid curve). � �� �3 1K K, , Td � 25. ,K ~ ,l � � �100 10 8�ex s. The points correspond to the experi- mental data from [9]. A R R c G A R c l c c 00 00 2 12 ( ~, ~,~ ) ( ~ ( ~,~ ) ~ ~ ~ isl isl isl� � � 0 0 0 1 0 exp( ( ~) )) ( ~ ~ ) ~ ( ~ ~ ) � � � /R /T K R/l l K R/l , (22) A R c G c /R /T l K R/ 00 10 1 ( ~,~ ) ( ~ ~ exp( ( ~ ) )) ~ ( ~ ~isl � � � � l K R/l l c) ( ~ ~ ) ~ ~ � ��0 22 isl , (23) ~R is the mean radius, which coincides with most probable one. Let us consider the ensemble of identical systems in which may be realized the state with different number of islands. In system with many islands the distribu- tion function depends on radiuses (or numbers of excitons) of all islands and determined by the func- tion f R R R R Ni N( ~ , ~ ~ ~ ; )1 2 � � � �� � . The equation (1) should be rewritten for many-islands system. In the model of self-consistent field the distribution function can be present as product of distribution functions for separate islands f R R R R N s R ci N i i N ( ~ , ~ ~ ~ ; ) exp ( ~ ,~ )1 2 1 � � � � � � �� � isl isl . (24) After integrating of this value over radiuses of all islands we obtain the probabilities of a realization of the state with N islands (or the probability of the state with the density ~cisl ) F c Sc s R c(~ ) exp( ~ ( ~,~ ))isl isl isl isl� , (25) s R c c R RG c / fi fi isl isl( ~,~ ) ln ~( ~) ~ exp( ( � 2 2 2 10 � � � � � ~) ) ~ ~ ~ ~ R /T R RdR R R s � � � � � �� . (26) The most probable density of islands is determined by the condition � � � ( ( ~,~ )) ~ c s R c c isl isl isl isl 0. (27) Intensity of the emission, created by excitons of condensed phase is proportional to the number of excitons, which islands contain I R cT � ~ ~2 isl . (28) In approximation (28) we neglect by the stimulating emission. The results of temperature dependence calculation of emission intensity according (28) are presented in the Fig. 3. The dependence is described by straight line. The line crosses the axis at the temperature for which the considered pumping has threshold value. Such behavior of intensity coincides with empirical formula obtained from experiment in the papers [9] I T TT c 1 . (29) So, we have explained the experiments without sug- gestion about Bose–Einstein condensation of excitons. 4. Formation of luminescence from the ring outside the laser spot As it was mentioned in Introduction, the authors of [11–16] observed the emission from the ring outside the laser spot at the distances from a center of the spot much larger than exciton diffusion length. At low temperatures, the ring fragmented into a periodical structure [11,12,15]. The mechanism of ring appearance was suggested [15,16] basing on two assumptions: 1) without light irradiation the well is populated with a certain density of electrons; 2) holes are captured by the well with a 1454 Fizika Nizkikh Temperatur, 2006, v. 32, No. 11 V.I. Sugakov 1.4 1.5 1.6 1.7 1.8 1.9 0.02 0.04 0.06 0.08 TI 2 3 1 0.1 T, K Fig. 3. The dependence of luminescence intensity on tem- perature at different pumping, ~ G: 0085. (1), 0.11367 (2), 0.1644 (3). Parameters are the same as in Fig. 2. larger probability than electrons. As the result, the ir- radiated structure develops two differently charged spatial regions. The laser spot and the region around it are charged positively while the region far away re- mains negative. Electrons and holes can recombine only where they meet, therefore, the sharp lumines- cence ring is formed on the boundary between regions of opposite charges. The model of the fragmentation observed in [11,12,15] was bild in [29,30] on the base of the tradi- tional phase transition theory, generalized for the case of a nonequilibrium system, in which particles (excitons) have finite lifetime. The recombination of electrons and holes on the ring determines the exciton pumping. Ac- cording to [15,16] the exciton density has sharp maxi- mum at the some distance from laser spot. If the density exceeds the threshold value the islands of condensed phase are generated on the ring (see insert in Fig. 4). The system can be described by kinetic equations (1) with interaction between islands through exciton con- centration fields. The study of sizes of islands (frag- ments) and the distance between them is similar to the consideration of island parameters at uniform pumping presented above. Results of such study are presented in [29,30]. The probability for system to have N islands is described by the formula f N f S N( ) exp( ( ))� 0isl isl . (30) The most probable number of islands is determined by extremum of function S Nisl ( ). The results of taken from [29,30] calculations of temperature dependence of island radius, distance between islands and fluctua- tions of the distances are presented in Fig. 4. The dis- tance from the laser spot to fragmented ring (r0) equals 200 �m. In this section we shall consider the case when the condensed phase of excitons is shaped as a ring around the laser spot in region of the largest electron-hole re- combination. Let the internal and external radii of the ring formed by the condensed phase are equal to r1 and r2, respectively, and r0 is the radius of the maximum of the exciton production rate G r( ) (see the insert in Fig. 5). The ring grows due to the exciton creation by pumping and due to the exciton inflow from the envi- ronment. The ring narrows as a result of the exciton decay and the exciton escape. The consideration of the exciton condensation in the shape of a ring is similar to the study of the forma- tion of exciton islands considered in the previous sec- tions. Therefore, we shall set out the problem briefly. Similarly to the consideration of the islands, let us in- troduce the distribution function fn , which deter- mines the probability that the condensed phase shaped as a ring contains n excitons. The distribution function obeys to an equation similar to the equation (1), in which the probability current is equal to j W r c r r c r fn fi n n n n n� � � � � �2 1 1 1 1 2 1 2 1 1� ( ( ) ( )), , , , � � �2 21 0 1 1 1 2� ��r r G r f W r r c fn n n if n n i n( ) ( ) � �( ), ,r r c f /n n i n2 2 1 2 ex , (31) �where c r n( ),1 and c r n( ),2 are the exciton density on the internal and the external circular boundaries of the ring, �r r rn n n� 2 1, , is the thickness of the ring, G rn( ) is the mean value of the pumping G r rG r dr/ r rn n r r n n ( ) ( ) ( ) , � � � 1 2 0 . Other designations have been defined in the previ- ous sections. Exciton condensation in quantum wells Fizika Nizkikh Temperatur, 2006, v. 32, No. 11 � 1455 d R �d/d 14 12 10 8 6 4 2 0 R ,d , m � 1.6 2.0 2.4 2.8 T, K 0.3 0.2 0.1 0 � d /d r0 2 /N� Fig. 4. Dependences of the island radius R, distance be- tween islands d and ratio of the island fluctuation shift to the distance between islands �d/d on temperature. The insert shows positions of exciton condensed phase islands around the laser excitation spot. 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.5 1.0 1.5 2.0 2.5 3.0 2.4 2.0 1.6 1.2 0.8 /S S is l r T, K r, m m r1 r2 r0 r 1 2 Fig. 5. Dependences of the thickness of condensed phase ring �r (1) and ratio of values Sisl and Sr (2) on temperature. The insert shows the condensed phase in the form of a ring. Further we take into account the fact that n �� 1 and obtain a Fokker–Planck equation for the distribu- tion function. We assume also that the thickness of the ring is narrower than the ring radius r rn0 �� � . At such approximation c r c rn n( ) ( ), ,1 2� . After reductions similar to those carried out in the subsection 2, we obtain the following expression for the distribution function: f f W c r W c r c / G r W cn fi n if i i n fi � 0 1 2 2 2 exp ( ( ) ) ( ( ) ( , � �ex ( ) ) ( ( ),r W c r c / G r dn n if i i n n 1 0 � � � � � ��� � �ex . (32) Coefficients Wfi and Wif are connected by the de- tailed balance principle of Eq. (8). As the radius of the ring is significantly larger than the radius of an island of the previous section we may neglect the surface en- ergy and put �0 0� . The exciton density beyond the region of the con- densed phase obeys Eq. (13). Under condition of the accepted approximation that the thickness of the ring is significantly smaller than its radius, the solution of this equation takes the form c r n c r/ls( ) exp( )� ex 0 , (33) where nex is the solution of the nonuniform Eq. (13). It describes the exciton density created by the electron-hole recombination in the absence of the condensed phase. The value of cs0 is determined by the boundary condition of Eq. (14). For the ring, this condition may be written in the form D c r r W c r W cfi n if iex � � � ( ) ( ),1 . (34) The solution of Eq. (33) with the boundary condi- tion of Eq. (34), assuming that the thickness of the ring of the condensed phase is smaller than the diffu- sion length (�r l� ), reads c r n c /T W l/D c /Tn fi ( ) exp( ) exp( ).,1 10 101 � ex ex � � (35) Further we shall use the thickness distribution func- tion which is defined as f r f dn/d r r c fn i n( )� �� � 2 0� . Using dimensionless variables, we obtain the following formula for the thickness distribution function from Eqs. (32) and (35): f r f S rr r( ) exp( ( ))� �� 0 , (36) where S r r n c /T r G r /l r( ) { [ exp( )] ( ( )( � � � � 4 2 1 10 10 0� � � �ex )} [ exp( )] ( ( ))( ) d r n c /T r G r /l c � � � � 2 1 1 410 0� � � �ex 10 0 1( ) exp( ) � � �/l /T r� . (37) The most probable thickness of the ring is deter- mined by the condition � � � S r r r( )� � 0. (38) Using Eq. (38), we have calculated the most prob- able thickness of the ring if the condensed phase is realized in the shape of a ring at the following values of parameters: � �� �10 30K K, , ci � �1011 2cm , m m* . ,� 0 37 0 �ex s� �10 7 . The same values of parame- ters were used in [30] during analysis of a fragment formation. As seen from Fig. 5, the thickness of the ring decreases with rising the temperature. The question to answer is what type of the con- densed phase is realized? Is it the fragmented ring or is it the continuous ring? In order to find the solution of the problem we have performed the following estimations. We have compared the probabilities of the appearance of these two states. The curve 2 in Fig. 5 shows the ratio of the values Sisl and Sr , which determine the probabilities according to the formulas of Eqs. (30) and (36), correspondingly. It should be noted that absolute values of Sisl and Sr are large for exception the region of critical tempera- ture. It means that one of two states (fragmental ring or continues ring) can been realized with high proba- bilities depending on the meaning of ratio S /Srisl is larger or less than unity. It is seen that the state with the phase distribution in the form of fragments is more probable than the distribution in the form of the continuous ring at chosen values of parameters. In the vicinity to the critical temperature the frag- mented ring is transformed into continues ring. Such transformation was observed in experiment [15]. 1456 Fizika Nizkikh Temperatur, 2006, v. 32, No. 11 V.I. Sugakov Conclusion In statistical theory of exciton condensed phase for- mation in two-dimensional system it is suggested that condensed phase arises as a result of attractive interac- tion between excitons. The investigated system is nonequilibrium one as a result of presence pumping and finite value of exciton lifetime. It forms some spe- cific properties such as restricted sizes of condensed phase regions (islands), a correlation in positions of islands, so on. The theory was applied to explanation of experimental manifestation of condensed phase in quantum wells and also to explanation of the periodi- cal fragmentation, which was observed in lumines- cence spectrum from a ring around a laser spot in a crystal with double quantum wells. For such explana- tions the theory does not require the exciton Bose–Einstein condensation. 1. S.V. Moskalenko, Bose–Einstain Condensation of Exci- tons and Biexciton, Academy of Science of Moldovian SSR, Kisheney (1970). 2. D. Hulin, A. Mysyrowicz, and a la Guillaume Benoitm, Phys. Rev. Lett. 45, 1970 (1980). 3. L. Keldysh, Excitons in Semiconductors, Nauka, Moscow (1971). 4. Ya. Pokrowskii, Phys. Status Solidi 11, 385 (1972). 5. T.M. Bragina, Yu.S. Lelikov, and Yu.G. Shreter, Zh. Exp. Theor. Phys. 79, 1838 (1980). 6. T.M. Rice, J.C. Hensel, T.G. Phillips, and G.A. Tho- mas, The Electron-Hole Liquid in Semiconductors, Aca- demic Press, New York (1977). 7. N.V. Zamkovets, N.N. Sybeldin, and V.A. Tsvetkov, Zh. Exp. Theor. Phys. 105, 1066 (1994). 8. K.E. O’Hara and J.P. Wolfe, Phys. Rev. B62, 12909 (2000). 9. A.V. Larionov and V.B. Timofeev, JETP Lett. 73, 342 (2001); V.B. Timofeev, Usp. Fiz. Nauk 125, 315 (2005). 10. V. Krivolapchuk, E.S. Moskalenko, and A.L. Zhmo- dikov, Phys. Rev. B64, 045313 (2001). 11. L.V. Butov, A.C. Gossard, and D.S. Chemla, Nature 418, 751 (2002). 12. L.V. Butov, Solid State Commun. 127, 89 (2003). 13. D. Snoke, S. Denev, Y. Liu, L. Pfeiffer, and K. West, Nature 418, 754 (2002). 14. D. Snoke, Y. Liu, S. Denev, L. Pfeiffer, and K. West, Solid State Commun. 127, 187 (2003). 15. L.V. Butov, V.B. Levitov, A.V. Mintsev, B.D. Simons, A.C. Gossard, and D.S. Chemla, Phys. Rev. Lett. 92, 117404 (2004). 16. R. Rapaport, G. Chen, D. Snoke, S.H. Simon, L. Pfeiffer, K. West, Y. Liu, and S. Denev, Phys. Rev. Lett. 92, 117405 (2004). 17. O.L. Berman, Yu.E. Lozovik, D.W. Snoke, and R.D. Coalson, Phys. Rev. B70, 235310 (2004). 18. E.A. Cornell and C.E. Wieman, Rev. Mod. Phys. 74, 875 (2002). 19. D. Birkedal, J. Singh, V.G. Lyssenko, J. Erland, and J.M. Hvam, Phys. Rev. Lett. 76, 672 (1996). 20. Yu.E. Lozovik and O.L. Berman, JETP Lett. 64, 573 (1996). 21. Yu.E. Lozovik and O.L. Berman, JETP 84, 1027 (1997). 22. J. Fernandez-Rossier and C. Tejedor, Phys. Rev. Lett. 78, 4809 (1997). 23. I.M. Lifshits and V.V. Slyosov, J. Phys. Chem. Solids 19, 35 (1961). 24. R.N. Silver, Phys. Rev. B11, 1569 (1975). 25. P.M. Westerwelt, Phys. Status Solidi 74, 727 (1976). 26. V.S. Bagaev, N.V. Zamkovets, L.V. Keldysh, N.M. Sibeldin, and V.A. Tsetkov, JETP 70, 1501 (1976). 27. V.I. Sugakov, Mol. Cryst. Liq. Cryst. 384, 11 (2002). 28. V.I. Sugakov, Phys. Solid State 46, 1496 (2004). 29. V.I. Sugakov, Ukr. J. Phys. 49, 1117 (2004). 30. V.I. Sugakov, Solid State Commun. 134, 63 (2005). Exciton condensation in quantum wells Fizika Nizkikh Temperatur, 2006, v. 32, No. 11 1457