Heterostructures with CdTe and CdTe1-xSex (x ∼ 0.01) absorbers between two wider-band-gap Cd1-xMgxTe barriers (x ∼ 0.25–0.3) were grown by molecular beam epitaxy to study carrier generation and recombination in bulk materials with passivated interfaces. Using a combination of confocal photoluminescence (PL), time-resolved PL, and low-temperature PL emission spectroscopy, two extended defect types were identified and the impact of these defects on charge-carrier recombination was analyzed. The dominant defects identified by confocal PL were dislocations in samples grown on (211)B CdTe substrates and crystallographic twinning-related defects in samples on (100)-oriented InSb substrates. Low-temperature PL shows that twin-related defects have a zero-phonon energy of 1.460 eV and a Huang-Rhys factor of 1.50, while dislocation-dominated samples have a 1.473-eV zero-phonon energy and a Huang-Rhys factor of 1.22. The charge carrier diffusion length near both types of defects is ∼6 μm, suggesting that recombination is limited by diffusion dynamics. For heterostructures with a low concentration of extended defects, the bulk lifetime was determined to be 2.2 μs with an interface recombination velocity of 160 cm/s and an estimated radiative lifetime of 91 μs.

Model systems utilizing epitaxial films can provide critical insights into material properties and performance limitations for solar cells and other electronic devices. Double heterostructures (DHs) are used to study bulk properties with well-passivated interfaces and have proven to be very useful for the development of III–V materials. For example, Olson et al. achieved low recombination velocity and high lifetimes in GaAs1 that quickly led to high-efficiency tandem solar cells.2 Progress for II–VI semiconductors, on the other hand, has been much slower. Recently, however, several groups have shown that CdTe DHs grown by molecular beam epitaxy (MBE) have interface recombination velocities comparable to GaAs/AlGaAs DHs.3,4 Defect analysis based on intensity-dependent photoluminescence (PL) revealed interface state density as low as 1010 cm−2 for CdTe.5 Reports on II–VI DHs also showed that charge-carrier lifetimes were limited by Shockley-Read-Hall (SRH) recombination5,6 but lifetime-limiting defects were not identified.

Epitaxial CdTe can have various types of extended defects with densities varying from 104 to 108 cm−2, depending on substrate surface preparation and lattice mismatch for heteroepitaxy. Both theory and experiment have shown that CdTe growth on (100) surfaces can promote the formation of twins whereas growth on (211) suppresses twins.5,7–11 In this letter, we use photoluminescence (PL) to identify the critical concentration of extended defects that dominates Shockley-Read-Hall recombination and limits charge carrier lifetime in epitaxial films and to show that recombination at extended structural defects begins to limit carrier lifetimes when defect density is above 106 cm−2. We also explore the recombination rates, lifetimes, and low-temperature PL emission spectra in films with various types of defects. Growth of high-lifetime, low-defect epitaxial CdTe and CdSeTe demonstrate that II–VI materials have excellent electronic qualities that, in principle, should allow the development of higher-performing semiconductor devices. This is particularly critical for CdTe solar cells, where efficiency is limited by the open-circuit voltage, which depends strongly on the minority-carrier lifetime in the bulk material.

DHs used in this study were grown by MBE on InSb (100) and CdTe (211)B substrates at Texas State University.12 For the InSb substrates, an InSb buffer was grown before a thin CdTe buffer layer. Each DH consisted of a 0.5–1.5-μm-thick CdTe buffer, a 30-nm CdMgTe layer, a 0.5–1.5-μm thick CdTe absorber layer, a 30-nm CdMgTe layer, and a 10-nm CdTe capping layer. The incorporation of 0.8% Se in the CdTe absorber and buffer for some of the samples provided better lattice matching with the InSb substrates and led to an increase in critical thickness beyond ∼3 μm, allowing for thicker layers without strain relaxation.13,14

After growth, confocal PL (c-PL) imaging (Olympus confocal microscope) was used to quantify defect density and distinguish defect types.14,15 A 635-nm laser diode was used to illuminate the samples. Pinholes placed in the optical path restricted out-of-focus luminescence, significantly enhancing both lateral and depth resolution. A 60× magnification oil-immersion PlanApo objective lens with a numerical aperture of 1.40 was used to obtain a lateral resolution of 0.35 μm and a depth resolution of 0.25–0.50 μm.16 To measure the defect density (ρd), c-PL images were processed with the ImageJ software package.17 The quantified ρd is very similar to that measured by cathodoluminescence but with faster data acquisition. This approach also yields more reproducible data than etch-pit techniques.18 

Low-temperature photoluminescence (LTPL) emission spectra were then acquired at 4 K using a continuous-wave HeNe laser at 632.8 nm and an excitation intensity of 0.2–2 W/cm2. Charge-carrier lifetime measurements were made using time-resolved photoluminescence (TRPL) with a regenerative, amplified Yb:KGW laser system with an optical parametric amplifier set at 640 nm with 0.3-ps pulses at a 1.1-MHz repetition rate, an avalanche photodiode detector, and time-correlated single photon counting. Excitation intensity varied but remained at low injection conditions (less than 1012 photons/pulse/cm2). Single-exponential TRPL lifetimes, τeff, were independent of excitation intensity and therefore show minority-carrier lifetimes.

The c-PL data for our DHs show two types of extended defects. (211)B substrates had a high defect density (ρd > 107 cm−2) with the majority of these defects attributed to individual dislocations. Twin-related defects dominated on (100)-oriented substrates and had lower defect densities (104–107 cm−2). CdTe and its ternary alloys have traditionally been grown on (211) substrates to suppress twin formation and facilitate heteroepitaxy.5,7–11 But dislocation density was initially very high for samples on (211)B substrates. We attributed this to poor interface quality and a need for better substrate preparation which, when implemented, led to improved growth. On the other hand, polar (100) surfaces have much higher surface energy than non-polar surfaces and act as nucleation centers for twin-related defects.11 Twinning during growth on (100)-oriented substrates occurs about a ⟨111⟩ axis, resulting in inclined twin lamella.8 Bonding across the twin plane is satisfied at the atomic level, resulting in no significant recombination centers. However, bonds cannot be satisfied at the edges of the twin lamella, resulting in paired dislocation/extended defect formation with resulting non-radiative recombination as seen in the signatures in Figure 1(a). Figure 1(b) shows an atomic-force microscopy (AFM) scan of a (100)-oriented layer grown under conditions enhancing growth of the twin lamellae compared to the epitaxial layer. The intersection of the lamella edges leads to the observed dark spot “couplets” that are characteristic of twinning. This is analogous to paired dislocations seen at the edges of smaller stacking faults (as seen by transmission electron microscopy (TEM)19).

FIG. 1.

(a) A c-PL image and (b) an AFM micrograph of a 2-μm thick CdTe/CdMgTe DH grown on InSb (100) with large twin density. Double dark spots are the intersection of twin lamellae with the matrix. The height range of the micrograph for the AFM is ∼8 to 12 nm.

FIG. 1.

(a) A c-PL image and (b) an AFM micrograph of a 2-μm thick CdTe/CdMgTe DH grown on InSb (100) with large twin density. Double dark spots are the intersection of twin lamellae with the matrix. The height range of the micrograph for the AFM is ∼8 to 12 nm.

Close modal

Localized dark features in c-PL images in Figure 2 indicate non-radiative recombination at extended defects. The feature size is related to the rate of carrier diffusion towards the defect in combination with the recombination rate at the defect.20,21 The recombination rate can be evaluated using the contrast ratio C(r). Chen et al. have shown that for local excitation and local collection of luminescence signals, in the case of a fixed photocarrier lifetime τ and diffusion coefficient D, the contrast near an extended defect can be found from22 

C(r)=I0I(r)I=K02(r/Ld)K02¯(0),
(1)

where I0 is the PL signal from a homogeneous area, I(r) is the signal at a distance r from the defect, K0 is a modified Bessel function, K0¯ is the function's average within a laser spot, and the diffusion length is Ld=Dτ. Fitting the data in Figure 2(d) to Equation (1) suggests Ld = 6.3 ± 0.3 μm near dislocations and Ld = 6.2 ± 0.3 μm for twin-related defects. The similar values suggest that the recombination rate of both types of defects is large enough that recombination is limited by diffusion dynamics, with the defects' effect on overall recombination only limited by how quickly the carriers can diffuse to the defect. Because the c-PL measurements were taken under high optical injection conditions (with an excess carrier density around 3 × 1018/cm3), this value is not representative of the minority-carrier diffusion coefficient and the assumptions used in deriving Equation (1) are not strictly valid. However, as shown here and noted by Chen et al.,22 the functional form fits the distribution well with Ld as an effective diffusion length, and the excess carrier distribution will have the form

δn(r)=δp(r)(G2πD)K0(rLd),
(2)

with G being the generation rate.

FIG. 2.

(a) Dislocation-dominant and (b) twin-dominant c-PL images for CdTe epilayers on (211)B and (100) substrates. The scale is the same in both images. (c) High-magnification images and linear cross-sections of the PL signal for each feature were taken. (d) The contrast ratio C(r) for both defect types were fit to Eq. (1) to determine charge-carrier diffusion length.

FIG. 2.

(a) Dislocation-dominant and (b) twin-dominant c-PL images for CdTe epilayers on (211)B and (100) substrates. The scale is the same in both images. (c) High-magnification images and linear cross-sections of the PL signal for each feature were taken. (d) The contrast ratio C(r) for both defect types were fit to Eq. (1) to determine charge-carrier diffusion length.

Close modal

The variation of the dominant defect type allowed the use of LTPL emission spectra to identify the defect-related spectroscopic transitions and electron-phonon coupling. In addition to exciton peaks (1.57–1.60 eV)23–25 and emission due to donor-acceptor pairs and point defects (∼1.55 eV),23–25 LTPL data contained a Y-luminescence band due to extended defects26–28 with a zero phonon line (ZPL) between 1.45 eV and 1.50 eV and longitudinal (LO) phonon replicas with a 21-meV separation.27,29 To determine the ZPL energy and the electron-phonon coupling, which was distinctly different for samples with different dominant defects, we fit these data with Gaussian functions.23,24 This analysis, shown in Figure 3, indicates a ZPL energy of 1.473 eV for dislocation-dominant samples on CdTe (211)B substrates and 1.460 eV for twin-related defect-dominant films on InSb (100) substrates. The ZPL energy for the dislocation-dominated sample is in excellent agreement with correlative PL and transmission electron microscopy (TEM) analysis.28 In contrast, the ZPL energy for twins was 13 meV lower. The intensity of Gaussian sub-bands I(n) was also fit to a Poisson distribution in the Huang-Rhys model30 

I(n)=I0eSHRSHRn/n!,
(3)

where I0 is the scaling factor, SHR is the Huang-Rhys factor, and n indicates the number of phonon replicas emitted for each sub-band. Fits give SHR = 1.22 for dislocations and SHR = 1.50 for twin-related defects, which is in good agreement with previous studies on single-crystal CdTe (SHR = 1.58 (Ref. 31) and 1.60 (Ref. 32)). The Huang-Rhys factor is sensitive to the lattice structure, and similar coupling between the excitons and phonons is expected in single crystals and lattice-matched epilayers.

FIG. 3.

Low-temperature PL emission spectra (4 K) of the extended defect region for CdTe DHs—twin-dominant on top (ρd = 8.4 × 104 cm−2, circles in inset) and dislocation-dominant on the bottom (ρd = 1.5 × 106 cm−2, squares). Intensity of Gaussian sub-bands as a function of phonon replica n is fit to the Huang-Rhys model in the inset.

FIG. 3.

Low-temperature PL emission spectra (4 K) of the extended defect region for CdTe DHs—twin-dominant on top (ρd = 8.4 × 104 cm−2, circles in inset) and dislocation-dominant on the bottom (ρd = 1.5 × 106 cm−2, squares). Intensity of Gaussian sub-bands as a function of phonon replica n is fit to the Huang-Rhys model in the inset.

Close modal

Next, to consider the impact of extended defects on charge-carrier recombination, we compared charge-carrier lifetimes as measured by TRPL to total defect density ρd determined by c-PL. Figure 4(a) shows several of the longest-lifetime TRPL decays for CdTe and CdSeTe DHs. The 2.5-μm-thick CdSeTe DH on InSb with a 640 ns measured lifetime had a defect density ρd = 7.0 × 104 cm−2, the 1-μm-thick CdTe on InSb had a 240 nm lifetime and ρd = 5.8 × 105 cm−2, and the 1.5–μm-thick CdTe on CdTe had a 42 ns lifetime and ρd = 1.6 × 106 cm−2. For the samples with sufficiently low defect density, ρd < 5 × 105 cm−2, interface recombination dominated and recombination rates in the bulk (1/τbulk) and at the interfaces (2S/d) were separated from33 

1/τeff=1/τbulk+2S/d,
(4)

where τeff is the effective lifetime, S is the interface recombination velocity, and d is the absorber thickness. Measured effective lifetimes for samples with ρd < 5 × 105 cm−2, which excluded the CdTe on CdTe substrates, fit to Equation (4) yielded τbulk = 2.2 ± 0.3 μs and S = 160 ± 10 cm/s as seen in Figure 4(b).

FIG. 4.

(a) TRPL decay curves for CdTe and CdSeTe DHs with the highest measured lifetimes for each sample type. (b) Effective recombination rate of samples with ρd below 1 × 106 cm−2 plotted against 2/d used to extract τbulk (2.2 μs) and interface recombination velocity (160 cm/s).

FIG. 4.

(a) TRPL decay curves for CdTe and CdSeTe DHs with the highest measured lifetimes for each sample type. (b) Effective recombination rate of samples with ρd below 1 × 106 cm−2 plotted against 2/d used to extract τbulk (2.2 μs) and interface recombination velocity (160 cm/s).

Close modal

CdTe typically has a background dopant concentration on the order of NA ≈ 1014 cm−3 (Ref. 34) and a radiative recombination coefficient from B = (2–4) × 10−9 cm−3 s−1 (Refs. 35–37) to B = (1–2) × 10−10 cm−3 s−1.5,38–40 Using B = 1.1 × 10−10 cm3/s,5,40 an estimated radiative lifetime in samples with d = 1 μm can range from τrad = 1/((1−γ)B NA) ≈ 91–455 μs, where the lower value assumes no photon recycling (γ = 0) and the higher value is for photon recycling γ = 80%.41 Therefore, most recombination can be attributed to non-radiative Shockley-Read-Hall (SRH) processes.42 And because 2S/d ≫ 1/τbulk, most SRH recombination occurred at the interfaces. Similar results were also obtained from temperature-dependent TRPL analysis6 and excitation-dependent PL intensity measurements.5 Despite non-radiative recombination being a significant loss mechanism, these values represent state-of-the-art II–VI DHs and indicate excellent bulk material and interface quality.3,42

We note that the data in Figure 4(b) are for samples with ρd < 1 × 106 cm−2, which is adequate for analysis that considers only interface and bulk recombination. Above ρd ∼ 1 × 106 cm−2, the overall recombination rate 1/τeff increases rapidly at higher ρd. This is indicated by the decreased lifetime shown for the CdTe/CdTe sample in Figure 4(a), which had a ρd = 1.6 × 106 cm−2. For DHs with high defect density, τeff decreased by almost a factor of 100 as ρd approached 1 × 107 cm−2.

In summary, recombination at extended defects in undoped CdTe and CdSeTe DHs grown on CdTe and InSb substrates with (211)B and (100) orientations, respectively, was studied to determine the impact of defects on bulk recombination. Confocal PL shows that dislocations are the dominant defect type in films grown on (211)B substrates, twin-related defects dominate for (100), and the ambipolar diffusion length is 6 μm for both. LTPL spectra can be used to differentiate between dominant defect types. Dislocation-dominant films have a zero-phonon line at 1.473 eV and a Huang-Rhys factor of 1.22, and films dominated by twin-related defects have a ZPL at 1.460 eV with a Huang-Rhys factor of 1.50. From c-PL and TRPL measurements, we determined that extended defects dominate recombination at a defect density above 106 cm−2. Below this limit, carrier lifetimes and diffusion lengths are not limited by extended defects and most recombination occurs at the interfaces. A bulk lifetime of 2.2 μs and an interface recombination velocity of 160 cm/s were extracted from TRPL data, with epitaxial CdTe and CdSeTe films exhibiting lifetimes that are sufficient for high efficiency solar cells and electronic devices. Similarly, high-quality single-crystal bulk materials have enabled recent breakthrough in CdTe solar cell voltage,40 providing a foundation for the physical understanding of other II–VI electronic devices.

The authors would like to thank Jim Sites for many useful discussions. This research was supported by the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, under Contract No. DE-AC36-08GO28308. The U.S. Government retains and the publisher, by accepting the article for publication, acknowledges that the U.S. Government retains a nonexclusive, paid up, irrevocable, worldwide license to publish or reproduce the published form of this work, or allow others to do so, for U.S. Government purposes.

1.
J. M.
Olson
,
R. K.
Ahrenkiel
,
D. J.
Dunlavy
,
B.
Keyes
, and
A. E.
Kibbler
,
Appl. Phys. Lett.
55
,
1208
(
1989
).
2.
J. M.
Olson
,
S. R.
Kurtz
,
A. E.
Kibbler
, and
P.
Faine
, in
21st IEEE Photovoltaic Specialists Conference, Kissimmee, FL
(
1990
), p.
24
.
3.
M. J.
DiNezza
,
X.-H.
Zhao
,
S.
Liu
,
A. P.
Kirk
, and
Y.-H.
Zhang
,
Appl. Phys. Lett.
103
,
193901
(
2013
).
4.
S.
Liu
,
X.-H.
Zhao
,
C.
Campbell
,
M. J.
DiNezza
,
Y.
Zhao
, and
Y.-H.
Zhang
,
J. Vac. Sci. Technol. B
33
,
011207
(
2015
).
5.
C. H.
Swartz
,
M.
Edirisooriya
,
E. G.
Leblanc
,
O. C.
Noriega
,
P. A. R. D.
Jayathilaka
,
O. S.
Ogedengbe
,
B. L.
Hancock
,
M.
Holtz
,
T. H.
Myers
, and
K. N.
Zaunbrecher
,
Appl. Phys. Lett.
105
,
222107
(
2014
).
6.
X. H.
Zhao
,
M. J.
Dinezza
,
S.
Liu
,
P. A. R. D.
Jayathilaka
,
O. C.
Noriega
,
T. H.
Myers
, and
Y. H.
Zhang
, in
40th IEEE Photovoltaic Specialists Conference, Denver, CO
(
2014
)
, pp.
3272
3275
.
7.
A.
Million
,
J. Vac. Sci. Technol., A
6
,
2813
(
1988
).
8.
R. J.
Koestner
and
H. F.
Schaake
,
J. Vac. Sci. Technol., A
6
,
2834
(
1988
).
9.
A. G.
Cullis
,
N. G.
Chew
,
J. L.
Hutchison
,
S. J. C.
Irvine
, and
J.
Giess
, in
Microscopy of Semiconductor Confernce, Oxford
(
1985
), pp.
29
34
.
10.
A. W.
Vere
,
S.
Cole
, and
D. J.
Williams
,
J. Electron. Mater.
12
,
551
(
1983
).
11.
W. J.
Yin
,
J. H.
Yang
,
K.
Zaunbrecher
,
T.
Gessert
,
T.
Barnes
,
Y.
Yan
, and
S. H.
Wei
,
Appl. Phys. Lett.
107
,
141607
(
2015
).
12.
J.
Chai
,
K. K.
Lee
,
K.
Doyle
,
J. H.
Dinan
, and
T. H.
Myers
,
J. Electron. Mater.
41
,
2738
(
2012
).
13.
C.
Fontaine
,
J. P.
Gailliard
,
S.
Magli
,
A.
Million
, and
J.
Piaguet
,
Appl. Phys. Lett.
50
,
903
(
1987
).
14.
O. C.
Noriega
,
A.
Savage
,
T. H.
Myers
,
P. J.
Smith
,
R. N.
Jacobs
,
C. M.
Lennon
,
P. S.
Wijewarnasuriya
, and
Y.
Chen
, in
II–VI Workshop Chicago
(
2013
).
15.
E. G.
Leblanc
,
P. A. R. D.
Jayathilaka
,
M.
Edirisooriya
,
O. S.
Ogedengbe
,
C.
Swartz
,
O. C.
Noriega
, and
T. H.
Myers
, in
SunShot Workshop Anaheim
, CA (
2014
).
16.
D. W.
Piston
,
Biol. Bull.
195
,
1
(
1998
).
17.
W. S.
Rasband
,
ImageJ
(
U. S. National Institutes of Health
,
Bethesda, MD
,
1997–2015
).
18.
J.
Chai
,
O. C.
Noriega
,
A.
Dedigama
,
J. J.
Kim
,
A. A.
Savage
,
K.
Doyle
,
C.
Smith
,
N.
Chau
,
J.
Pena
,
J. H.
Dinan
,
D. J.
Smith
, and
T. H.
Myers
,
J. Electron. Mater.
42
,
3090
(
2013
).
19.
C.
Li
,
J.
Poplawsky
,
Y.
Wu
,
A. R.
Lupini
,
A.
Mouti
,
D. N.
Leonard
,
N.
Paudel
,
K.
Jones
,
W.
Yin
,
M.
Al-Jassim
,
Y.
Yan
, and
S. J.
Pennycook
,
Ultramicroscopy
134
,
113
(
2013
).
20.
T. H.
Gfroerer
,
Y.
Zhang
, and
M. W.
Wanlass
,
Appl. Phys. Lett.
102
,
012114
(
2013
).
21.
T. H.
Gfroerer
,
C. M.
Crowley
,
C. M.
Read
, and
M. W.
Wanlass
,
J. Appl. Phys.
111
,
093712
(
2012
).
22.
F.
Chen
,
Y.
Zhang
,
T. H.
Gfroerer
,
A. N.
Finger
, and
M. W.
Wanlass
,
Sci. Rep.
5
,
10542
(
2015
).
23.
C.
Kraft
,
H.
Metzner
,
M.
Hädrich
,
U.
Reislöhner
,
P.
Schley
,
G.
Gobsch
, and
R.
Goldhahn
,
J. Appl. Phys.
108
,
124503
(
2010
).
24.
D. S.
Albin
,
D.
Kuciauskas
,
J.
Ma
,
W. K.
Metzger
,
J. M.
Burst
,
H. R.
Moutinho
, and
P. C.
Dippo
,
Appl. Phys. Lett.
104
,
092109
(
2014
).
25.
D. P. P.
Halliday
,
M. D. G.
Potter
,
J. T.
Mullins
, and
A. W.
Brinkman
,
J. Cryst. Growth
220
,
30
(
2000
).
26.
K.
Zanio
, in
Semiconductors and Semimetals
, edited by
R. K.
Willarson
and
A. C.
Beer
(
Academic
,
New York
,
1972
), Vol.
13
.
27.
P. J.
Dean
,
G. M.
Williams
, and
G.
Blackmore
,
J. Phys. D: Appl. Phys.
17
,
2291
(
1984
).
28.
S.
Hildebrandt
,
H.
Uniewski
,
J.
Schreiber
, and
H. S.
Leipner
,
J. Phys. III France
7
,
1505
(
1997
).
29.
E.
Molva
,
K.
Saminadayar
,
J. L.
Pautrat
, and
E.
Ligeon
,
Solid State Commun.
48
,
955
(
1983
).
30.
K.
Huang
and
A.
Rhys
,
Proc. R. Soc. A
204
,
406
(
1950
).
31.
D.
Kuciauskas
,
P.
Dippo
,
A.
Kanevce
,
Z.
Zhao
,
L.
Cheng
,
A.
Los
,
M.
Gloeckler
, and
W. K.
Metzger
,
Appl. Phys. Lett.
107
,
234906
(
2015
).
32.
W.
Stadler
,
D.
Hofmann
,
H.
Alt
,
T.
Muschik
,
B.
Meyer
,
E.
Weigel
,
G.
Müller-Vogt
,
M.
Salk
,
E.
Rupp
, and
K.
Benz
,
Phys. Rev. B
51
,
10619
(
1995
).
33.
R. K.
Ahrenkiel
, in
Semiconductors and Semimetals
, edited by
R. K.
Ahrenkiel
and
M. S.
Lundstrom
(
Elsevier Science Publishing Co. Inc.
,
1993
), Vol.
39
, pp.
40
146
.
34.
X.-H.
Zhao
,
S.
Liu
,
Y.
Zhao
,
C. M.
Campbell
,
M. B.
Lassise
,
Y.-S.
Kuo
, and
Y.-H.
Zhang
,
IEEE J. Photovoltaics
6
,
552
(
2016
).
35.
R.
Cohen
,
V.
Lyahovitskaya
,
E.
Poles
,
A.
Liu
, and
Y.
Rosenwaks
,
Appl. Phys. Lett.
73
,
1400
(
1998
).
36.
R. K.
Ahrenkiel
,
B. M.
Keyes
,
D. L.
Levi
,
K.
Emery
,
T. L.
Chu
, and
S. S.
Chu
,
Appl. Phys. Lett.
64
,
2879
(
1994
).
37.
X.-H.
Zhao
,
M. J.
DiNezza
,
S.
Liu
,
S.
Lin
,
Y.
Zhao
, and
Y.-H.
Zhang
,
J. Vac. Sci. Technol. B
32
,
040601
(
2014
).
38.
A. E.
Rakhshani
,
J. Appl. Phys.
81
,
7988
(
1997
).
39.
A. P.
Kirk
,
M. J.
DiNezza
,
S.
Liu
,
X.-H.
Zhao
, and
Y.-H.
Zhang
, in
39th IEEE Photovoltaic Specialists Conference, Tampa, FL
(
2013
), p.
2515
.
40.
J. M.
Burst
,
J. N.
Duenow
,
D. S.
Albin
,
E.
Colegrove
,
M. O.
Reese
,
J. A.
Aguiar
,
C.-S.
Jiang
,
M. K.
Patel
,
M. M.
Al-Jassim
,
D.
Kuciauskas
,
S.
Swain
,
T.
Ablekim
,
K. G.
Lynn
, and
W. K.
Metzger
,
Nat. Energy
1
,
16015
(
2016
).
41.
P.
Asbeck
,
J. Appl. Phys.
48
,
820
(
1977
).
42.
X.-H.
Zhao
,
M. J.
DiNezza
,
S.
Liu
,
S.
Lin
,
Y.
Zhao
, and
Y.-H.
Zhang
,
Appl. Phys. Lett.
105
,
252101
(
2014
).