少子寿命测试判断是否有外延

Abruptness of a-Si:H/c-Si interface revealed by carrier lifetime measurementsStefaan De Wolf and Michio KondoCitation: Appl. Phys. Lett. 90, 042111 (2007); doi: 10.1063/1.2432297View online: /10.1063/1.2432297View Table of Contents: /resource/1/APPLAB/v90/i4Published by the AIP Publishing LLC.Additional information on Appl. Phys. Lett.Journal Homepage: /Journal Information: /about/about_the_journalTop downloads: /features/most_downloadedInformation for Authors: /authorsAbruptness of a-Si:H/c-Si interface revealed by carrierlifetime measurementsStefaan De Wolf a͒and Michio KondoNational Institute of Advanced Industrial Science and Technology(AIST),Central2,1-1-1Umezono,Tsukuba,Ibaraki305-8568,Japan͑Received27September2006;accepted15December2006;published online26January2007͒Intrinsic hydrogenated amorphous siliconfilms can yield outstanding electronic surface passivation of crystalline silicon wafers.In this letter the authors confirm that this is strongly determined by the abruptness of the interface.For completely amorphousfilms the passivation quality improves by annealing at temperatures up to260°C,most likely byfilm relaxation.This is different when an epitaxial layer has been grown at the interface duringfilm deposition.Annealing is in such a case detrimental for the passivation.Consequently,the authors argue that annealing followed by carrier lifetime measurements allows determining whether the interface is abrupt.©2007American Institute of Physics.͓DOI:10.1063/1.2432297͔Hydrogenated amorphous silicon͑a-Si:H͒films depos-ited on crystalline silicon͑c-Si͒surfaces have increasingly attracted attention over the past20years.Initially,it was discovered that abrupt electronic heterojunctions can be cre-ated with such structures.1Soon afterwards applications fol-lowed,including bipolar transistors,2imaging devices,3and solar cells.4For the latter it was recognized that the output parameters benefit substantially from inserting a few nano-meter thin intrinsic a-Si:H͑i͒film between the doped amor-phous emitter and c-Si substrate.For solar cells that feature a similar heterostructure back surfacefield,impressive energy conversion efficiencies exceeding21%have been reported.5 The role of the a-Si:H͑i͒buffer layer has been discussed in literature͑see,e.g.,Refs.6–12͒:It is known that suchfilms can yield outstanding surface passivation for c-Si surfaces,13 but also that growth of an epitaxial interface during a-Si:H͑i͒deposition is detrimental for heterojunction device performance.12For hot wire chemical vapor deposited ͑CVD͒a-Si:H,where no ion bombardment takes place, abrupt interfaces have been obtained either by limiting the deposition temperature T depo͑Ref.14͒or by terminating the c-Si surface with a SiN x monolayer prior to a-Si:Hdeposition.15The abruptness of the interface,i.e.,whetherinstant a-Si:H deposition on c-Si occurred without initialepitaxial growth,was in these studies determined either bytransmission electron microscopy͑TEM͒͑Refs.12,14,and15͒or by͑in situ͒spectroscopic ellipsometry͑SE͒,16forwhich mirror polished surfaces are desirable.To gain know-ledge about the electronic surface passivation properties ofthese interfaces,the most straightforward technique is bymeasuring the effective carrier lifetime␶eff of the samples. Such measurements are known to be extremely sensitive, allowing for detection of bulk defect densities as low as 109–1011cm−3in a simple,contactless technique at room temperature.17In this letter,we show that by low temperature͑up to260°C͒postdeposition annealing,the surface passivationquality of direct plasma enhanced͑PE͒CVD a-Si:H͑i͒films improves when the a-Si:H/c-Si interface is abrupt.This contrasts with the case when an epitaxialfilm has been grown at the interface,where the surface passivation quality is seen to degrade significantly by a similar annealing treat-ment.Consequently,we argue that annealing followed by carrier lifetime measurements allows accurate determination of the onset of epitaxial growth in an easy-to-use way which is not restricted to polished c-Si surfaces.For the experiments,300␮m thick relatively low resistivity͑ϳ3.0⍀cm͒boron-dopedfloat zone͑100͒͑FZ͒-Si͑p͒wafers have been used.Both surfaces of the sub-strates were mirror polished to eliminate the influence of substrate surface roughness on the passivation properties18 and to allow for SE measurements.For predeposition surface cleaning,the samples werefirst immersed in a ͑H2SO4:H2O2͒͑4:1͒solution for10min to grow a chemical oxide,which was followed by a rinse in de-ionized water. The oxide was then stripped off in a dilute HF solution͑5%͒for30s.After this the samples were immediately transferred to the load lock of the deposition system.Forfilm deposi-tion,a parallel plate direct PECVD reactor operated at radio frequency͑rf͒͑13.56MHz͒power was used,in which the samples were mounted at the top electrode.The electrode distance and diameter were respectively20and230mm.An undiluted SiH4flow of20SCCM͑SCCM denotes cubic cen-timeter per minute at STP͒was used and the chamber was maintained at low pressure͑0.5Torr͒.The value for T depo was varied from105to255°C.The rf power absorbed by the plasma was5W.This is the minimal power required to maintain a stable plasma at the given deposition conditions. To evaluate the surface passivation quality,identicalfilms of about50nm thick were deposited on both wafer surfaces. After deposition,the samples were consecutively annealed in a vacuum furnace͑30min,with annealing temperatures T ann ranging from120to260°C͒.In between the annealing steps,the value for␶eff of the samples was measured with a Sinton Consulting WCT-100quasi-steady-state photocon-ductance system,19operated in the so-called generalized mode.Since high quality FZ-Si wafers have been used throughout the experiments,the contribution of the bulk to the total recombination expressed by␶eff can be neglected.In such a case,the effective surface recombination velocity S eff, which value can be regarded as a direct measure for the passivation quality of thefilms present at the surfaces,maya͒Electronic mail:stefaan.dewolf@aist.go.jpAPPLIED PHYSICS LETTERS90,042111͑2007͒0003-6951/2007/90͑4͒/042111/3/$23.00©2007American Institute of Physics90,042111-1be approximated by S eff =d *͑2␶eff ͒−1,with d being the wafer thickness.17All reported values for ␶eff and S eff are evaluated at a constant minority carrier injection density,⌬n ,of 1.0ϫ1015cm −3.The thickness of the deposited films was ex situ determined by measuring ellipsometry spectra ͑␺,⌬͒using a Woollam M-2000rotating-compensator instrument.In order to verify the SE analyses,high-resolution ͑HR ͒cross-sectional TEM micrographs were acquired by a Jeol JEM2000EX system operated at an acceleration voltage of 200kV.Figure 1shows the imaginary part of the pseudodielec-tric function 20͗␧1͘+i ͗␧2͘of thin a -Si:H ͑i ͒films on c -Si wa-fers obtained from SE measurements at the initial deposition stages ͑here after 108s ͒and for different values of T depo .For reference,data for bare c -Si are given too.The figure dis-plays how samples featuring films for which T depo ജ205°C ͑group a in Fig.1͒show an evident crystalline signature,suggesting that those films were epitaxially grown.Con-versely,for samples with films for which T depo ഛ180°C ͑group b ͒,the value for ͗␧2͘deviates significantly from that of bare c -Si substrates,leading us to conclude that these films are most likely amorphous,as expected.Figure 2re-veals how the surface passivation quality of similar films,again for different values of T depo ,changes as a function of T ann .For the films deposited at the lowest temperature ͑T depo =105°C ͒,initially the passivation is poor,but im-proves to a remarkable extent by annealing.The surface pas-sivation quality of as-deposited films improves with higher T depo up to 180°C.Postannealing up to 260°C increases the values for ␶eff further,well in excess of 1ms.This situation is different for films deposited at 205°C,where annealing does not give rise anymore to an improvement,and for even higher values of T depo ,the figure shows that the passivation quality actually drastically goes down.Figure 3shows how the onset of epitaxial film growth and the phenomenon where postannealing of the samples results in passivation losses are related to each other.Part ͑a ͒of this figure gives the a -Si:H ͑i ͒film thickness,d bulk ,during the initial deposition stages,as function of T depo .These data are obtained from fitting the measured SE data of the different samples to a two-layer model,to take surface roughness on top of the a -Si:H film into account ͓see the inset in Fig.3͑a ͔͒.For allfilms deposited at T depo ജ205°C,hardly any film growth can be observed with SE during at least the first 72s.The difficulty to fit the initial film thickness to the SE data sug-gests that for the complete surface the deposited material is crystalline.21This area is crosshatched in the figure.From the slopes of the exponential fits in the Arrhenius plot as given in Fig.2,an activation energy E A can be extracted.This is given in Fig.3͑b ͒as function of T depo .The area with negative values for E A coresponds to films for which postannealing is harmful for the passivation quality and has been cross-hatched in the figure too.A good correspondence exists be-tween the onset where crystalline material is grown at the full interface during film deposition and the point where the value of E A becomes negative.Figure 4shows aHRTEMFIG.1.͑Color online ͒Imaginary part of the pseudodielectric function ͗⑀1͘+i ͗␧2͘for thin a -Si:H ͑i ͒films deposited at different temperatures on c -Si surfaces,obtained from ͑ex situ ͒SE measurements.The deposition time for all films was 108s.Group a shows data of samples featuring films with T depo ജ205°C,whereas for group b,T depo ഛ180°C.For reference,data for a bare c -Si substrate are giventoo.FIG.2.͑Color online ͒Influence of the postdeposition annealing temperature T ann on the surface passivation quality for PECVD a -Si:H ͑i ͒films deposited at different temperatures T depo .All deposition times were 12min,whereas annealing times were 30min.Starting from the T ann onset of annealing induced passivation changes,the shown lines are exponential fits of thedata.FIG.3.͑a ͒Calculated film thickness d bulk from ex situ SE measurements as function of T depo ,given for several deposition times t depo .The used two-layer model is shown in the inset.Crosshatched area shows films with epitaxially grown interface.͑b ͒Extracted activation energy E A as function of T depo for the films shown in Fig.2.Crosshatched area shows films for which the surface passivation degrades by annealing.The lines are guides for the eye.micrograph of one of the a -Si:H ͑i ͒/c -Si interfaces of the sample with films deposited at 230°C,as indicated in Fig.3͑b ͒.This micrograph confirms that indeed a substantial part of the film was epitaxially grown.The interpretation of E A may be that it represents an energy barrier for a relaxed a -Si:H ͑i ͒/c -Si interface.Since E A is extracted from ␶eff measurements,its value will mainly be determined by the density of electronically active defects of the film.These are the defects that are within reach of the c -Si minority carrier wave function.The impact of these de-fects is determined by their energetic position within the band gap and their electron and hole capture cross sections.This contrasts with electron spin resonance ͑ESR ͒measure-ments of a -Si:H ͑i ͒layers,which rather reveal paramagnetic defect densities in the bulk of relatively thick ͑typically a few micrometers ͒films deposited on quartz substrates.Nev-ertheless,for 4␮m thick films deposited at 25°C,Biegelsen et al.observed in their ESR studies activation energies of about 0.5eV for the decrease of the Si dangling bond density by postdeposition annealing at values for T ann up to 250°C.22Considering the differences in measurement tech-niques,a good correspondence exists between this value for E A and the ones obtained from ␶eff measurements for the films deposited at low T depo as presented in Fig.3͑b ͒.From this it must be concluded that for completely amorphous films deposited on c -Si surfaces,most likely amorphous net-work relaxation is taking place during the low temperature anneal,without any bond rupture,but which is capable of annealing out defects that are harmful for the passivation quality of the films.Consequently,films with E A =0eV cor-respond to fully relaxed a -Si:H/c -Si interfaces.The shift of the optimal value for T depo from about 205°C before anneal-ing to about 155°C after annealing ͑as seen in Fig.2͒may originate from a competition between a higher hydrogen content and a relaxed film network for high quality surface passivation.The origin of negative E A values for epitaxially grown interfaces is at present less understood.Since the applied annealing temperatures must be considered to be too low for Si–H bond rupture,a possible explanation may be that for as-grown epitaxial films,most of the hydrogen at the inter-face is not bonded but present as atomic H +,perhaps yielding field effect passivation of the surface rather than physicalinterface passivation.Subsequent annealing then leads to fast H +diffusion away from the interface into the c -Si matrix,resulting in passivation losses:e.g.,for an anneal at 260°C for 30min,whereas in a -Si:H ͑with D 0=1.17ϫ10−2cm 2s −1and E a =1.53eV ͒23the hydrogen diffu-sion depth is in the order of nanometers,in c -Si ͑with D 0=1.45ϫ10−3cm 2s −1and E a =0.497eV ͒24this is about five orders of magnitude larger.D 0is the preexponen-tial term,E a is the activation energy,and k is the Boltzmann constant in the expression for the diffusivity D ͑T ͒=D 0exp ͑−E a /kT ͒.In conclusion,in this letter we have shown that high quality passivation for c -Si by means of direct PECVD a -Si:H ͑i ͒deposition has a sharp processing boundary which is defined by epitaxial growth conditions.In the case that the interface is abrupt,the deposited films benefit tremendously from a low temperature anneal,which is believed to yield film relaxation.Moreover,we argued that annealing followed by carrier lifetime measurements is a valuable tool,which is not restricted to surfaces that are mirror polished,to deter-mine the abruptness of the c -Si/a -Si:H ͑i ͒interface.1H.Matsuura,T.Okuno,H.Okushi,and K.Tanaka,J.Appl.Phys.55,1012͑1984͒.2J.Symons,M.Ghannam, A.Neugroschel,J.Nijs,and R.Mertens,Solid-State Electron.30,1143͑1987͒.3H.Mimura and Y .Hatanaka,J.Appl.Phys.61,2575͑1987͒.4M.Tanaka,M.Taguchi,T.Matsuyama,T.Sawada,S.Tsuda,S.Nakano,H.Hanafusa,and Y .Kuwano,Jpn.J.Appl.Phys.,Part 131,3518͑1992͒.5M.Taguchi, A.Terakawa, E.Maruyama,and M.Tanaka,Prog.Photovoltaics 13,481͑2005͒.6M.W.M.van Cleef,J.K.Rath,F.A.Rubinelli,C.H.M.van der Werf,R.E.I.Schropp,and W.F.van der Weg,J.Appl.Phys.82,6089͑1997͒.7M.Taguchi,K.Kawamoto,S.Tsuge,T.Baba,H.Sakata,M.Morizane,K.Uchihashi,N.Nakamura,S.Kiyama,and O.Oota,Prog.Photovoltaics 8,503͑2000͒.8N.Jensen,R.M.Hausner,R. B.Bergmann,J.H.Werner,and U.Rau,Prog.Photovoltaics 10,1͑2002͒.9M.Schmidt,A.Schoepke,L.Korte,ch,and W.Fuchs,J.Non-Cryst.Solids 338,211͑2004͒.10H.Fujiwara and M.Kondo,Appl.Phys.Lett.86,032112͑2005͒.11S.De Wolf and G.Beaucarne,Appl.Phys.Lett.88,022104͑2006͒.12T.H.Wang,E.Iwaniczko,M.R.Page,D.H.Levi,Y .Yan,H.M.Branz,and Q.Wang,Thin Solid Films 501,284͑2006͒.13S.Dauwe,J.Schmidt,and R.Hezel,in Proceedings of the 29th IEEE Photovoltaic Specialists Conference ,New Orleans,Louisiana ͑IEEE,Pis-cataway,NJ,2002͒,p.1246.14J.Thiesen, E.Iwaniczko,K.M.Jones, A.Mahan,and R.Crandall,Appl.Phys.Lett.75,992͑1999͒.15Y .Yan,M.Page,T.H.Wang,M.M.Al-Jassim,H.M.Branz,and Q.Wang,Appl.Phys.Lett.88,121925͑2006͒.16D.H.Levi,C.W.Teplin,E.Iwaniczko,Y .Yan,T.H.Wang,and H.M.Branz,J.Vac.Sci.Technol.A 24,1676͑2006͒.17D.K.Schroder,Semiconductor Material and Device Characterization ,3rd ed.͑Wiley-Interscience,Hoboken,NJ,2006͒.18S.De Wolf,G.Agostinelli,G.Beaucarne,and P.Vitanov,J.Appl.Phys.97,063303͑2005͒.19R.A.Sinton and A.Cuevas,Appl.Phys.Lett.69,2510͑1996͒.20G.E.Jellison,in Handbook of Ellipsometry ,edited by H.G.Tompkins and E.A.Irene ͑Springer,Heidelberg,2005͒.21H.Fujiwara and M.Kondo,Appl.Phys.Lett.90,013503͑2007͒.22D.K.Biegelsen,R.A.Street,C.C.Tsai,and J.C.Knights,Phys.Rev.B 20,4839͑1979͒.23D.E.Carlson and C.W.Magee,Appl.Phys.Lett.33,81͑1978͒.24C.Herring,N.M.Johnson,and C.G.Van de Walle,Phys.Rev.B 64,125209͑2001͒.FIG.4.HRTEM micrograph of one of the interfaces of the sample deposited at 230°C as indicated in Fig.3.The PECVD a -Si:H ͑i ͒film was deposited on a mirror polished c -Si ͑100͒surface.The deposition time was 12min.。