The Role of Sodium as a Surfactant and Suppressor of Non-Radiative Recombination at Internal Surfaces in Cu 2 ZnSnS 4

FULL P

APER

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim (1 of 8) 1400849wileyonlinelibrary.com

The Role of Sodium as a Surfactant and Suppressor of Non-Radiative Recombination at Internal Surfaces in Cu 2 ZnSnS 4

Talia Gershon , Byungha Shin , Nestor Bojarczuk , Marinus Hopstaken , David B. Mitzi , and Supratik Guha *

Dr. T. Gershon, N. Bojarczuk, Dr. M. Hopstaken, Dr. D. B. Mitzi, Dr. S. Guha IBM TJ Watson Research Center Physical Sciences Department 1101 Kitchawan Rd , Yorktown Heights , NY 10598 , USA E-mail: [email protected] Prof. B. Shin Dept. of Materials Science and Engineering Korea Advanced Institute of Science and Technology Daejeon , Republic of Korea

DOI: 10.1002/aenm.201400849

glass have larger grain size and give better performance due to in-diffusion of alkali species from the glass.

There have been a number of studies aimed at understanding the role of sodium in CIGS and CZTS, and the mechanisms through which it infl uences electronic and microstructural properties. Some studies highlight the importance of sodium in enhancing grain size in both CIGS and CZTS. [ 7,8 ] Other studies have shown that sodium increases the carrier concentration and/or mobility of CIGS and CZTS, [ 5,9–11,29 ] and that it can passivate deep defects. [ 3,12 ] Device improvements have been demon-strated even when sodium is introduced to CIGS via a low-temperature process after fi lm growth (i.e., such that samples are not allowed to undergo microstruc-tural changes). [ 6 ] Sodium has been shown

to mainly enhance the open circuit voltage ( V oc ) and fi ll fac-tors of devices, both of which are heavily sensitive to defects and traps. [ 3,4,9,11,13 ] Kelvin probe studies comparing CIGS and CZTS grain boundaries have indicated potential “spikes” at grain boundaries, which have been proposed to behave as hole-repelling and electron-attracting regions, aiding the separation and transport of the minority electrons. [ 14,15 ] However, these spikes may not exist in materials that do not contain sodium, [ 14 ] and some studies have found that the sign of potential spike at the grain boundary can even reverse, depending on sample preparation. [ 16 ]

An open question remains whether unpassivated grain boundaries in CZTS contain non-radiative trap states, and whether sodium effectively passivates these traps. First-princi-ples calculations have shown that the grain boundaries in CIGS are electrically benign due to large atomic relaxations that result in the absence of deep states in the band gap. [ 14 ] By contrast, fi rst principles calculations of the grain boundaries of CZTS suggest that deep states do exist in the gap, which are expected to act as non-radiative recombination centers. [ 17 ]

This study demonstrates that non-radiative traps unequivo-cally exist in CZTS fi lms without sodium, and points to sur-faces and grain boundaries as their most likely location. As shown both in this paper, and in prior work, [ 12,18,19 ] Na

It is well-known that sodium improves the performance of Cu 2 ZnSnS 4 (CZTS) devices, yet the mechanism of the enhancement is still not fully understood. This work aims to present a unifi ed account of the relationships between grain boundaries in CZTS, sodium content at these boundaries, non-radiative recombination, and surfactant effects that produce large microstructural changes. Using temperature-dependent photoluminescence measurements, it is demonstrated that samples containing dramatically different grain sizes display identical radiative and non-radiative decay characteristics when suffi cient sodium is present in the fi lm. It is also shown that the sodium concentration needed to effi ciently passivate non-radiative defects is signifi -cantly less that the quantity needed to obtain micrometer-sized CZTS grains. Finally, the high densities of donor-acceptor pairs that are observed in CZTS fi lms appear to reside within the grains themselves, rather than at grain boundaries.

1. Introduction

The earth-abundant semiconductor Cu 2 ZnSnS 4 (CZTS) is an attractive candidate absorber material for large-scale photovol-taics deployment. [ 1 ] Much of what is known to be benefi cial for CZTS solar cell performance has been adopted from research on the related material Cu(In,Ga)Se 2 (CIGS). [ 1,2 ] In particular, the benefi cial role of sodium in enhancing device performance and the material properties of both CIGS and CZTS has been widely reported in the literature. The benefi t of sodium is evi-dent in both CIGS and CZTS devices from direct comparisons on different substrates, [ 3–6 ] where materials grown on soda-lime

Adv. Energy Mater. 2014, 1400849

www.MaterialsViews.comwww.advenergymat.de

http://doi.wiley.com/10.1002/aenm.201400849

FULL

PAPER

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim1400849 (2 of 8) wileyonlinelibrary.com

diffuses to internal and external surfaces in the fi lm. By cor-relating the sodium content of CZTS fi lms with temperature-dependent photoluminescence data, we fi nd that sodium renders the grain boundaries passive (within our detection limits). This is supported by the nearly identical radiative and non-radiative recombination profi les of samples with ≈325 nm grains and 1 µm grains, i.e., in samples with signifi cantly dif-ferent grain boundary densities. We also demonstrate that the quantity of sodium required for passivating this type of non-radiative recombination is different from the quantity required to observe micron-sized grains in CZTS thin fi lms. Thus, the two phenomena of grain boundary passivation and surfactant activity, though both associated with Na, have separate causes and concentration requirements.

The presence of sodium in large concentrations is corre-lated with a large increase in surface oxygen content, even for minimal exposure to ambient; we note, however, that a direct causal relationship between the presence of oxygen and defect passivation has not been determined although it has been hypothesized. [ 12 ] Finally, the fact that samples with small grains (≈325 nm diameter) have similar photoluminescence spectra to samples with much larger grains (≈1 µm diameter) at low tem-peratures indicates that the high concentration of point defects observed in CZTS materials [ 20 ] lie predominantly in the bulk and not in the grain boundaries.

2. Experimental Section

Fine-ground yttria-stabilized zirconia (YTZP) substrates (Insaco, Inc.) were selected for this study to ensure that the sodium con-tent in the CZTS could be controlled strictly through intentional addition of NaF, instead of diffusion from the substrate. YTZP substrates also eliminate other potential contaminants from the glass substrate (e.g., potassium, calcium), thereby isolating the effect of the sodium. The YTZP substrates were coated with 700 nm of sputtered Mo. Different thicknesses of NaF were coated onto the YTZP/Mo substrates using a thermal evapo-rator: 1, 2.5, 4.5, 8, 15 and 23 nm, with one uncoated sample kept as a control (i.e., no NaF). The addition of NaF to CZTS is a well-known strategy for Na incorporation. [ 6,21 ] Throughout the manuscript, samples are referenced by the initial thick-ness of NaF, although we note that some sodium is expected to volatilize out of the high-sodium samples during annealing. We have verifi ed through secondary ion mass spectrometry (SIMS), however, that the sodium content in the fi lms increases in accordance with the above sequence.

CZTS was co-evaporated from resistively heated cells con-taining elemental Cu, Zn, Sn, and a valved-cracker sulfur source onto the substrates held at 150 °C (base pressure ≈ 1 × 10 −8 Torr). The growth rate was approximately 10 nm per minute. The source materials all had 6N purity except for Sn, which had 5N purity. After low-temperature growth, the sam-ples were subjected to a high-temperature anneal (>580 °C) in the presence of excess sulfur and in a nitrogen-fi lled glove box. [ 22 ] The fi nal compositions of the samples were nominally Cu/Sn = 1.80, Zn/Sn = 1.20 for the samples on YTZP. These samples were deposited together in the same thin fi lm growth, with sample rotation enabled and are therefore expected to be

of identical composition. The composition of a reference sam-ples grown on glass (no NaF added) was Cu/Sn = 1.8, Zn/Sn = 1.25. These measurements were performed using X-ray fl uores-cence, calibrated with a sample measured separately by Ruther-ford backscattering. We note that the preparation conditions used for this study are the same conditions used to prepare high-effi ciency co-evaporated CZTS devices. [ 22 ]

One by one, samples were removed from the glove box and immediately loaded into an X-ray photoelectron spectroscopy (XPS) measurement tool with minimal exposure to the ambient atmosphere (≈5–10 min). Monochromatic Mg Kα radiation was used for the XPS measurements. The XPS spectra presented in this paper were collected with a take-off angle of 90°. Binding energies were shifted using the common C 1s peak at 284.6 eV.

After XPS characterization, the samples were examined by scanning electron microscopy (SEM) using a 10 kV accelera-tion voltage. Temperature-dependent photoluminescence (PL) spectra were recorded using a Hamamatsu single-photon counting system with a pulsed 532-nm laser (15 kHz) and a pulse-duration of 1 ns (4.5 × 10 −5 J cm –2 ). Our detector is an InGaAs PMT, cooled to −80 °C to reduce noise. The sample temperature was controlled using a Janis Super-Tran VP contin-uous He fl ow cryostat, equipped with a PID-controlled resistive heater with thermocouples.

Compositional depth profi les for both bulk and trace ele-ments were acquired by SIMS depth profi ling, using a primary 3 keV Cs + ion beam. Secondary ions were collected as positive CsM + cluster ions and normalized to Cs + reference ion to mini-mize matrix effects. The signature of the Mo back electrode is used to provide depth perspective through the fi lms. During the fi rst ≈600 s, mass interference with other elements results in the measurement of a background Mo signal. From 600–800 s, the Mo signal increases and the S signal remains high, signi-fying the beginning of the interfacial MoS 2 region of the thin fi lm stack. At this time, the metal signals all decrease. After an additional 300 s, the sulfur signal decreases and the Mo signal increases again, implying that the Mo metal has been reached (see Figure 1 ).

3. Results

Figure 1 shows SIMS depth profi les of the Cu, Zn, Sn, and S constituent elements (Figure 1 a,b), as well as the sodium content (Figure 1 c,d) of fi ve CZTS samples. Four of the samples were grown on YTZP substrates with varying thicknesses of NaF, and one sample was grown on glass as a reference. NaF was not intentionally added to the sample grown on the glass substrate, thus all sodium in this sample originated from the soda-lime glass. Figure 1 a shows that at fi rst glance the Cu, Zn, Sn, and S profi les appear similar across the fi ve samples under consid-eration. However, upon closer inspection (Figure 1 b) it can be seen that a zinc-rich region is increasingly present at the surface (<200 s) in the samples with 0, 1, and 4.5 nm of NaF, respectively, while the sample with 23-nm NaF contains a zinc-rich region near the Mo back contact (450–700 s). Since all of these sam-ples were nominally “zinc-rich” and “copper-poor,” in accordance with optimal device results, it is not surprising to fi nd regions of segregated ZnS. It appears that the quantity of alkali species in



FULL P

APER


the fi lm infl uences the location of the precipitated ZnS. Notably, the Zn signal in the middle of the CZTS fi lm region (200–500 s) is higher in the 0 and 1 nm NaF samples, implying that ZnS precipitates could be trapped within the CZTS fi lm region when the sodium concentration is low. A closer inspection of the zinc and sodium distributions throughout these fi lms can be found in the Supporting Information.

The SIMS data confi rms that the sodium distribution throughout the sample (surface, fi lm region, and MoS 2 layer) correlates with the quantity of sodium introduced via NaF (Figure 1 c,d, Table 1 ). Approximately 9–33% (depending on the sample) of the total sodium measured by SIMS is found at the sample surface after the anneal (Table 1 ). This result is consistent with earlier reports that sodium segregates to CZTS surfaces and grain boundaries. [ 12,18,19 ]

XPS measurements of the samples containing varying quan-tities of sodium are shown in Figure 2 . The data are again consistent with previous observations of sodium segregation to CZTS surfaces and grain boundaries. [ 12,18,19 ] A large Na 1s peak is visible for the sample with 23 nm of NaF, and is accompanied by a large O 1s peak despite efforts taken to minimize the expo-sure of these samples to ambient atmosphere before the meas-urement. The correlation between sodium and oxygen at the CIGS and CZTS surface has been observed previously. [ 12,18,19 ] Similarly, faint but visible sodium and oxygen peaks can be seen in the sample with 4.5 nm of NaF. These results suggest that the sodium attracts oxygen to the CZTS surface within minutes, while oxygen is not immediately observed at surfaces without sodium. Additionally, there is a strong signal from the zinc 3p core levels in the samples with 4.5 nm or less of NaF.



Figure 1. SIMS depth profi les of CZTS samples: no NaF, 1 nm NaF, 4.5 nm NaF, and 23 nm NaF on YTZP substrates, and one sample on glass (no NaF intentionally added). a) Cu, Zn, Sn, and S of all samples overlaid with Mo signal shown for depth reference. b) Enlarged view of the Zn signals from the YTZP samples. c) Na signals of the fi ve samples, with Mo overlaid for depth reference. d) Comparison of sodium distribution in different regions of the samples based on integrated raw SIMS data.

Table 1. Calculations of sodium distribution and concentration in various regions of the thin fi lm stack. The concentration calculations assume CZTS fi lms approximately 700 nm thick. The absolute concen-trations are estimated based on a reference sample consisting of Mo/NaF, known quantity/CZTS where the sample was not annealed. The known sodium concentration was then used to correlate SIMS counts with known sodium concentration.

Sodium distribution in CZTS

Starting NaF thickness Percent of total sodium found at the surface

Concentration of Na in CZTS fi lm region (approx.)

1 nm 8.7% 3.9 × 10 18 cm −3

4.5 nm 8.7% 1.0 × 10 19 cm −3

23 nm 33.2% 2.0 × 10 19 cm −3

FULL

PAPER


This is consistent with the SIMS data, indicating that zinc has segregated to the surface in these samples, likely in the form of ZnS.

Top-down and cross-sectional SEM images ( Figure 3 ) reveal that a fi ne-grained layer (70–100 nm grain diameters) is present at the surface of samples which started with 4.5 nm or less of NaF. Based on the SIMS and XPS data, we identify this fi ne-grained material as ZnS. Beneath this layer, CZTS grains with diameters of 270–330 nm are visible for the samples starting with 4.5 nm NaF or less, while much larger grains (≈1 µm in diameter) are observed in the sample starting with 23 nm of NaF (Figure 3 , bottom). A comparison of the average grain diameters in these and other similarly prepared samples is displayed in Figure 4 (excluding the surface region). Each data point in this fi gure represents an average of measurements per-formed on over 40 grains. The sodium content of the 4.5 nm NaF sample is not suffi cient to produce micron-sized grains in CZTS, but appears suffi cient to begin the surfactant process; the average grain diameter in this sample is ≈20% larger than the sample with less NaF. Figure 4 also suggests that this pro-cess is a “threshold”-type of process, where a critical concen-tration of sodium is required to observe large grains. This is

judged from the large discontinuity in the grain size vs NaF thickness relationship after ≈5 nm.

The SIMS data suggest that excess ZnS is present in the CZTS region of the lower-sodium-content samples in Figure 3 a–c. As CZTS grains grow, excess Zn not incorpo-rated into the CZTS structure is expelled and must precipi-tate as ZnS. Precipitates are known to impede grain boundary motion by a process known as solute drag. [ 23 ] This effect could be limiting the grain growth in the sodium-free fi lms. Sodium appears to help the zinc accumulate at the surface or back inter-faces of the fi lm at the same time as grain growth is enhanced (Figure 1 b, 4 ). We suspect that these two phenomena are related. For example, at least one ternary Na-Zn-S compound goes through a eutectic point above 600 °C. [ 24 ] An enhanced-diffusivity liquid phase could help remove the solutes from an advancing grain boundary and thereby increase its mobility. [ 23 ]

Figure 5 displays the temperature-dependent photolumines-cence spectra of the CZTS samples with varying sodium con-tent: a) no NaF, b) 1 nm NaF, c) 4.5 nm NaF, and d) 23 nm NaF. A full explanation of the states contributing to the observed PL at 4 K in our materials has been published previously. [ 20 ] Briefl y, the luminescence corresponding to photon energies between



Figure 3. Top-down (top) and cross-sectional (bottom) SEM images of the CZTS samples on YTZP substrates with: a) no NaF, b) 1 nm NaF, c) 4.5 nm NaF, and d) 23 nm NaF.

Figure 2. XPS measurements (after annealing) of samples containing varying amounts of sodium, as determined from starting NaF thickness: a) Na signal, b) O signal, and c) Cu and Zn signals.

FULL P

APER


≈1–1.3 eV at 4 K involves carriers trapped in localized donor and acceptor states, while the luminescence corresponding to the high-energy shoulder involves carriers residing in the band tails. The temperature-evolution of the PL spectra of samples (a–c) are different, while the spectra of samples (c,d) are almost identical. In samples (a–c), as the Na content increases, the radiative PL loss at the highest temperatures decreases, i.e., the samples with more sodium display less non-radiative recom-bination at the highest temperatures. The similarity in the temperature-dependent PL spectra of samples (c,d) is remark-able, considering the large difference in grain boundary density between these two samples. If the grain boundaries in samples (c,d) were effi cient at trapping photogenerated carriers non-radiatively, then the large difference in grain boundary density would have produced a difference in the measured lumines-cence. This leads to the conclusion that the grain boundaries in these samples (with higher Na concentrations) cannot be effi -cient non-radiative recombination centers.

In contrast, the samples containing less sodium (Figure 5 a,b) display a reduced luminescence signal while having a



Figure 4. Average CZTS grain diameter as a function of starting NaF thickness. Data represent an average of over 40 grains for samples with varying sodium content. The small-grain region at the surface was not included in the averaging. Error bars represent standard deviations.

Figure 5. Temperature-dependent photoluminescence spectra of four CZTS samples of varying NaF content: a) no NaF, b) 1 nm NaF, c) 4.5 nm NaF, and d) 23 nm NaF.

FULL

PAPER


comparable grain size. This conclusively demonstrates that Na is strongly correlated with reduced non-radiative recombina-tion in CZTS. Importantly, we note that the quantity of sodium needed to suppress non-radiative recombination (4.5 nm starting NaF thickness) does not necessarily correspond to large-grained CZTS material (Figure 3,4 ). We also note that the fi ne-grained ZnS layer present on top of the lower-sodium sam-ples (Figure 3 ) is unlikely to interfere with these optical meas-urements as judged from the wide band gap of ZnS and the similarity of the spectra in (c,d) of Figure 5 .

Figure 6 presents a) raw and b) normalized integrated pho-toluminescence data for all four YTZP samples as a function of inverse temperature. From these plots, the similarity between the radiative recombination in the 4.5-nm NaF and 23-nm NaF samples is even more evident. It is also notable that the raw PL signal from the sample that did not contain sodium is lower

over the entire range of temperatures, despite the fact that it had a similar grain boundary density to the 1-nm and 4.5-nm NaF samples. Even at the lowest temperatures, the difference between the no-sodium sample and the 23 nm NaF sample is approximately 50%. If sodium does indeed passivate grain boundaries, then the unpassivated boundaries would trap car-riers generated within a minority carrier diffusion length, causing them to display a partially quenched PL signal. An opti-cally inactive grain boundary region as thin as 4–5 nm could account for the 50% reduction in PL signal observed from the no-sodium sample at 4 K (details of this calculation can be found in the Supporting Information).

Various regions of the normalized PL data (Figure 6 b) can be fi t to an Arrhenius relationship to estimate the activation energy for the dominant carrier redistribution process. [ 25 ] In general, the activation energies can be understood as the energy



Figure 6. a) Raw and b) normalized integrated photoluminescence of CZTS samples on YTZP substrates as a function of inverse temperature. Region 1: 0–80 K, region 2: 95–170 K, region 3: 185–250 K. c) Temperature-dependent integrated PL intensity of the 23 nm NaF and glass reference sample overlaid.

FULL P

APER


for de-trapping of carriers from localized states and their redis-tribution to different states. This activation energy is related to the depth of the state from the band edge. Thus, a higher activation energy implies the involvement of deeper states. In region 1 (0–80 K), there is little change in PL intensity as a function of temperature for any of the samples. Throughout region 2 (95–170 K), the PL signals of all four samples decrease with activation energies of approximately 25–30 meV (Table 2 ). Since this energy is similar in all samples, we infer that this process is related to the CZTS bulk. In region 3 (185–250 K), a higher-activation-energy process (i.e., a deeper recombination channel) becomes active in the samples with low sodium, while the activation energies for the high-sodium samples remain ≈25–30 meV. The process activated for the low-sodium samples in region 3, the highest-temperature region, is consistent with grain boundary effects, which would become increasingly dom-inant as carriers travel tens of nanometers from the bulk radia-tive states to the boundary region. This activation energy differ-ence at high temperatures further supports the conclusion that the sodium is active mainly along grain boundaries and may not necessarily impact bulk states.

Figure 6 c displays the temperature-dependent photolumines-cence data from the reference CZTS sample on soda-lime glass overlaid with the 23-nm NaF sample on YTZP. The sample on glass displays the same degree of passivation as those with higher sodium content (Figure 1 ). Additionally, the average grain diameter in this sample is ≈1 µm (data not shown). One explanation for why large grains and effi cient passivation are observed for the glass sample despite its lower Na content could be the presence of additional impurities from the glass, which are also known to be benefi cial (e.g., potassium). [ 26,27 ]

4. Conclusion

In conclusion, we have shown that CZTS samples without sodium contain non-radiative defects, which are effi ciently passi-vated by the addition of sodium (NaF starting thickness ≈ 4.5 nm). We further validate prior results, which indicate that the most likely location for sodium in CZTS thin fi lms is along surfaces and grain boundaries, and that surface sodium is correlated with the presence of oxygen. Once passivated, we demonstrate that large differences in the grain boundary density in CZTS thin fi lms produce no measurable differences in the radiative and non-radiative decay profi les (Figure 5 c,d and Figure 6 a,b). This surprising result indicates that the grain boundaries have

become passive, within the measurement accuracy of our apparatus.

We also show that the quantity of sodium needed to passi-vate grain boundaries (i.e., starting NaF thickness ≈ 4.5 nm) is not suffi cient to produce large grains (>1 µm) in CZTS. Instead, this quantity appears to be the critical concentration needed to begin a surfactant type of process as evidenced by a 20% increase in average grain diameter compared to a sample with 0–2.5 nm of starting NaF thickness.

The process of grain growth appears to be correlated with the re-distribution of Zn in the thin fi lm. Increases in sodium concentration correlate with segregation of zinc towards the front and back surfaces of the CZTS fi lm. As CZTS grains grow, excess zinc is expelled to the surfaces. We hypothesize that, in the absence of sodium, the precipitation of ZnS causes solute drag on the grain boundaries, which limits the grain size in CZTS fi lms. The re-distribution of the zinc out of the fi lm bulk could therefore be closely related to improved grain boundary mobility and larger grains. This surfactant behavior could be due to the formation of a liquid sodium-sulfur, as reported recently, [ 28 ] or sodium-zinc-sulfur [ 24 ] phase at elevated temperatures, which would aid in mass transport along the grain boundaries during annealing. Finally, the similarity in the shape of the photoluminescence spectra for fi lms with grains of substantially different sizes (e.g., Figure 5 c,d) confi rms that the donor-acceptor pair type recombination that has been observed in this system [ 20 ] takes place within the grain bulk and not the grain boundary region.

Supporting Information Supporting Information is available from the Wiley Online Library or from the author.

Acknowledgements This material was based upon work supported by the U.S. Department of Energy under Award Number DE-EE0006334. The information, data, or work presented herein was funded in part by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specifi c commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or refl ect those of the United States Government or any agency thereof. The authors would like to acknowledge M. Spector for the Mo depositions.

Received: May 22, 2014 Revised: June 26, 2014

Published online:

[1] D. B. Mitzi , O. Gunawan , T. K. Todorov , K. Wang , S. Guha , Sol. Energy Mater. Sol. Cells 2011 , 95 , 1421 .



Table 2. Activation energies ( E a ) of carrier redistribution processes in CZTS samples with varying sodium content in different temperature regimes.

Extracted activation energies

Sample E a in region 2 (95–170 K) [meV]

E a in region 3 (185–250 K) [meV]

No sodium 24 106

1 nm NaF 25 67

4.5 nm NaF 28 29

23 nm NaF 23 24

FULL

PAPER


[2] I. Repins , N. Vora , C. Beall , S.-H. Wei , Y. Yan , M. Romero , G. Teeter , H. Du , B. To , M. Young , R. Noufi , Kesterites and Chalco-pyrites: A Comparison of Close Cousins , U.S. Department of Energy, 2011 .

[3] P. T. Erslev , J. W. Lee , W. N. Shafarman , J. D. Cohen , Thin Solid Films 2009 , 517 , 2277 .

[4] J. V. Li , D. Kuciauskas , M. R. Young , I. L. Repins , Appl. Phys. Lett. 2013 , 102 , 163905 .

[5] T. Prabhakar , N. Jampana , Sol. Energy Mater. Sol. Cells 2011 , 95 , 1001 . [6] D. Rudmann , D. Brémaud , H. Zogg , A. N. Tiwari , J. Appl. Phys.

2005 , 97 , 084903 . [7] A. Rockett , Thin Solid Films 2004 , 480 , 2 . [8] W. M. Hlaing Oo , J. L. Johnson , A. Bhatia , E. A. Lund , M. M. Nowell ,

M. A. Scarpulla , J. Electron. Mater. 2011 , 40 , 2214 . [9] T. Nakada , D. Iga , H. Ohbo , A. Kunioka , Jpn. J. Appl. Phys. 1996 , 36 ,

732 . [10] U. Rao , H. W. Schock , Appl. Phys. A 1999 , 69 , 131 . [11] H. Zhou , T.-B. Song , W.-C. Hsu , S. Luo , S. Ye , H.-S. Duan , C.-J. Hsu ,

W. Yang , Y. Yang , J. Am. Chem. Soc. 2013 , 135 , 15998 . [12] L. Kronik , D. Cahen , H. W. Schock , Adv. Mater. 1998 , 10 , 31 . [13] T. Nakada , Electron. Mater. Lett. 2012 , 8 , 179 . [14] Y. Yan , C.-S. Jiang , R. Noufi , S.-H. Wei , H. R. Moutinho ,

M. M. Al-Jassim , Phys. Rev. Lett. 2007 , 99 , 235504 . [15] J. B. Li , V. Chawla , B. M. Clemens , Adv. Mater. 2012 , 24 , 720 . [16] A. R. Jeong , W. Jo , S. Jung , J. Gwak , J. H. Yun , Appl. Phys. Lett. 2011 ,

99 , 082103 . [17] J. Li , D. B. Mitzi , V. B. Shenoy , ACS Nano 2011 , 11 , 8613 .

[18] D. W. Niles , M. Al-Jassim , K. Ramanathan , J. Vac. Sci. Technol. A. 1999 , 17 , 291 .

[19] R. Haight , X. Shao , W. Wang , D. B. Mitzi , Appl. Phys. Lett. 2014 , 104 , 033902 .

[20] T. Gershon , B. Shin , N. Bojarczuk , T. Gokmen , S. Lu , S. Guha , J. Appl. Phys. 2013 , 114 , 154905 .

[21] K. Granath , M. Bodegard , L. Stolt , Sol. Energy Mater. Sol. Cells 2000 , 60 , 279 .

[22] B. Shin , O. Gunawan , Y. Znu , N. A. Bojarczuk , S. Jay Chey , S. Guha , Prog. Photovoltaics 2013 , 21 , 72 .

[23] Y.-M Chiang , D. Birnie III , W. David Kingery , Physical Ceramics , John Wiley & Sons, Inc ., New York 1997 .

[24] V. Tomashyk , P. Feychuk , L. Shcherbak , Ternary Alloys Based on II-VI Sem-iconductor Compounds , Taylor & Francis Group , Boca Raton, FL 2014 .

[25] Y.-H. Cho , G.-H. Gainer , A. J. Fischer , J. J. Song , Appl. Phys. Lett. 1998 , 73 , 1370 .

[26] A. Chirila , P. Reinhard , F. Pianezzi , P. Bloesch , A. R. Uhl , C. Fella , L. Kranz , D. Keller , C. Gretener , H. Hagendorfer , D. Jaeger , R. Erni , S. Nishiwaki , S. Buecheler , A. N. Tiwari , Nat. Mater. 2013 , 12 , 1101 .

[27] M. Johnson , S. V. Baryshev , E. Thimsen , M. Manno , X. Zhang , I. V. Veryovkin , C. Leighton , E. S. Aydil , Energy Environ. Sci. 2014 , 7, 1931 .

[28] C. M. Sutter-Fella , J. A. Stückelberger , H. Hagendorfer , F. L. Mattina , L. Kranz , S. Nishiwaki , A. R. Uhl , Y. E. Romanyuk , A. N. Tiwari , Chem. Mater. 2014 , 26 , 1420 .

[29] A. Nagaoka , H. Miyake , T. Taniyama , K. Kakimoto , Y. Nose , M. A. Scarpulla , K. Yoshino , Appl. Phys. Lett 2014 , 104 , 152101 .



Documents

The Role of Sodium as a Surfactant and Suppressor of Non-Radiative Recombination at Internal Surfaces in Cu 2 ZnSnS 4