Group-V element doping is promising for simultaneously maximizing the hole concentration and minority carrier lifetime in CdTe for thin film solar cells, but there are roadblocks concerning point defects including the possibility of self-compensation by AX metastability. Herein, we report on doping, lifetime, and mobility of CdTe single crystals doped with As between 1016 and 1020 cm−3 grown from the Cd solvent by the travelling heater method. Evidence consistent with AX instability as a major contributor to compensation in samples doped below 1017 cm−3 is presented, while for higher-doped samples, precipitation of a second phase on planar structural defects is also observed and may explain spatial variation in properties such as lifetime. Rapid cooling after crystal growth increases doping efficiency and mobility for times up to 20–30 days at room temperature with the highest efficiencies observed close to 45% and a hole mobility of 70 cm2/Vs at room temperature. A doping limit in the low 1017/cm3 range is observed for samples quenched at 200–300 °C/h. Bulk minority carrier lifetimes exceeding 20 ns are observed for samples doped near 1016 cm−3 relaxed in the dark and for unintentionally doped samples, while a lifetime of nearly 5 ns is observed for 1018 cm−3 As doping. These results help us to establish limits on properties expected for group-V doped CdTe polycrystalline thin films for use in photovoltaics.

CdTe is the leading absorber material for thin film solar cells in commercial production, and record devices have exceeded 22%.1 This demonstrated efficiency still lags the theoretical limit primarily because of low open circuit voltage (Voc) which can be explained by a low p-type carrier concentration of typically 1014–1015 cm−3 and short minority carrier lifetime less than several nanoseconds.2,3 Therefore, increasing p-type doping and minority carrier lifetime is priority for increasing CdTe device efficiency to 25% and above.3 The hole concentration >1017 cm−3 and lifetime >10 ns are expected to lead the device efficiency of 25% and VOC of more than 1 V. Additionally, if the depletion width is decreased by increasing doping, photocarrier mobility will additionally play a role in maintaining high short circuit current (Jsc) because of more stringent requirements on the minority carrier diffusion length.

Recently, the combination of group-V doping and Cd-rich stoichiometry has attracted attention as a means to simultaneously increase p-type doping and photocarrier lifetime.4 P-type doping with substitutional Cu (CuCd) typically suffers from low activation because of self-compensation by Cu interstitials (Cui).5 Te-rich conditions not only promote higher p-type doping with CuCd but also result in undesirable native defects such as Te on Cd antisites (TeCd) and Te interstitials (Tei) which are predicted to act as strong recombination centers. Additionally, the upper state of the double acceptor Cd vacancy (VCd) has also been shown to participate in recombination.6–9 Thus, Cd-rich defect chemistry is believed to result in fewer non-radiative recombination centers.

Group-V elements (N, P, As, Sb, and Bi) introduce acceptor states by substitution of Te sites with ionization energies of nearly (respectively) 60, 70, 90, 260, and 300 meV.10 Although it has lower ionization energy, incorporating N into II-VI group compounds is difficult. Sb and Bi are known to have very low activation at operating temperature due to their ionized energy more than 200 meV. Therefore, As and P are good candidates for p-type doping under Cd-rich conditions. The As atom diffuses slowly through the Te sublattice due to its larger atomic size than P.11 The post diffusion process such as annealing is needed for the doping in the CdTe thin film, but it is difficult for the P source to control equilibrium conditions because of its uncertainty in vapor pressure.12 Therefore, As seems to be a more appropriate dopant because it is suitable for high volume manufacturing compared to P.

However, p-type doping is difficult to achieve in CdTe because of the introduction of deep defect levels and self-compensation.5,9,13,14 Additionally, self-compensation of AX metastability by large lattice relaxation of a substitutional group-V element (e.g., AsTe) with conversion to a donor state13–16 is predicted to limit p-type below approximately 1016 cm−3. Recently, a number of experimental studies have confirmed that group-V doping under Cd-rich conditions can increase the hole concentration.4,14,17,18 Also, evidence of metastability has been obtained although the microscopic origin is not fully resolved. We have grown high-quality Cd-rich CdTe single crystals from the Cd solvent using the traveling heater method (THM) and reported an apparent As doping limit near 1017 cm−3.17 In addition, we reported evidence consistent with AX behavior such as persistent photoconductivity and increased doping efficiency after thermal quenching which decay over time.18 The understanding of group-V doping of CdTe is still rather rudimentary; the interplay of processing, group-V concentration, secondary phase formation, self-compensation, AX centers, minority carrier lifetime, and mobility still requires elucidation. In this letter, we focus on the effects of cooling processes on doping efficiency, mobility, and minority carrier lifetime. We demonstrate that rapid cooling from high temperature results in higher hole concentration, higher hole mobility, and higher photoelectron lifetime.

The As-doped Cd-rich CdTe single crystals were grown by Cd-solvent THM at a growth temperature of 950 °C as we have previously described.17 Grown crystals were cut into samples of approximately 5 mm × 5 mm × 0.5 mm with a diamond blade and polished mechanically with 0.1 μm Al2O3 powder and then etched with a 5% Br2/Methanol solution for 5 min to remove saw and polishing damage. The total incorporated As concentration in the crystals was determined using inductively coupled plasma atomic emission spectroscopy (ICP-AES). Ni-W Ohmic contacts of diameter 1 mm for the van der Pauw configuration Hall effect measurement were deposited by the evaporation method onto the corners of each sample to a thickness of 200–300 nm, and then, all samples were annealed at 250 °C under 96% N2–4% H2 mixed gas for 10 min to form the contacts. The hole concentration and mobility were measured at room temperature by resistivity and Hall effect measurements with a 0.54 T magnetic field in the dark. Prior to the Hall experiment, the samples were stored in the dark for 1 week after crystal growth in order to ensure relaxation. The bulk photocarrier lifetime was measured by two-photon excitation time-resolved photoluminescence (2PE-TRPL). The laser system provided 1120 nm excitation using 0.3 ps laser pulses with a 1.1 MHz repetition rate.19 

Figure 1 shows the hole concentration vs total As concentration for a series of Cd-rich samples quenched with an average cooling rate of 200–300 °C/h from the growth temperature of 950 °C to room temperature. A high doping activation (Hall-measured hole concentration per ICP-measured As concentration) of up to 45% can be observed for less than a few times the 1017 cm−3 atomic concentration of As, whereas for higher As concentrations, the activation efficiency decreases to less than ∼1%. Several studies have also shown the presence of a doping limit near 1017 cm−3 for Cd-rich group-V p-type doping.4,14,17 Several compensation mechanisms may contribute to the doping limit in As-doped CdTe with Cd-rich stoichiometry such as Cd interstitial donors (Cdi), Cd3As2 precipitation, and AX centers. Interstitial group-V element defect is expected to play self-compensation based on the diffusion process by the theoretical report.11 All results in our studies are consistent with an increasing fraction of self-compensation by AX centers with increasing As doping as would be expected from formation enthalpy arguments. We note that if AX metastability is the only cause of compensation, the maximum doping efficiency possible would be exactly 50%. Our highest observed doping efficiency of 45% approaches this limit and for practical reasons can only be measured after some time has elapsed after quenching. There are few reports on As or P doped samples exhibiting activation efficiency >50%.

FIG. 1.

Hole concentration versus measured As concentration in Cd-rich samples cooled from the growth temperature to room temperature at an average cooling rate of 200–300 °C/h. The dashed lines are isopleths of constant activation.

FIG. 1.

Hole concentration versus measured As concentration in Cd-rich samples cooled from the growth temperature to room temperature at an average cooling rate of 200–300 °C/h. The dashed lines are isopleths of constant activation.

Close modal

The predicted crossing of the formation enthalpies of the Td symmetric (AsTe acceptor) and distorted with lattice relaxation (AX donor) states as a function of Fermi energy implies that in thermal equilibrium, a sufficient density of AsTe centers will pin the Fermi level at the crossing point: the addition of two AsTe to the system will result in one adopting each configuration and thus self-compensation.9,13 This is because if the Fermi level lies above the crossing point, the formation energy of the Td configuration becomes lower than that of the AX center and vice versa. Calculation results for group-V doping in Cd-rich stoichiometry and Cu-doping under Te-rich stoichiometry indicate that the full CdTe defect equilibrium (including all native defects) tends to pin the Fermi level above the crossing point of the AX center.5,9,13

One potential strategy to overcome this pinning is non-equilibrium processing, which we implement in the form of quenching after crystal growth or annealing in order to manipulate the fraction of group-V dopants in the Td configuration. However, if the resulting Fermi level is below the crossing point, the AX center theory predicts that this configuration should be metastable and decay over time. Figure 2 shows dependence of doping efficiency (left red axis) and mobility (right blue axis) upon the cooling rate for 1016–1017 cm−3 of As doping. We note that increased cooling rates can result in doping efficiency as high as about 45%. In addition, Ablekim et al. reported that the doping efficiency in As-doped Cd-rich CdTe crystals was 0.02–0.04% with 7 °C/h and 1%–3% with 52 °C/h which is consistent with our results in Fig. 2.14 The fraction of self-compensation caused by AX instability (but possibly other mechanisms too) would be suppressed by quenching. Next, we investigate the effects of quenching on majority carrier mobility in order to attempt to provide differentiating evidence between possible origins of compensation reduction upon quenching.

FIG. 2.

Cooling rate dependence of doping efficiency (left red axis) and Hole mobility (right blue axis). Nominal amounts of As doping were 1016–1017 cm−3 for all samples. Square symbol data are reported in Ref. 14.

FIG. 2.

Cooling rate dependence of doping efficiency (left red axis) and Hole mobility (right blue axis). Nominal amounts of As doping were 1016–1017 cm−3 for all samples. Square symbol data are reported in Ref. 14.

Close modal

Figure 3 shows the hole mobility measured between 100 and 300 K for two samples grown under the same conditions with an As doping concentration of 1017 cm−3 but cooled at different rates after growth. We find that the mobility in both cases appears to be dominated by ionized defect scattering ∼Tk with k ≈ 3/2 (we observe k closer to 2 in these experiments). The fact that the mobility is higher after quenching compared to slow cooling is good evidence that the total ionized defect concentration is actually reduced upon quenching, i.e., it is consistent with the AX center hypothesis. Typically, concentrations of dominant point defects are higher at elevated temperatures and quenching would preserve them so this eliminates any mechanisms that produce more total point defects upon quenching. We have reported that the degree of compensation (the donor concentration per acceptor concentration) increases with decreased As doping efficiency in Cd-rich CdTe.18 All of these results are consistent with quenching, reducing the fraction of As acceptors in the distorted AX configuration and thus increasing doping efficiency and hole mobility. Using the mentioned strategies of limiting the nominal doping to less than a few times 1017 cm−3 and rapid cooling, a high quality As-doped Cd-rich CdTe single crystal with a high doping efficiency of ∼45% and a high hole mobility of 70 cm2/Vs at room temperature was obtained, which was stable for at least one month.

FIG. 3.

Temperature dependence of Hall mobility for two samples with an As concentration of 1017 cm−3 cooled at 50 or 350 °C/h and exhibiting 5% and 40% activation efficiencies, respectively.

FIG. 3.

Temperature dependence of Hall mobility for two samples with an As concentration of 1017 cm−3 cooled at 50 or 350 °C/h and exhibiting 5% and 40% activation efficiencies, respectively.

Close modal

The bulk lifetime is perhaps the most important parameter contributing to the photovoltaic performance of thin film photovoltaic absorber materials. Typical one photon excited TRPL probes rather shallow absorption depths in CdTe because of the large absorption coefficient, which (especially in the case of long minority carrier diffusion length) makes it strongly affected by surface/interface recombination. Two photon excited TRPL (2PE-TRPL) using sub-bandgap excitation can be focused >100 μm from the surface in bulk crystals, and thus, the bulk properties may be evaluated. 2PE-TRPL decay curves from 1016 and 1018 cm−3 As-doped Cd-rich and an unintentionally doped sample with no As added are shown in Fig. 4. All samples were grown by the cooling rate of 300 °C/h from the growth temperature of 950 °C to room temperature and relaxed in the dark for 1 week. Remarkably, the bulk lifetime of the 1016 cm−3 As-doped sample was measured to be >20 ns which is only slightly smaller than the lifetime of the unintentionally doped sample. This combination of high doping and lifetime again is consistent with the suppression of AX distorted As dopants and TeCd recombination centers by Cd-rich growth and rapid cooling, in agreement with theoretical predictions.5,8,9,13 We did find that the bulk lifetime can exhibit large spatial variation, illustrated by the second decay curve for the 1016 cm−3 As doped sample in Fig. 4 taken at a different location and yielding a lifetime of 5 ns. Our polishing procedure is not perfect, and deep scratches are sometimes present—we speculate that such scratches may contribute to the spatial lifetime variation in samples below the doping limit (which should presumably also be below the precipitation limit). Further investigation is ongoing.

FIG. 4.

Normalized 2PE-TRPL decay for bulk lifetime in three samples cooled from the growth temperature to room temperature at an average cooling rate of 300 °C/h and relaxed in the dark for 1 week. The sample with 1016 cm−3 of As doping was measured at two random locations: (a) and (b).

FIG. 4.

Normalized 2PE-TRPL decay for bulk lifetime in three samples cooled from the growth temperature to room temperature at an average cooling rate of 300 °C/h and relaxed in the dark for 1 week. The sample with 1016 cm−3 of As doping was measured at two random locations: (a) and (b).

Close modal

Further insight into another possible mechanism of spatially varying lifetime was provided by infrared (IR) transmission microscopy; however, this was possible in this study only for the 1018 cm−3 doped sample from Fig. 4 which is clearly above the doping limit. Indeed, as shown in Fig. 5, we observed a planar defect, likely a twin or stacking fault rather than a grain boundary as it did not meander (which we have observed often for grain boundaries), decorated by an ordered array of what are most probably Cd3As2 precipitates14 cutting diagonally through the thickness of the sample. Assuming based on the sample composition and the shape of the precipitates that these are indeed Cd3As2, this observation provides insight into the location of at least some of the As not contributing to the hole concentration for samples above the doping limit. Lastly, twins and stacking faults are the most prominent intragrain structural defects in thin polycrystalline films, suggesting that this dopant deactivation channel will probably be present in thin films at high doping also, especially in cases of slow cooling.

FIG. 5.

IR transmission image of a crystallographic twin passing diagonally through the thickness of the CdTe sample with 1018 cm−3 As doping shown in Fig. 4.

FIG. 5.

IR transmission image of a crystallographic twin passing diagonally through the thickness of the CdTe sample with 1018 cm−3 As doping shown in Fig. 4.

Close modal

In conclusion, we have grown As-doped Cd-rich CdTe single crystals and documented many of their properties. The main findings of this study are as follows: (1) We observe hole concentrations in the 1016–1017 cm−3 range and doping efficiency as high as 45% for As doping in quenched CdTe single crystals grown from the Cd solvent. (2) We observe long bulk lifetimes >20 ns with ∼45% of As doping efficiency in samples with 1016 cm−3. (3) We observe a doping limit in the low 1017 cm−3 range for quenching in the 200–300 °C/h range. (4) We document the effects of quenching on doping efficiency and mobility, and for samples below the doping limit of a few times 1017 cm−3, these results are consistent with AX metastability being a dominant cause of compensation. These properties were stable for at least one month after quenching. We consider that these doping properties in this study are reference for understanding As-doped polycrystalline thin film photovoltaics and the effects of cooling (quenching) provide improvement and controlling of group-V doping by the post diffusion process in the CdTe thin film sample.

A.N. acknowledges support from the JSPS for a Research Fellow Grant-in-Aid. M.A.S. acknowledges partial support from the Department of Energy through the Bay Area Photovoltaic Consortium under Award No. DE-EE0004946. At NREL, this research was supported by the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, under Solar Energy Technologies Office (SETO) Agreement Number 30306 and 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. We thank Professor Kelvin Lynn and Dr. Santosh Swain (Washington State University) for facilitating the IR transmission microscopy and 5 N Plus Semiconductors for suppling high purity Cd and Te materials.

1.
M. A.
Green
,
K.
Emery
,
Y.
Hishikawa
,
W.
Warta
,
E. D.
Dunlop
,
D. H.
Levi
, and
A. W. Y.
Ho-Baillie
,
Prog. Photovoltaics
25
,
3
(
2017
).
2.
M.
Gloeckler
,
I.
Sankin
, and
Z.
Zhao
,
IEEE J. Photovoltaics
3
,
1389
(
2013
).
3.
A.
Kanevce
,
M. O.
Reese
,
T. M.
Barnes
,
S. A.
Jensen
, and
W. K.
Metzger
,
J. Appl. Phys.
121
,
214506
(
2017
).
4.
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
).
5.
J. H.
Yang
,
W. J.
Yin
,
J. S.
Park
,
W. K.
Metzger
, and
S. H.
Wei
,
J. Appl. Phys.
119
,
045104
(
2016
).
6.
S. H.
Wei
and
S. B.
Zhang
,
Phys. Rev. B
66
,
155211
(
2002
).
7.
S. H.
Wei
and
S. B.
Zhang
,
Phys. Status Solidi B
229
,
305
(
2002
).
8.
J.
Ma
,
D.
Kuciauskas
,
D.
Albin
,
R.
Bhattacharya
,
M.
Reese
,
T.
Barnes
,
J. V.
Li
,
T.
Gessert
, and
S. H.
Wei
,
Phys. Rev. Lett.
111
,
067402
(
2013
).
9.
J. H.
Yang
,
W. J.
Yin
,
J. S.
Park
,
J.
Ma
, and
S. H.
Wei
,
Semicond. Sci. Technol.
31
,
083002
(
2016
).
10.
R.
Triboulet
and
P.
Siffert
,
CdTe and Related Compounds; Physics, Defects, Heteroand Nano-Structures, Crystal Growth, Surfaces and Applications Part I
(
Elsevier
,
2009
).
11.
E.
Colegrove
,
J. H.
Yang
,
S. P.
Harvey
,
M.
Young
,
J. M.
Burst
,
J. N.
Duenow
,
D. S.
Albin
,
S. H.
Wei
, and
W. K.
Metzger
,
J. Phys. D: Appl. Phys.
51
,
075102
(
2018
).
12.
K. J.
Bachmann
and
E.
Buehler
,
J. Electrochem. Soc.
121
,
835
(
1974
).
13.
J. H.
Yang
,
W. J.
Yin
,
J. S.
Park
,
J.
Burst
,
W. K.
Metzger
,
T.
Gessert
,
T.
Barnes
, and
S. H.
Wei
,
J. Appl. Phys.
118
,
025102
(
2015
).
14.
T.
Ablekim
,
S. K.
Swain
,
W.-J.
Yin
,
K.
Zaunbrecher
,
J.
Burst
,
T. M.
Barnes
,
D.
Kuciauskas
,
S. H.
Wei
, and
K. G.
Lynn
,
Sci. Rep.
7
,
4563
(
2017
).
15.
R. J.
Nelson
,
Appl. Phys. Lett.
31
,
351
(
1977
).
16.
D. V.
Lang
and
R. A.
Logan
,
Phys. Rev. B
19
,
1015
(
1979
).
17.
A.
Nagaoka
,
K. B.
Han
,
S.
Misra
,
T.
Wilenski
,
T. D.
Sparks
, and
M. A.
Scarpulla
,
J. Crystal Growth
467
,
6
(
2017
).
18.
A.
Nagaoka
,
D.
Kuciauskas
, and
M. A.
Scarpulla
,
Appl. Phys. Lett.
111
,
232103
(
2017
).
19.
D.
Kuciauskas
,
A.
Kanevce
,
J. M.
Burst
,
J. N.
Duenow
,
R.
Dhere
,
D. S.
Albin
,
D. H.
Levi
, and
R. K.
Ahrenkiel
,
IEEE J. Photovoltaics
3
,
1319
(
2013
).