搜档网
当前位置:搜档网 › ApplPhysLett_98_091906

ApplPhysLett_98_091906

Comprehensive insights into point defect and defect cluster formation in CuInSe2

Christiane Stephan,1,a?Susan Schorr,2Michael Tovar,1and Hans-Werner Schock1

1Helmholtz-Zentrum Berlin fuer Materialien und Energie,Hahn-Meitner Platz1,14109Berlin,Germany

2Faculty of Earth Science,Free University Berlin,Malteserstra?e74-100,12249Berlin,Germany

?Received16November2010;accepted6February2011;published online1March2011?

The concentration of native point defects in CuInSe2powder material as a function of stoichiometry

has been experimentally determined by neutron powder diffraction.A correlation between the Cu/In

ratio and the density of V Cu as well as In Cu has been established and their concentrations are

quanti?ed.It is demonstrated,that assuming the spontaneous formation of defect pairs,the density

of native point defects is reduced signi?cantly by an order of magnitude.The functionality of a

solar device,assuming same conditions like in the analyzed material,may be explained by a

neutralization due to the formation of electrically inactive defect complexes.?2011American

Institute of Physics.?doi:10.1063/1.3559621?

The electronic properties of a polycrystalline heterojunc-

tion thin?lm solar cell are strongly in?uenced by the pres-

ence of electrical active defects.Especially the native defects

in the most important component of the device,the com-

pound semiconductor CuInSe2?CISe?acting as absorber,are crucial.These defects,such as vacancies,interstitials and

antisites,are resulting from deviations from the stoichio-

metric composition.To achieve a p-type CISe absorber an

overall slightly copper-poor composition is necessary,1

whereas the region close to the buffer/absorber interface

should be of n-type character by exhibiting a very copper-

poor composition.2–4High ef?cient devices are actually fab-

ricated by a complex vacuum based multistage process using

three separate elemental sources of Cu,In,and Se.5,6The

carrier type and concentration is implemented by carefully

controlling the Cu/In and Se/M?M=Cu+In?ratio adjusting the stoichiometry with respect to Cu.Native point defects and the carrier type in CISe are still under discussion.It is generally believed that Cu vacancies?V Cu?cause p-type con-ductivity,whereas copper on interstitial positions?Cu i?or In Cu antisites act as donors and promote a n-type character.4 Numerous investigations of CISe thin?lms7,8and a few studies concerning the electronic defect levels in CISe single crystals9,10could not solve the problem of the structural ori-gin of defects,producing these levels.To date,no quantita-tive study has been made on the variation of native point defects as a function of stoichiometry.Besides the need of a high carrier concentration a crucial prerequisite for a high conversion ef?ciency is the chalcopyrite type structure ?space group I4ˉ2d?of the absorber.With deviations from stoichiometry the chalcopyrite type structure still consists over a relatively wide compositional range in CISe.11How-ever,structural parameters like the tetragonal distortion???=?1?c/2a?,a and c are the lattice constants?and the dis-placement of the anion from the ideal tetrahedral position, expressed by the parameter u with u=?0.25?x??x is the an-ion position coordinate?,are changing.An investigation of the lattice parameter in nonstoichiometric CuInSe2bulk samples grown by the Bridgman method has been reported by Merino et al.12They assumed without experimental evi-dence the existence of copper on interstitial positions and the generation of copper vacancies with increasing copper de?-ciency.

In this letter,we report on experimental correlations be-tween the stoichiometry of CISe polycrystalline bulk mate-rial and the concentration of native point defects,such as copper and indium vacancies,interstitials,and antisites.

Powder samples of Cu1?y In y Se0.5+y with various compo-sition?0.49?y?0.60?have been prepared by solid state re-action of the pure elements.Details of the growth procedure are described by Stephan et al.11During the synthesis the selenium partial pressure was not controlled,therefore all detailed results presented here are valid just for the synthesis method https://www.sodocs.net/doc/5318679900.html,positional investigations were carried out by electron microprobe analysis?wavelength dispersive x-ray?WDX?measurements using a JEOL-JXA8200sys-tem?.In order to obtain reliable results from the WDX mea-surements,the system was calibrated using elemental stan-dards of Cu,In,and Se.By means of the decided fabrication and characterization of the polycrystalline bulk material a series of CISe powder reference samples with various well determined chemical compositions was produced.The com-position and phase content of this sample series is summa-rized in Table I.Thus a systematic study of structural trends in CISe bulk material in dependence on the stoichiometry could be performed over a wide compositional range.

In order to determine the cation distribution in the crys-tal structure of the chalcopyrite type phase,which is the basis for the calculation of the point defect concentrations,the

a?Author to whom correspondence should be addressed.Electronic mail: christiane.stephan@helmholtz-berlin.de.TABLE https://www.sodocs.net/doc/5318679900.html,position and phase content for the powder samples studied by neutron powder diffraction?Cu1?y In y Se0.5+y,Ch-chalcopyrite type phase?.

Sample y Cu/In Phases

A0.49?1? 1.04?1?Ch,Cu-selenides

B0.51?1?0.95?1?Ch

C0.52?1?0.94?1?Ch

D0.54?1?0.84?1?Ch

E0.57?1?0.75?1?Ch,CuIn3Se5

APPLIED PHYSICS LETTERS98,091906?2011?

0003-6951/2011/98?9?/091906/3/$30.00?2011American Institute of Physics

98,091906-1

method of the average neutron scattering length has been adopted.13A detailed introduction to this method applied to

CuB III X 2VI absorber materials is explained elsewhere.14

In the present study this powerful approach,which is based on neu-tron powder diffraction,is used to determine defect concen-trations with structural origin in nonstoichiometric CISe.The approach is based on the fact,that vacancies ?V Cu and V In ?as well as antisite defects ?In Cu and Cu In ?will change the neu-tron scattering length of the cation sites 4a ?copper site ?and 4b ?indium site ?in the chalcopyrite type structure signi?-cantly,because the neutron scattering lengths of copper and indium are different ?b Cu =7.718?4?fm ,b In =4.065?2?fm ?Ref.15??.Thus the distribution of copper and indium on the both cation sites of the chalcopyrite type structure can be revealed on the basis of the cation site occupancy values determined by Rietveld analysis of the neutron diffraction data.Neutron powder diffraction data were collected at the Berlin Research Reactor BERII using the ?ne resolution powder diffracto-meter E9??=1.79776??.The data treatment was per-formed by full pattern Rietveld re?nement 16using the FULL-PROF suite software package.17

Free parameters of the structural re?nement of the chalcopyrite type phase have been lattice constants ?a and c ?,the anion x -coordinate,site occupation numbers for the copper ?occ 4a ?and indium ?occ 4b ?position as well as anisotropic atomic displacement parameters for all three structural sites.The R Bragg values of the re?nements are between 5and 6.The cation site occupa-tion numbers were used to determine the experimental aver-age neutron scattering lengths b ˉj

exp ?with j =4a ,4b ?of the cation sites 4a and 4b according to

b ˉ4a exp =oc

c 4a ·b Cu b ˉ4b exp =occ 4b ·b In .

?1?

The experimental average neutron scattering length as a function of stoichiometry ?see Fig.1?shows a decrease of b ˉ4a exp and an increase of b ˉ4b exp with decreasing Cu/In ratio.The latter can only be caused by the formation of Cu In antisite

defects ?because b Cu ?b In ?.To realize a decrease of the b ˉ4a

exp values,the structural site 4a has to be occupied by copper,indium ?In Cu ?and/or copper vacancies ?V Cu ?.Only the copper-rich sample ?A ?with Cu /In ?1shows no indication for an In Cu antisite defect,because there is no decrease of the

b ˉ4a exp value.Thus the structural site 4a seems to be fully oc-cupied by copper.On the other hand a slight increase in the b ˉ4b exp value can be observed.

For the quantitative evaluation of the cation distribution

from b ˉj

exp of the cation sites 4a and 4b ,the following proce-dure was developed.Assuming copper and indium as well as vacancies would occupy the same structural site j ,the average neutron scattering length of the site ?b ˉj ?is given by Eq.?2?

b ˉj =N Cu j ·b Cu +N In j ·b In +V j .

?2?

Here N Cu j and N In j are the fractional amounts of copper and indium on the corresponding site,b Cu and b In are the neutron scattering lengths of copper and indium and V j is the fraction of vacancies on the site.

The fractional amounts of copper and indium on the both cation sites 4a and 4b have to be in agreement with the chemical composition of the chalcopyrite phase.On the basis of Eq.?2?and the additional requirement N Cu j +N In j +V j =1,as well as assuming a certain cation distribution,a theoretical average neutron scattering length ?b ˉj calc ?can be calculated.In the beginning a simple distribution model without cation va-cancies was applied,just assuming In Cu antisite defects for copper-poor samples ?B-E ?.The comparison between experi-mental and theoretical average neutron scattering lengths

showed discrepancies like b ˉ4a exp ?b ˉ4a calc and b ˉ4b exp ?b ˉ4b calc .In the

next step the cation distribution model was modi?ed by in-troducing copper vacancies and Cu In antisite defects to

achieve the b ˉ4a exp and b ˉ4b exp values,respectively.Thereby the

total amount of copper in the chalcopyrite phase,was distrib-uted on the two cation sites 4a and 4b ,thus the occupation of interstitial positions by copper ?Cu i ?could be excluded.For the copper-rich sample A it can be concluded,that a Cu In antisite defect,which would act as a donor,with a maximal site fraction of ?3?2?%on the 4b site has to be the reason

for the increase in the b ˉ4b exp

value.This means the chalcopyrite type crystal structure tolerates an additional copper incorpo-ration for copper-rich CISe of up to 3%into the cation sub-structure.

Finally the defect concentrations per cubic centimeter can be calculated using the extracted fractional amounts of V Cu ,Cu In ,and In Cu antisite defects.The unit cell volume was determined by the obtained lattice constants.The concentra-tion of intrinsic point defects as a function of stoichiometry in polycrystalline CISe bulk samples is summarized in Fig.2,where the development of the different kinds of intrinsic point defects in dependence on the Cu/In ratio is shown.The observed trend of an increasing amount of copper vacancies with increasing copper de?ciency corresponds to the defect formation energies calculated by Zhang et al.18Due to the negative defect formation energy ??1.9eV ?of V Cu in copper-poor CISe it is traceable to observe high concentra-tions of copper vacancies in copper-poor samples.Cu In and In Cu antisite defects are present for Cu /In ?0.95and in-crease with decreasing Cu/In ratio but with a different slope.The increment of native defects in CISe with decreasing Cu/In ratio gives rise to the formation of the ordered vacancy compound ?OVC ?CuIn 3Se 5.The latter appears as secondary phase besides the chalcopyrite type phase in the very copper-poor sample E.The growth of the OVC phase can be

re-

FIG.1.?Color online ?Experimentally observed average neutron scattering

length of the cation sites 4a and 4b of the chalcopyrite type structure as a function of stoichiometry.The horizontal dotted lines show the neutron scat-tering length for Cu ?b Cu ?and In ?b In ?for comparison.

garded as an accumulation of copper vacancies,leading to the generation of very small CuIn 3Se 5exsolutions like do-mains within the chalcopyrite type phase,resulting in more complex electronic properties.

The depicted experimental correlation between stoichi-ometry of CISe polycrystalline bulk material and the concen-tration of native point defects can be used to get an impres-sion of the situation in nonstoichiometric CISe thin ?lms.The calculated defect concentrations are relatively high and increase with increasing copper de?ciency.On the other hand,high ef?cient devices exhibit a copper-poor composi-tion of the CISe absorber layer,especially the very copper de?cient CISe surface.3The formation of defect arrays as proposed by Zhang et al.18,19can be considered to explain the functionality of CISe absorbers with Cu /In ?1.They showed by ab initio calculations,that the defect pairs

?2V Cu

?+In Cu 2+?and ?Cu In 2?+In Cu 2+?in CISe have very low forma-tion energies,moreover these defect pairs are predicted to be electrically inactive.Therefore the calculated defect concen-trations were used to estimate the defect pair density and the resulting carrier type ?Table II ?.It can be shown,that assum-ing the spontaneous formation of defect pairs,the density of native defects is reduced signi?cantly by an order of magni-tude.Thus the surprising electrical tolerance of CISe to its huge concentrations of native defects could be explained.

In conclusion,we have shown that the concentration of native cation point defects in polycrystalline CISe bulk ma-

terial can be experimentally determined by neutron powder diffraction.Assuming the formation of defect pairs,it is pos-sible to deduce a change in the carrier type character from p-type to n-type if the material is very copper-poor,whereas the chalcopyrite type phase,coexisting with the OVC,is p-type again.Assuming a similar scenario for CISe thin ?lms,it is possible to realize that non-stoichiometric CISe is well performing as absorber material in thin ?lm solar cells despite a high concentration of native defects.

1

R.Nou?,R.Axton,C.Herrington,and S.K.Deb,Appl.Phys.Lett.45,668?1984?.2

A.Niemegeers,M.Burgelman,R.Herberholz,U.Rau,D.Hariskos,and H.W.Schock,Prog Photovoltaics 6,407?1998?.3

R.Klenk,Thin Solid Films 387,135?2001?.4

S.Siebentritt,M.Igalson,C.Persson,and https://www.sodocs.net/doc/5318679900.html,ny,Prog.Photovoltaics 18,390?2010?.5

C.A.Kaufmann,A.Neisser,R.Klenk,and R.Scheer,Thin Solid Films 480–481,515?2005?.6

I.Repins,M.A.Contreras,B.Egaas,C.DeHart,J.Scharf,C.L.Perkins,B.To,and R.Nou?,Prog.Photovoltaics 16,235?2008?.7

M.Igalson,P.Zabierowski,D.Przado,A.Urbaniak,M.Edoff,and W.N.Shafarman,Sol.Energy Mater.Sol.Cells 93,1290?2009?.8

A.L.Li and I.Shih,J.Electron.Mater.22,195?1993?.9

L.Kaplan,G.Leitus,V .Lyakhovitskaya,F.Frolow,H.Hallak,A.Kvick and D.Cahen,Adv.Mater.?Weinheim,Ger.?12,366?2000?.10

G.Zahn and P.Pau?er,Cryst.Res.Technol.23,499?1988?.11

C.Stephan,S.Schorr,and H.W.Schock,Proceedings of the 2009MRS Spring Meeting Symposium M,2009,V ol.M1165.12

J.M.Merino,J.L.M.deVidales,S.Mahanty,R.Diaz,F.Rueda,and M.Leon,J.Appl.Phys.80,5610?1996?.13

A.Furrer,J.Mesot,and T.Str?ssle,Series on Neutron Techniques and Applications,V ol.4?2009?.14

S.Schorr,C.Stephan,T.Th?rndahl,and R.Mainz,in X-Ray and Neutron Diffraction on Materials for Thin Film Solar Cells ,edited by D.Abou-Ras,T.Kirchhartz,and U.Rau ?Wiley-VCH,Weinheim,2011?.15

A.-J.Dianoux and https://www.sodocs.net/doc/5318679900.html,nder,Institute Laue-Langevin 1?2001?.16

H.M.Rietveld,J.Appl.Crystallogr.2,65?1969?.17

J.Rodriguez-Carvajal and T.Roisnel,www.ill.eu/sites/fullprof/.18

S.B.Zhang,S.H.Wei,A.Zunger,and H.Katayama-Yoshida,Phys.Rev.B 57,9642?1998?.19

S.B.Zhang,S.H.Wei,and A.Zunger,Phys.Rev.Lett.78,4059?1997?

.

FIG.2.?Color online ?Concentration of different kinds of intrinsic point

defects as a function of nonstoichiometry.The values always refer to the chalcopyrite phase in the samples.The dotted line indexes the increase of the In Cu antisite defect and is just a guide to the eye.The solid lines mark the region where different phases occur.

TABLE II.Estimated defect pair density and residual concentration of na-tive point defects in the chalcopyrite type phase ?normalized to the total defect concentration ?.Cu/In

0.75

0.84

0.94

0.95

1.05

Defect pairs 96%90%60%00Native defects 4%Cu In 10%In Cu 40%Cu In 100%V Cu 100%Cu In Type p-type n-type p-type p-type p-type

相关主题