Carrier density and lifetime for different dopants in single crystal and polycrystalline cdte

Carrier density and lifetime for different dopants in single crystal and polycrystalline CdTe Carrier density and lifetime for different dopants in single crystal and polycrystalline CdTe James M Burs[.]

Carrier density and lifetime for different dopants in single-crystal and polycrystalline CdTe

James M Burst, Stuart B Farrell, David S Albin, Eric Colegrove, Matthew O Reese, Joel N Duenow, Darius Kuciauskas, and Wyatt K Metzger

Citation: APL Mater 4, 116102 (2016); doi: 10.1063/1.4966209

View online: http://dx.doi.org/10.1063/1.4966209

View Table of Contents: http://aip.scitation.org/toc/apm/4/11

Published by the American Institute of Physics

Articles you may be interested in

Long carrier lifetimes in large-grain polycrystalline CdTe without CdCl2

APL Mater 108, 263903263903 (2016); 10.1063/1.4954904

Carrier density and lifetime for different dopants

in single-crystal and polycrystalline CdTe

James M Burst, Stuart B Farrell, David S Albin, Eric Colegrove,

Matthew O Reese, Joel N Duenow, Darius Kuciauskas,

and Wyatt K Metzger

National Renewable Energy Laboratory (NREL), Golden, Colorado 80401, USA

(Received 25 August 2016; accepted 13 October 2016; published online 1 November 2016)

CdTe defect chemistry is adjusted by annealing samples with excess Cd or Te vapor with and without extrinsic dopants We observe that Group I (Cu and Na) elements can increase hole density above 1016cm−3, but compromise lifetime and stability By post-deposition incorporation of a Group V dopant (P) in a Cd-rich ambient, lifetimes of 30 ns with 1016 cm−3 hole density are achieved in single-crystal and polysingle-crystalline CdTe without CdCl2 or Cu Furthermore, phosphorus doping appears to be thermally stable This combination of long lifetime, high carrier concentration, and improved stability can help overcome historic barriers for CdTe solar cell development C 2016 Author(s) All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).[http://dx.doi.org/10.1063/1.4966209]

By increasing cell efficiency toward 25%, CdTe solar technology can further outperform mul-ticrystalline silicon with costs below traditional electricity sources.1,2Currently, CdTe photovoltaics relies on CdCl2 and Cu treatments to achieve high efficiency; however, this processing leads to

a stubborn compensated defect chemistry manifested by the open-circuit voltage (Voc) of record-efficiency cells being limited to 840–890 mV for over two decades.3 , 4 The CdTe layer has been grown by electrodeposition, screen printing, physical vapor deposition, close-spaced sublimation (CSS), sputtering, vapor transport, and other methods, with grain size typically ranging from several hundred nanometers to several micrometers.5 , 6 Interestingly, these different deposition methods produce similar material properties Before processing, carrier concentration is typically less than

1014cm−3and lifetimes are on the order of tens of ps.7 9Consequently, a CdCl2treatment is applied

to increase aggregate carrier lifetimes to several ns, and Cu is applied to increase hole density to

1013–1015cm−3 However, Cu can move spatially and energetically, which can create stability issues and change carrier lifetime.5 , 10 , 11

Additional efficiency advances will require increasing Vocand fill factor by further improving the CdTe material properties Modeling indicates that 25% efficiency can be reached by increasing the CdTe hole density beyond 1016cm−3with lifetimes above 10 ns Dopants can be incorporated both during (in situ) and after (post- or ex situ) deposition Recent experimental results demonstrate that

in situdoped CdTe single crystals can reach radiatively limited lifetimes and support Voc> 1 V.12

In situdoping is ideal for uniformly introducing dopants during growth and will be pursued in other studies But a disadvantage is that film deposition and dopant incorporation must be done simulta-neously in a manner specialized to the growth technique

Here, post-deposition annealing methods are applied to increase hole density and examine doping stability and carrier lifetime in different CdTe materials Post-deposition doping is currently used to introduce Cl and Cu, and it can be applied to films deposited by different methods Disadvantages are that often a diffusion step is required and different bulk grain-boundary and bulk diffusion rates can introduce inhomogeneity By progressing from pure single-crystal (sX) CdTe, to multicrystalline (mX) samples with 300 µm−3 mm grains, to polycrystalline (pX) films with grains tens of microns in diameter produced by fast manufacturing methods, we examine how structural defects and impurities inherent in thin-film processing alter lifetime and hole density Doping is adjusted by annealing these

2166-532X/2016/4(11)/116102/6 4, 116102-1 © Author(s) 2016.

116102-2 Burst et al. APL Mater 4, 116102 (2016)

sX, mX, and pX samples with excess Cd or Te vapor without extrinsic dopants, or with Cu, Na, or P included Several approaches achieve a hole density >1016cm−3, but lifetime varies significantly The next three paragraphs describe methods The sX and mX CdTe samples were made by the traveling-heater method at JX Nippon Mining and Metals and cut to 5 mm × 5 mm × 800 µm.13

Most samples were nominally undoped (see thesupplementary materialfor purity details); however, Nippon also provided samples with Na incorporated to 1017cm−3during crystal growth CSS was performed at NREL to deposit nominally undoped pX CdTe films with 50–100-µm thickness at moderate vacuum conditions (20 Torr) using 5N-pure CdTe powders.14 The pX substrates were 800-nm molybdenum sputtered on Corning 7059 borosilicate glass.5 , 15

The equilibrium phase diagram shows that Cd and Te vacancy solubility in single-phase CdTe steadily increases with temperature.16By annealing samples in Te or Cd overpressures, it is possible to increase Cd and Te vacancies and form substitutional acceptors17with Group I and Group V elements

on Cd and Te sites, respectively The CdTe samples were placed in the center of ampoules evacuated

to ∼10−4–10−5Torr Cd, Te, or P+ Cd shot was placed separately at a necked end of ampoules about

10 cm3in volume The samples are described according to the dopant and ambient (e.g., Cu+ Te) Cu + Te samples were formed by diffusing 10 nm of evaporated Cu at 250◦C for 1 h and then annealing

in Te Na+ Te samples were made by annealing samples with Na incorporated during sX growth in

Te Te, Cd, and P+ Cd samples were made by annealing nominally undoped samples in overpressures

of Te, Cd, and P+ Cd vapor, respectively The overpressure anneals were isothermal and performed for 2–4 h at 600◦C–750◦C in a calibrated, three-zone Mellen tube furnace; the ampoules were removed and quenched in water to condense vapor away from the samples

Hall-effect measurements were taken with graphite contacts in the van der Pauw configuration

to provide information on carrier mobility, type, and concentration (BioRad HL5500) Impurity concentrations were determined from secondary-ion mass spectrometry (SIMS) measurements For capacitance-voltage (CV) analysis on CSS samples, we fabricated diodes by sputter-depositing CdS/ZnO/ZnO:Al or In/ITO for the front contact and Cu/Mo or graphite for the back contact; values

at 0 V bias are presented in Fig.3 A regeneratively amplified Yb:KGW laser and optical parametric amplifier tuned to 1120 nm provided two-photon excitation (2PE) with laser pulses 300 fs in tempo-ral width fired at a rate of 1.1 MHz A silicon avalanche photodiode counted single photons passed through interference filters with 10-nm bandwidth at 840-nm center wavelength Bulk lifetime was determined by exponential fits to the time-resolved photoluminescence (TRPL) decay curves.18 Results plotted in Figure1show the variation of carrier concentration and lifetime on sX and

mX crystals The carrier density and bulk lifetime for the as-grown (center of Figs.1(a)and1(b)) p-type crystals are about 2 × 1014cm−3and 80 ns, respectively By annealing samples in Cd (brown symbols), we convert the CdTe from p- to n-type This result clearly indicates that we can reset the original Te-rich defect chemistry by post-deposition processes Upon annealing samples in Te, we achieve a hole density exceeding 1016cm−3with lifetimes of 5–10 ns

Hole density above 1016 cm−3 is also achieved with Cu after annealing the samples in Te; however, the lifetime decreases by three orders of magnitude Na+ Te processing produces hole density from 1016 to just over 1017 cm−3and lifetimes of 1–2 ns (not shown) After the P + Cd anneal, the hole density increases to 1016cm−3with lifetimes as high as 40 ns, which is significantly better than current technology and approaches the results necessary to approach Voc> 1 V.12

The mX and sX data shown by open and closed squares, respectively, in Figs 1(a)and1(b)

indicate that methods that increase the hole density in single crystals produce similar results in mX CdTe Therefore, grains at this scale do not affect aggregate doping Hall-effect mobility values (not shown) are all in the range of 60–80 cm2/V s Thus, large grains and/or hole-density variation from low-1014to mid-1016cm−3do not limit CdTe mobility here

The PL data shown in Fig.1(c), viewed in light of the hole-density measurements and literature, indicate that the overpressure anneals form vacancies that can be occupied by extrinsic dopants such as

P or Cu, if present Previous calculations and experiments indicate that vacancies have lower formation energy than interstitials or antisites, and they can remain after samples are cooled, so vacancies are more likely to influence hole density.19 , 20After the Te anneal, a 1.54–1.55-eV peak emerges in the low-temperature PL data that have been identified as a donor-acceptor pair (DAP) transition involving

an acceptor Cd-vacancy complex.21 – 24

FIG 1 Comparison of the (a) carrier concentration and (b) bulk minority-carrier lifetime for sX (closed circles) and mX (open squares) samples for each processing condition Black and brown symbols represent p- and n-type doping, respectively (c) PL emission spectra at 4.25 K Dashed lines represent Gaussian fits indicating longitudinal optical (LO) phonon replicas (d) Representative decays of single-photon counting time-resolved photoluminescence.

When samples are annealed in Te with Cu present, Cu can occupy Cd sites to form CuCd acceptors Effective-mass theory calculations indicate that the CuCdlevel is 146–147 meV, and they accurately describe transition energies measured by infrared absorption between many CuCdstates such as 1S3/2 to 2P5/2(Γ7).17 , 25 – 27 First-principles theoretical estimates have ranged from 160 to

300 meV.28 – 30Electrical device measurements observe many trap levels, and thus, CuCdassignments range from 150 to 400 meV.31 – 34However, for single crystals, correlated Hall, admittance spectros-copy, and PL measurements corroborate the CuCdlevel at 135–157 meV.32Similarly, for pX and sX CdTe with ion-implanted or diffused Cu, resonantly excited luminescence, infrared absorption, and

PL assign the CuCdlevel to 145–150 meV with a zero-phonon DAP transition at 1.45 eV.26,27,32,35–40 The emergence of the zero-phonon DAP peak at 1.453 eV and replicas for Cu+ Te in Fig.1(c)is consistent with the formation of 145–150-meV CuCdacceptors

When samples are annealed in Cd with P present, P can occupy Te sites to form PTe This is observed in Fig.1(c)for the P+ Cd samples by the zero-phonon DAP peak at 1.531 eV and DAP

LO phonon replicas at lower energies Hall and theoretical studies indicate that PTesubstitution forms

a shallow singly ionized acceptor state at 35–70 meV above the valence band, and PL spectroscopy indicates 60–80 meV.17 , 25 , 28 , 41 – 46In agreement, the measured acceptor level from the PL spectra in Fig.1(c), after subtracting the zero-phonon energy from the bandgap and estimating 10–15 meV donor energy,17 , 47is about 60–70 meV Effective-mass theory for PTepredicts 68 meV.17 , 25 , 43The increased hole density, anneal environment, and 1.543-eV zero-phonon peak indicate the formation of shallow

PTeacceptors

The lifetime trends in Fig.1are consistent with theoretical and experimental studies where Cd-rich stoichiometry resulted in longer lifetimes due to fewer TeCdrecombination centers.19,48Incorporation and activation levels are given in Fig S1 (supplementary material); incorporation ranges from 1016

to 1018cm−3with less than 10% Cu activation and P activation approaching 50% The latter matches the best activation levels in recent samples incorporated with As during molecular beam epitaxy or P during Bridgman growth, where the lifetime increased with activation.12 , 49Cu has been observed to increase lifetime at low concentration and can replace defects such as TeCdor VCd, as observed in PL spectra by Grecu et al.10 , 21At larger Cu concentrations, lifetime has been observed to decrease.10 , 50 , 51

The severe lifetime reduction by Cu shown in Fig.1(d)cannot be attributed to the CuCdor Cuisites based on recent experimental and theoretical capture cross section estimates.37 , 52So a different Cu defect is the most plausible explanation, and improved activation could increase lifetime

116102-4 Burst et al. APL Mater 4, 116102 (2016)

FIG 2 Variation of carrier concentration in sX CdTe with (a) time and (b) temperature The solid lines are guides to the eye.

In addition to increasing lifetime and hole density, it is necessary to establish a stable defect chemistry Figure 2(a) indicates doping stability probed by measuring hole density for samples stored in the dark over several months The Cu+ Te samples degrade in the dark, which is consis-tent with Cu incorporation above room-temperature saturation limits.53 , 54The hole density obtained

in the Te or P+ Cd samples did not change in the dark over the same time period

Samples were also subjected to 30-min anneals at fixed temperatures from 100◦C to 300◦C in

a tube furnace with flowing He to probe stability at temperatures frequently encountered during de-vice fabrication processes Figure2(b)indicates that for every dopant other than P, the hole density decreases by about two orders of magnitude at 250◦C or higher Stability issues have been linked with Na and Cu in solar cells.5 , 10 , 11 , 55Consistent with Refs.55and56, these sX results indicate that Group I dopants induce stability problems because of their bulk defect chemistry, rather than just grain-boundary effects Combined, the data indicate that attempting to increase hole density above

1016cm−3with Cu can present significant stability and lifetime obstacles Group V dopants diffuse less readily,54,57and here, the Group V dopant, P, appears to be the most stable

Next, we discuss results when transferring the methods that worked on sX and mX materials

to pX CdTe grown by CSS To avoid compensation and assess bulk recombination with reduced influence from grain boundaries and interfaces, thick and consequently very large-grained pX CdTe was deposited by CSS without intentional Cl, Cu, S, or O Figure 3(a) illustrates that in the as-grown pX CdTe, the average bulk minority-carrier lifetime is 10 ns, which is remarkably high, and it approaches values achieved in thinner devices after CdCl2and Cu processing.4Here, large high-quality grains enable long lifetimes in rapidly deposited material without the CdCl2 treatment.14This suggests that bulk CdTe lifetime is not limited a priori by fast manufacturing rates

or intrinsic impurities in the raw material Furthermore, CdCl2may be most critical to passivate grain-boundary and interface recombination in standard small-grain thin films

Figure3illustrates that for the Cu+ Te samples, the lifetime decreased by orders of magnitude just as for single crystals; yet, the hole density did not exceed 1015cm−3, even though incorporation

FIG 3 (a) Hole density and lifetime for sX and CSS pX samples The P + Cd annealing results show a clear increase

in lifetime and hole density compared to as-grown films (b) Time-correlated single-photon counting decay curves for representative samples from (a).

levels were similarly around 1018 cm−3 Te vapor pressure at temperatures >600◦C was insu ffi-cient to prevent thin CdTe films from evaporating partially or entirely, so data are not reported for this condition On the other hand, the P + Cd anneal increases lifetime to about 20 ns and hole density to 1016 cm−3 in CSS pX films, similar to the sX films Achieving these properties simultaneously without Cu or Cl in a pX material indicates a potential and fundamentally different path for improving CdTe efficiency

In summary, we demonstrate methods that alter the defect chemistry of as-grown CdTe material and increase carrier concentration and bulk lifetime Group I dopants Cu and Na, annealed with Te excess, increase hole density above 1016cm−3but decrease lifetime to less than 2 ns These Group

I dopants are not stable in the dark or under moderate anneal temperatures Annealing samples in

Te achieves hole density and lifetime values above 1016cm−3and 5 ns, respectively; however, it has been difficult to transfer these results to pX thin films By incorporating the Group V element P on

Te sites, bulk lifetime values of 20–40 ns with a hole density of 1016cm−3are obtained in both CdTe crystals and CSS pX films, and the hole density is more stable This represents an improvement of two orders of magnitude in hole density relative to standard pX CdTe solar cells while still retaining good lifetime The post-deposition approaches can be applied to material produced by different methods These approaches provide a path to overcome longstanding CdTe material limits and to improve stability for CdTe solar applications

Seesupplementary materialfor discussion of impurity levels in nominally undoped single crys-tals and comparison of the SIMS atomic concentration and hole density for Cu, Na, and P sX and

pX samples

This work was supported by the U.S Department of Energy under Contract No DE-AC36-08-GO28308 with the National Renewable Energy Laboratory

1 See www.firstsolar.com /en/solutions/utility-scale-generation for CdTe module information and comparison with conven-tional energy sources.

2 See http://www.greentechmedia.com/articles/read/First-Solar-is-Reaching-16.3-Efficiency-in-Production-PV-Modules for

a discussion of CdTe and Si module performance.

3 M A Green, K Emery, Y Hishikawa, W Warta, and E D Dunlop, “Solar cell efficiency tables (version 48),” Prog Photo-voltaics 24, 905 (2016).

4 M Gloeckler, I Sankin, and Z Zhao, J Photovoltaics 1, 1389 (2013).

5 B E McCandless and J R Sites, “CdTe solar Cells,” in Handbook of Photovoltaic Science Engineering, edited by A Luque and S Hegedus (John Wiley & Sons West Sussex, 2007).

6 H R Moutinho, R G Dhere, M M Al-Jassim, D H Levi, and L L Kazmerski, J Vac Sci Technol A 17, 1793 (1999).

7 W K Metzger, D Albin, D Levi, P Sheldon, X Li, B M Keyes, and R K Ahrenkiel, J Appl Phys 94, 3549 (2003).

8 W K Metzger, D Albin, M J Romero, P Dippo, and M Young, J Appl Phys 99, 103703 (2006).

9 L Kranz, C Gretener, J Perrenoud, R Schmitt, F Pianezzi, F La Mattina, P Blösch, E Cheah, A Chiril˘a, C M Fella, H Hagendorfer, T Jäger, S Nishiwaki, A R Uhl, S Buecheler, and A N Tiwari, Nat Commun 4, 2306 (2013).

10 T A Gessert, W K Metzger, P Dippo, S E Asher, R G Dhere, and M R Young, Thin Solid Films 517, 2370 (2009).

11 C R Corwine, A O Pudov, M Gloeckler, S H Demtsu, and J R Sites, Sol Energy Mater Sol Cells 82, 481–489 (2004).

12 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).

13 A Noda, H Kurita, and R Hirano, “Bulk growth of CdZnTe /CdTe crystals,” in Mercury Cadmium Telluride: Growth, Properties, and Applications , edited by P Caper and J Garland (John Wiley & Sons, UK, 2011).

14 S A Jensen, J M Burst, J N Duenow, H L Guthrey, J Moseley, H R Moutinho, S W Johnston, A Kanevce, M M Al-Jassim, and W K Metzger, Appl Phys Lett 108, 263903 (2016).

15 D H Rose, F S Hassoon, R G Dhere, D S Albin, R M Ribelin, X S Li, Y Mahathongdy, T A Gessert, and P Sheldon,

Prog Photovoltaics 7, 331 (1999).

16 J H Greenberg, J Cryst Growth 197(3), 397 (1999).

17 J L Pautrat, J M Francou, N Magnea, E Molva, and K Saminadayar, J Cryst Growth 72, 194 (1985).

18 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).

19 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).

20 M Wienecke, H Berger, and M Schenk, Mater Sci Eng B16, 219 (1993).

21 D Grecu and A D Compaan, Appl Phys Lett 75, 361 (1999).

22 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).

23 J M Figueroa, F Sánchez-Sinencio, J G Mendoza-Alvarez, O Zelaya, C Vázquez-López, and J S Helman, J Appl Phys 60, 452 (1986).

24 D P Halliday, M D G Potter, J T Mullins, and A W Brinkman, J Cryst Growth 220, 30 (2000).

25 M Said and M A Kanehisa, J Cryst Growth 101, 488–492 (1990).

116102-6 Burst et al. APL Mater 4, 116102 (2016)

26 W Stadler, D M Ho ffman, H C Alt, T Muschik, B K Meyer, E Weigel, G Müller-Vogt, M Salk, E Rupp, and K W Benz, Phys Rev B 51, 10619 (1995).

27 E Molva, J P Chamonal, G Milchberg, K Saminadayar, B Pajot, and G Neu, Solid State Commun 44, 351 (1982).

28 S.-H Wei and S B Zhang, Phys Rev B 66, 155211 (2002).

29 J H Yang, W J Yin, J S Park, W Metzger, and S H Wei, J Appl Phys 119, 045104 (2016).

30 D Krasikov, A Knizhnik, B Potapkin, S Selezneva, and T Sommerer, Thin Solid Films 535, 322–325 (2013).

31 A Castaldini, A Cavallini, B Fraboni, P Fernandez, and J Piqueras, J Appl Phys 83, 2121 (1998).

32 J P Laurenti, G Bastide, M Rouzeyre, and R Triboulet, Solid State Commun 67, 1127 (1988).

33 B Biglari, M Samimi, M Hage-Ali, J M Koebel, and P Si ffert, J Cryst Growth 89, 428–434 (1988).

34 A Balcioglu, R K Ahrenkiel, and F Hasoon, J Appl Phys 88, 7175 (2000).

35 K M James, J L Merz, and C E Jones, J Vac Sci Technol A 6, 2664 (1988).

36 J P Chamonal, E Molva, and J L Pautrat, Solid State Commun 43(11), 801 (1982).

37 E Molva, J P Chamonal, and J L Pautrat, Phys Status Solidi B 109, 635 (1982).

38 D Kuciauskas, P Dippo, A Kanevce, Z B Zhao, L Cheng, A Los, M Gloeckler, and W K Metzger, Appl Phys Lett.

107, 43906 (2015).

39 T A Kuhn, W Ossau, A Waag, R N Bicknell-Tassius, and G Landwehr, J Cryst Growth 117, 660 (1992).

40 D Grecu, A D Compaan, D Young, U Jayamaha, and D H Rose, J Appl Phys 88, 2490 (2000).

41 F A Selim and F A Kroger, J Electrochem Soc 124, 401 (1977).

42 J Gu, T Kitahara, K Kawakami, and T Sakaguchi, J Appl Phys 46, 1184 (1975).

43 A Baldereschi and N O Lipari, Phys Rev B 9, 1525 (1974).

44 C Kraft, A Brömel, S Schönherr, M Hädrich, U Reislöhner, P Schley, G Gobsch, R Goldhahn, W Wesch, and H Metzner,

Thin Solid Films 519, 7153 (2011).

45 J Saraie, H Shinohara, H Edamatsu, and T Tanaka, J Lumin 21, 337–351 (1980).

46 E Molva, J I Pautrat, K Saminadayar, G Milchberg, and N Magnea, Phys Rev B 30, 3344 (1984).

47 R N Bhargava, J Cryst Growth 59, 15 (1982).

48 M O Reese, C L Perkins, J M Burst, S Farrell, T M Barnes, S W Johnston, D Kuciauskas, T A Gessert, and W K Metzger, J Appl Phys 118, 155305 (2015).

49 S Farrell, T Barnes, W K Metzger, J H Park, R Kodama, and S Sivananthan, J Electron Mater 44, 3202 (2015).

50 S H Demtsu, D S Albin, J R Sites, W K Metzger, and A Duda, Thin Solid Films 516, 2251 (2008).

51 X Wu, J Zhou, A Duda, Y Yan, G Teeter, S Asher, W K Metzger, S Demtsu, S H Wei, and R Noufi, Thin Solid Films

515, 5798 (2007).

52 J H Yang, W K Metzger, and S H Wei, “Carrier providers or carrier killers: The Case of Cu Defects in CdTe Solar Cells” (unpublished).

53 J Perrenoud, L Kranz, C Gretener, F Pianezzi, S Nishiwaki, S Buecheler, and A N Tiwari, J Appl Phys 114, 174505 (2013).

54 G Teeter and S Asher, in Proceedings of the 33th IEEE Photovoltaic Specialist Conference, San Diego, CA, 2008.

55 J N Duenow, J M Burst, D S Albin, D Kuciauskas, S W Johnston, R C Reedy, and W K Metzger, Appl Phys Lett.

105, 53903 (2014).

56 C Gretener, J Perrenoud, L Kranz, E Cheah, M Dietrich, S Buecheler, and A N Tiwari, Sol Energy Mater Sol Cells

146, 51 (2016).

57 E Colegrove, S Harvey, J H Yang, J Burst, D Albin, S H Wei, and W K Metzger, Phys Rev Appl 5, 054014 (2016).