AX-type defects in zinc-doped GaAs…1−x…P…x…on GaAs T.H.Gfroerer,1,a͒D.G.Hampton,1P.R.Simov,1and M.W.Wanlass2
不可磨灭的番号1Davidson College,Davidson,North Carolina28035,USA
2National Renewable Energy Laboratory,Golden,Colorado80401,USA
͑Received16February2010;accepted27April2010;published online30June2010͒
GaAsP alloys are potential candidates forϳ1.5to1.8eV photovoltaic converters in multijunction solar cells.We u thermally stimulated capacitance,deep level transient spectroscopy,and photocapacitance to characterize defects in p-type GaAs0.83P0.17and GaAs0.72P0.28grown lattice-mismatched on GaAs substrates.We obrve veral features typically associated with DX centers,including persistent photocapacitance,nonexponential thermally-activated capture and escape transients,and large Stokes shifts for optical thresholds.We u condary ion mass spectroscopy and capacitance versus voltage measurements to ascertain the sulfur and zinc doping profiles in the n+/p diodes.The dramatic decrea in the effective doping concentration with temperature in the unilluminated GaAs0.72P0.28diode and the magnitude of the capacitance change with illumination indicate that the defect concentration is comparable to the zinc doping,suggesting that zinc may facilitate the formation
of AX complexes in this alloy.©2010American Institute of Physics.͓doi:10.1063/1.3436590͔
I.INTRODUCTION
肺火旺吃什么药Technological improvement of the conversion efficiency in state-of-the-art,ultrahigh efficiency solar cells may de-pend,in part,on the development and refinement of high-band-gap alloys that can be grown on GaAs substrates.The band gap in alloys like GaAsP can be tuned to match design parameters that capitalize on the power available in the high-energy portion of the solar spectrum.However,the atomic spacing in GaAsP differs from that of the underlying GaAs, so a higher concentration of crystalline defects is expected in epitaxial GaAsP layers.The defects impair solar cell per-formance by trapping photoexcited charge carriers and pro-viding new pathways for the carriers to recombine rather than generate electricity.
An important type of defect in n-type III-V alloys,in-cluding GaAs͑1−x͒P͑x͒,1is the DX center.Characteristic features of this donor complex include thermally-activated nonexponential capture and escape transients into and out of the traps and augmented optical escape thresholds.The features are explained by lattice relaxation during the capture process,which produces a momentum-dependent energetic barrier to the occupied state.2Since photons carry little
mo-mentum relative to phonons,the optical energy required for escape can be much larger than the thermal threshold.It has also been shown theoretically that acceptors in large-band gap p-type miconductors can form similar complexes͑AX centers͒,but the theoretical work has focud on II-VI miconductors,3–5and the experimental evidence for the centers is spar.AX behavior has been reported in Mg-doped p-type GaN.6We report herein evidence for AX cen-ters in Zn-doped GaAs͑1−x͒P͑x͒,which becomes more pro-nounced as band-gap crossover͑from the⌫to X conduction band minimum͒is approached with increasing x.We note that band-gap crossover also plays an important role in the formation of DX centers in this alloy.7Our obrvation of similar behavior in Zn-doped Ga0.58In0.42P on GaAs will be reported elwhere.8
II.DEVICE DETAILS
Becau electrons have higher mobility,the collection efficiency of photoexcited minority carriers is augmented in n+/p solar cells,where the majority of the light is absorbed in the lightly-doped p-type ba.We u temperature-and illumination-dependent capacitance measurements to charac-terize defects in Zn-doped GaAs͑x͒P͑1−x͒grown on GaAs, where a step grade is ud between the substrate and the diode to reduce the concentration of lattice defects in the active layer of the device.
The GaInP/GaAsP heterostructures were grown by atmospheric-pressure metalorganic vapor-pha epitaxy on mi-insulating GaAs substrates oriented͑100͒2°toward ͗110͘.The multilayer heterostructures were grown in a con-tinuous quence at700°C in a hydrogen ambient.The GaAsP compositionally step-graded layers were grown using trimethylgallium,arsine,and phosphine as the primary pre-cursors,and diethylzinc͑DEZ͒as the p-doping precursor. During the growth of the graded layer in the xϭ0.28struc-ture,the vapor-pha͓V͔/͓III͔incread fromϳ4at the be-ginning toϳ16at the end as the P mole fraction was in-cread.The nϩ/p GaAsP diode was grown atop the graded layer using the same precursors and growth conditions as tho ud for thefinal step of the graded layer,with DEZ for p-type doping and hydrogen sulfide for n-type doping.GaInP minority-carrier confinement layers,included on either side of the GaAsP diode,were grown using triethylgallium,trim-ethylindium,and phosphine,with DEZ for p-type doping, hydrogen sulfide for n-type doping,and a vapor-pha ͓V͔/͓III͔ofϳ60.Thefinal n-GaAsP cap layer,ud for fa-cilitating the top contact to the diode,was doped heavily with sulfur.
a͒Electronic mail:tigfroerer@davidson.edu.
JOURNAL OF APPLIED PHYSICS107,123719͑2010͒
0021-8979/2010/107͑12͒/123719/4/$30.00©2010American Institute of Physics
107,123719-1
The diode structure in the two test systems is shown in Fig.1and condary ion mass spectroscopy ͑SIMS ͒mea-surements of the S and Zn concentrations in the x=0.28de-vice are shown in Fig.2.Since GaAs standards are ud for the SIMS relative nsitivity factors,the quantification of S and Zn in our GaAsP diodes may be somewhat inaccurate,but the deduced Zn concentrations are consistent with tho obtained via capacitance versus voltage ͑CV ͒measurements.All devices have an active area of 0.040cm 2.III.EXPERIMENTAL RESULTS
We u a Boonton 7200capacitance meter with a 1MHz test signal for all capacitance measurements.Temperature-dependent data from the GaAs 0.72P 0.28diode before and after broadband optical illumination are shown in Fig.3.The pre-cipitous decrea in capacitance with decreasing temperature beginning near 150K is due to the capture of holes into defect-related traps at the edge of the depletion region,which necessitates an expansion of the space-charge zone.The magnitude of the capacitance reduction indicates that the trap concentration is comparable to the zinc doping.When the
diode is illuminated with broadband light,the holes are op-tically activated out of the traps and the ca
江苏省淮北中学
pacitance returns to a level that is consistent with the more systematically reduced room temperature result.The low-temperature pho-tocapacitance is persistent,meaning that the capacitance will maintain this elevated level even when the illumination is removed.Persistent photocapacitance at low temperature is an unambiguous sign that an energetic barrier is obstructing thermal capture.
CV measurements on the cold ͑90K ͒diode before and after illumination indicate that the depletion width doubles ͑from about 0.5to 1.0m near zero bias ͒and the effective doping concentration is reduced by a factor of five ͑from ϳ1.5ϫ1016to ϳ3ϫ1015cm −3near zero bias ͒in the cold,unilluminated state.We u a linearly-extrapolated dielectric constant of 12.35for computing the concentrations in the GaAs 0.72P 0.28.In one-sided abrupt junctions,the depletion width is proportional to the square root of the doping con-centration,but the SIMS data ͑Fig.2͒shows sulfur diffusion into the ba,which results in compensation and a more graded junction.However,since the grading is much steeper on the heavily-doped n-side,capacitance changes are domi-nated by changes in depletion on the lightly-doped p-side and we u a one-sided junction treatment in our concentra-tion calculations.Figure 3also shows how the dark capaci-tance changes with temperature after broadband illumination.We obrve a decrea in capacitance with increasing tem-perature between
100and 125K as thermally activated cap-ture repopulates the traps.Above 125K,thermally activated escape ensues and the capacitance ris back to its nominal level.
This interpretation of the photocapacitance results is supported by complementary deep level transient spectros-copy ͑DLTS ͒measurements.We apply square waves,alter-
(Zn)GaAsP Step Grade
GaAs Substrate
p (Zn)GaAs(1-x)P(x)Ba (S)GaInP Window (S)GaAsP Conduction Layer
(Zn)GaInP Confinement /Conduction
x =0.17
n+(S)GaAsP Emitter x =0.28500Å500Å500Å500Å4µm 500Å10µm
7µm
0.1µm 4µm 0.15µm 500ÅFIG.1.Schematic of our test
structures ͑not to scale ͒.Nominal layer thick-ness for the GaAs 0.72P 0.28and GaAs 0.83P 0.17devices are given on the left and right,respectively.
FIG.2.͑Color online ͒SIMS data on the GaAs 0.72P 0.28device.While diffu-sion
of sulfur into the ba is clearly evident,the sulfur concentration is below the nsitivity of the measurement ͑ϳ1015cm −3͒at the edge of the zero-bias depletion region.
FIG.3.͑Color online ͒Capacitance vs temperature in the GaAs 0.72P 0.28de-vice.Under dark conditions,a precipitous decrea in capacitance is ob-rved with decreasing temperature.The drop can be reverd by a short period of illumination at low temperature.From this persistent photocapaci-tance state,increasing temperature in the dark produces another dip in the capacitance before the device returns to the high temperature condition.
nating between Ϫ10and Ϫ1V with 80and 800ms periods,to monitor the hole capture and escape.Figure 4shows thermally-activated capture and escape transients with expo-nential and stretched exponential 9͓C ͑t ͒=Ae −͑kt ͒d
我的书包作文
志愿活动策划案͔fits.The GaAs 0.83P 0.17diode appears to have nonexponential tran-sients with comparable rates,but the signals are small and difficult to analyze quantitatively.The nonexponential re-spon is another signature of local lattice reconfiguration during the capture and escape process.10An Arrhenius plot of the stretched exponential rates k for capture and escape are shown in Fig.5.For the escape analysis,a fixed ampli-tude A ,and stretching parameter d =0.33were ud.We ud the sam
e d for the capture analysis,but we found that a linearly decreasing amplitude with temperature was required to obtain good fits.It is interesting to note that the deduced capture rates are slower than the escape rates for the traps.This nonintuitive result may be due to the large electric field that is prent during the high rever bias escape measure-ment.
五年级上册数学期末考试卷
Figure 6shows how the photocapacitance respon de-pends on the energy of the incident light.For the measure-ments,we u a 20W tungsten halogen lamp coupled to an Optometrics grating monochromator,which produces ap-proximately 0.1mW of optical power with a full-width half-maximum bandwidth of approximately 10nm.A longpass filter with a cut-on wavelength of 950nm is ud to remove any cond order diffraction from the grating.Referring to Fig.6,we note that when the energy of the light is reduced,the optical escape rate decreas as expected,but the tran-sient amplitude is also reduced.In particular,the faster com-ponent of the signal is quenched at lower energies.Since the rate and amplitude of the respon depend strongly on the photon energy,we do not obtain a unique optical activation
energy in this system.The reduced amplitude with decreas-ing energy indicates that holes in some traps are not optically activated by the lower energy light.IV.MODEL
张治中将军
Our thermally-stimulated capacitance and DLTS mea-surements can be explained by an occupation-dependent
lat-
FIG.4.͑Color online ͒Reprentative fits of conventional and stretched exponentials to GaAs 0.72P 0.28transients measured at 162.5K ͑the escape results are shifted vertically for clarity ͒.The capture transient is measured as the bias on the diode is changed from Ϫ10to Ϫ1V and the escape transient is measured as the bias is returned to Ϫ10V
.
FIG.5.͑Color online ͒Arrhenius plot of the stretched exponential rates ud to fit the GaAs 0.72P 0.28transients.The deduced thermal activation energies are 0.37eV for capture and 0.39eV for
escape.
FIG.6.͑Color online ͒Photocapacitance respon of the GaAs 0.72P 0.28de-vice to monochromatic light at the optical energies indicated.The photon flux varies between approximately 2ϫ1016and 3ϫ1016cm −2s −1as we tune the light source through this range of energies.
tice configuration model like the one shown in Fig.7.Since our monochromatic photocapacitance results imply that the defects in our device have a range of optical activation en-ergies,we add a distribution of defect level energies to the more conventional DX center picture 11as shown.For a given defect level,the optical cross-ction ͑which controls the op-tical escape rate ͒depends on the energy of the incident light.Low energy light can only facilitate escape from the highest energy levels in the distribution,and the optical cross-ction for this transition will be small.High energy light will acti-vate escape from the full distribution of levels.And since the energy of this light will be well above the activation thresh-old for the highest energy defect levels,the optical cross-ction for the transitions will be large.The cross-ction decreas for lower levels where the optical energy ap-proaches the activation threshold.V.CONCLUSION
We have performed a variety of capacitance-bad mea-surements on Zn-doped GaAs ͑x ͒P ͑1−x ͒grown on GaAs.
We obrve a host of features that are usually associated with defect complexes having an occupation-dependent configu-ration.The magnitude of the capacitance change in the more phosphorous-rich alloy ͑GaAs 0.72P 0.28͒shows that the trap concentration is nearly as large as the zinc doping,suggest-ing that the zinc is playing a role in the formation of the defects.While donor-related configurational electron traps ͑i.e.,DX centers ͒have received a great deal of attention in the literature,we find little discussion of their counterpart in p-type material,namely,acceptor-related configurational hole traps.But if our interpretation involving AX centers is correct,device engineers may benefit from considering other elements for p-type doping in GaAsP/GaAs structures,par-ticularly at high phosphorous concentration.ACKNOWLEDGMENTS
The authors would like to thank J.J.Carapella for per-forming the MOVPE growth and processing the devices.Ac-knowledgment is made to the Davidson Rearch Initiative,funded by The Duke Endowment,and the donors of the American Chemical Society—Petroleum Rearch Fund for support of this rearch.
1
E.Calleja,
F.J.Sanchez,E.Munoz,E.Vigil,F.Omnes,P.Gibart,J.M.Martin,and
G.Gonzalez Diez,Phys.Rev.B 53,7736͑1996͒.2
D.J.Chadi and K.J.Chang,Phys.Rev.Lett.61,873͑1988͒.3
D.J.Chadi and K.J.Chang,Appl.Phys.Lett.55,575͑1989͒.4
C.H.Park and
D.J.Chadi,Phys.Rev.Lett.75,1134͑1995͒.5
C.H.Park,S.B.Zhang,and S.-H.Wei,Phys.Rev.B 66,073202͑2002͒.6
J.Z.Li,J.Y .Lin,H.X.Jiang,A.Salvador,A.Botchkarev,and H.Morkoc,Appl.Phys.Lett.69,1474͑1996͒.7
E.Calleja,E.Munoz,B.Jimenez,A.Gomez,
F.Garcia,and F.Kellert,J.Appl.Phys.57,5295͑1985͒.8
T.H.Gfroerer,D.G.Hampton,and M.W.Wanlass,Proceedings of the 35th IEEE Photovoltaics Specialists Conference,22June 2010͑to be pub-lished ͒.9
S.I.Simdyankin,N.Mousau,and E.R.Hunt,Phys.Rev.E 66,066205͑2002͒.10
I.D.Henning and H.Thomas,Solid-State Electron.25,325͑1982͒.11
D.V .Lang and R.A.Logan,Phys.Rev.Lett.39,635͑1977͒
.
Enlarged View of Optical Cross Sections
High Energy Energy Light
Arbitrary Configuration Coordinate
E n e r g y
FIG.7.͑Color online ͒Valence and defect bands as a function of the lattice distortion ͑which is reprented graphically by an arbitrary configuration coordinate ͒in the occupation-dependent model.The circular symbols show the variation in optical cross ction with changing photon and/or defect level energy.
卤什么好吃
