ARTICLE

pubs.acs.org/cm

Earth Abundant Element Cu2Zn(Sn1
xGex)S4 Nanocrystals for Tunable

Band Gap Solar Cells: 6.8% Efficient Device Fabrication
Grayson M. Ford,† Qijie Guo,† Rakesh Agrawal,*,† and Hugh W. Hillhouse*,†,‡
†
Chemical Engineering, Purdue University, West Lafayette, Indiana 47907, United States
bS Supporting Information
ABSTRACT: Cu2Zn(Sn1
xGex)S4 nanocrystals have been synthesized via batch reaction in
oleylamine with no additional surfactants present. The nanocrystals are knife-coated on
molybdenum substrates and then selenized to form a dense layer of Cu2Zn(Sn1
xGex)-
(S,Se)4, which is then used as the photoabsorbing layer in a thin film solar cell. The band gaps
Downloaded by MARTIN LUTHER UNIV HALLE WITT at 08:17:45:551 on July 04, 2019

of the nanocrystals and the resulting solar cells are demonstrated to be controlled by adjusting
the Ge/(GeþSn) ratio of the nanocrystal synthesis precursors. Solar cells fabricated from
Cu2ZnGeS4 nanocrystal films yielded a power conversion efficiency of 0.51%. However, Cu2Zn
(SnxGe1
x)S4 nanocrystals with a Ge/(GeþSn) ratio 0.7 yielded devices with an efficiency of
6.8% when synthesized to be Cu-poor and Zn-rich. This result opens the possibility of forming
Ge gradients to direct minority carriers away from high recombination interfaces and significantly
from https://pubs.acs.org/doi/10.1021/cm2002836.

improve the device efficiency of CZTSSe-based solar cells.

KEYWORDS: CZTS, CZTGS, CZTGSSe, CZTSSe, germanium, thin film, photovoltaic, low-cost

E arth abundant materials are a requirement for solar cells to

contribute power on the multiterawatt scale.1 Cu2ZnSn-
(S,Se)4 is of particular interest due to the achievement of 9.6%
previous CZTS syntheses involved sulfur addition anywhere
between room temperature and 300 °C. In contrast, we find
the temperature at which sulfur is added plays a critical role in
efficient laboratory scale devices2 using a nonvacuum, hydrazine- achieving the desired crystal structure and stoichiometry.15,16
based deposition process. Although hydrazine poses safety Powder X-ray Diffraction (PXRD, Figure 1) shows the
challenges from a manufacturing perspective, an alternative prominent peaks observed in the synthesized nanocrystals
nonvacuum deposition approach has achieved 7.2% efficiency correspond with those of tetragonal CZTS (JCPDS no.
via the selenization of Cu2ZnSnS4 (CZTS) nanocrystals depos- 26
0575) and CZGS (JCPDS no. 25
0327). CZTS has been
ited by roll coating.3 The resulting Cu2ZnSn(S,Se)4 (CZTSSe) reported to have two related tetragonal crystal stuctures, known
film has a band gap of approximately 1.1 eV, similar to CuInSe2. as stannite (I42m) and kesterite (I4).17 It is interesting to note
CuInSe2 device performance is improved by incorporation that the stannite phase has the same symmetry as a specific
of Ga,4
6 which widens the band gap. We hypothesized that disordering of the kesterite phase; the I42m symmetry is the
similar bandgap tuning and improvements to CZTS devices result of a relatively low-energy Cu
Zn cation exchange occur-
may be possible by substituting a lower atomic number group IV ring randomly in the kesterite phase.18 Although the similarity in
element for some of the Sn. Substitution with germanium is an atomic numbers between Cu, Zn and Ge contribute to the
attractive route, as germanium’s crustal abudance of 1.5 mg/kg7 difficultly in distinguishing between stannite and kesterite using
(or 1.6 mg/kg8) compares favorably with the 2.3 mg/kg PXRD,18 peak broadening11,13 further obscures proper identifi-
crustal abundance of Sn, the rarest element in CZTS.7 Here we cation. The tetragonal CZGS structure found in experimental
report a synthesis of Cu2Zn(Sn1
xGex)S4 (CZTGS) nanocryst- studies has been typically assigned to the stannite phase,19,20 but
als and the performance of Cu2Zn(Sn1
xGex)(SySe1
y)4 the values of c/a e 2 indicate the kesterite phase (possibly with
(CZTGSSe) solar cells fabricated by sintering the nanocrystal cation disorder) may be a more accurate description.18 The most
films with elemental selenium vapor. We find that the band intense peaks shared by the kesterite and stannite phases agree
gap of the CZTGS nanocrystals and the CZTGSSe solar cell well with the most intense peaks observed in Figure 1, indicating
can be rationally controlled by adjusting the Ge/(SnþGe) the nanocrystals have a tetragonal crystal structure. A shoulder
ratio. observed to the left of the (112) peak could be caused by a Cu2S
There have been previous reports of band gap control in impurity phase,21 stacking faults related to cation disordering
(ZnS)x(Cu2SnS3)1
x nanocrystals.9,10 However, such structures similar to the faulting in CuInSe2,22 or an orthorhombic CZGS
do not preserve the kesterite cation ordering observed in CZTS impurity phase.23
25 CZTGS nanocrystals were synthesized by
(or CZTGS) which appears to be critical to obtain good replacing GeCl4 in the CZGS synthesis with an equivalent molar
photovoltaic performance. CZTS nanocrystal syntheses pre-
viously reported11
13 involve the combination of metal salt Received: January 27, 2011
precursors and sulfur in the presence of oleylamine, and the Revised: March 23, 2011
CZTGS synthesis procedure reported here is similar.14 The Published: April 28, 2011

r 2011 American Chemical Society 2626 dx.doi.org/10.1021/cm2002836 | Chem. Mater. 2011, 23, 2626–2629
Chemistry of Materials ARTICLE

Figure 3. Various atomic ratios determined by energy-dispersive spec-

Figure 1. PXRD of Cu2Zn(Sn1
xGex)S4 nanocrystals for x = 0 (red), troscopy (EDS) showing the nanocrystals are close to stoichiometric.
0.5 (green), and 1.0 (blue). Also included for comparison are the simulated See text for molecular formulas.
PXRD patterns for the kesterite structures of CZTS and CZGS.

Figure 4. UV
vis data depicting the increasing onset of absorption

with increasing Ge content in CZTS (red), CZTGS (green), and CZGS
(blue).

matrix effects of the nanocrystal film, the lower EDS accuracy for
lighter elements (i.e., sulfur) or organosulfur compounds formed
during nanocrystal synthesis.
Figure 2. TEM images of (a) CZTS, (b) CZTGS, and (c) CZGS. Figure 4 depicts UV
vis data for CZTGS nanocrystals with
(d) HR-TEM of CZGS nanocrystals. Ge/(SnþGe) ratios of 0.0, 0.5, and 1.0. The absorption is not
expected to be influenced by quantum confinement effects, as the
amount of Sn(acac)2Cl2.26 The expected (112) peak shift due to majority of the nanocrystals (with average diameters of 15.4,
Sn incorporation (increasing the lattice constants) is observed by 13.3, and 8.6 nm (N = 35) for CZTS, CZTGS and CZGS,
PXRD and shown in the inset of Figure 1. respectively) are much larger than the Bohr exciton radius of bulk
The TEM images in Figure 2 reveal that the CZTGS nano- CZTS, estimated to be 2.6 nm using the values reported by
crystals vary in size from 5 to 30 nm, with the relative number of Persson.27 The band gap is observed to increase with increasing
smaller sized particles increasing with increasing Ge content. The Ge content of the synthesis precursors. The expected band gap
composition of the synthesized CZTGS nanocrystals as deter- for CZTS is ∼1.5 eV, whereas the band gap for CZGS has been
mined by energy-dispersive X-ray spectroscopy (EDS) is shown reported to be 2.05
2.25 eV.28,29 Linear extrapolation of the
in Figure 3, with the observed elemental ratios normalized to nanocrystal light absorption versus the photon energy is typically
equal 1.0 for stoichiometric CZTGS. The molecular formulas used to determine the effective band gap. As shown in Figure 4, if
were determined to be Cu2Zn0.92Sn1.11S4.37, Cu2Zn0.97(Sn0.69- the CZTS data are extrapolated to yield a bandgap of 1.5 eV, then
Ge0.31)1.12S4.73, Cu2Zn1.02(Sn0.41Ge0.59)1.10S4.68, and Cu2Zn1.09- the CZGS nanocrystals would have a band gap of 1.94 eV. Note
Ge1.03S4.80 for nanocrystals synthesized with Ge/(SnþGe) pre- that this band gap is slightly lower than those previously reported
cursor ratios of 0.0, 0.3, 0.5, and 1.0, respectively. The cation for bulk CZGS and may be due to the inherent error in the
ratios of the nanocrystals are close to the expected stoichiometry, extrapolation. However, this value is distinctly different from the
but there is an apparent excess of sulfur. This could be due to band gaps of Cu2GeS3 (0.5 eV30), ZnS (>3.4 eV), or GeS2
2627 dx.doi.org/10.1021/cm2002836 |Chem. Mater. 2011, 23, 2626–2629
Chemistry of Materials ARTICLE

Figure 5. Current
voltage behavior of champion CZGSSe solar cell in Figure 6. Current
voltage behavior of champion CZTGSSe solar cell
dark and under simulated AM1.5G. Inset: EQE. under AM1.5G simulated illumination using a Ge/(GeþSn) ratio of
0.70. Inset: comparison of EQE for CZTSSe3 and CZTGSSe devices.
(3.4 eV31), and is consistent with Ge being incorporated into the
CZTS crystal structure. Band gaps determined from the EQE39 (Figure 6 inset) of the
To determine the suitability of the nanocrystals for solar cells, CZTGSSe device and a similarly made CZTSSe device3 are 1.40
we fabricated CZGS nanocrystal films into devices using meth- and 1.11 eV, respectively. Note that the same technique to
ods explained in detail elsewhere.3,32,33 Briefly, the nanocrystals determine band gap cannot be applied to the EQE of the CZGSSe
were thoroughly washed, and a paste of the nanocrystal ink was device since the decay in the EQE is most likely dominated by
formed by adding hexanethiol to the dried nanocrystals to yield a low minority carrier lifetime and mobilities, not the functional
formulation with ∼200 mg of nanocrystals per mL of solution. form of the absorption coefficient. The S/(SþSe) ratio in the
The paste was then coated onto a Mo-coated soda-lime glass devices were similar and were determined to be 0.54, 0.54, and
(SLG) substrate with a doctor-blading technique before being 0.50 for the CZTSSe, CZTGSSe and CZGSSe devices, respec-
annealed under Se atmosphere at 500 °C for 20 min. After tively (note that this analysis overestimates S content due to the
annealing, standard device fabrication steps were employed to presence of the CdS layer and the overlap of the S KR and Mo LR
obtain devices with a SLG/Mo/CZGSSe/CdS/i-ZnO/ITO/ X-ray emission peaks). As a result, variation in band gap is not
Ni
Al architecture and an area of 0.47 cm2. The current
attributed to selenium content. The data presented here demon-
voltage behavior (Figure 5) under AM1.5G simulated illumina- strate Ge incorporation can be used for band gap control of both
tion revealed a total area efficiency of 0.51%. The External CZTGS nanocrystals and the resulting solar cells, while retaining
Quantum Efficiency (EQE, Figure 5 inset) is observed to reach promising device efficiency.
0% at 950 nm, indicating the band gap is at most no larger than In summary, semiconductor nanocrystals were synthesized
1.3 eV. This agrees well with one reported CZGSe band gap of using Cu, Zn, Sn and Ge salts and were concluded to span
1.29 eV,34 but is lower than another report of 1.52 eV.35 Model- the CZTGS tetragonal system based on PXRD, TEM, EDS,
ing of CZGSe18 reveals the band gap can vary from 1.21
and UV
vis absorption. Tin-free CZGS nanocrystals fabricated
1.50 eV, with lower band gaps being indicative of lower quality into devices using a low cost roll coating approach had an
(higher disorder). This also indicates that comparisons of band efficiency of 0.51%. Device efficiency was improved to 6.8% using
gaps in the CZTGSSe material system should be made between Ge/(GeþSn) = 0.70 CZTGS nanocrystals with Cu-poor and
devices with similar minority carrier lifetimes and mobilities Zn-rich stoichiometry synthesized by an alternate method. This
(or at minimum similar quantum efficiency curves, EQE(λ)). important result establishes the prospect of significant device
Additionally, a relative loss of Ge was observed during annealing efficiency improvements by adjusting or layering the Ge content
under Se atmosphere.36 Another reason for the low efficiency is (similar to Ga gradients in CIGSe).
the use of relatively stoichiometric nanocrystals (stoichiometric
cation ratios). Typical high efficiency devices in this material class ’ ASSOCIATED CONTENT
are Cu-poor and Zn-rich,2,3,37 and our previous work with CZTS
nanocrystals has shown that composition control can lead to bS Supporting Information. Experimental details, nanocrys-
significant improvements in device performance.3,11 tal composition at different sulfur injection temperatures, and
We find that this is also the case in the CZTGS system. Although supporting PXRD and EDX data (PDF). This material is available
composition is still being optimized for devices, significant free of charge via the Internet at http://pubs.acs.org.
improvements can be obtained with high-Ge content CZTGS
nanocrystals synthesized using an alternate procedure found to ’ AUTHOR INFORMATION
result in a similar Cu-poor and Zn-rich composition.11,38 CZTGS Corresponding Author
nanocrystals synthesized with Ge/(GeþSn), Cu/(ZnþSnþGe) *E-mail: [email protected] (R.A.); [email protected] (H.W.H.).
and Zn/(SnþGe) ratios of 0.70, 0.80, and 1.20, respectively,
were fabricated into solar cells using the same methods described Present Addresses
‡
above. Champion total area device efficiency was found to be Chemical Engineering, University of Washington, Seattle,
6.8%, with the current
voltage behavior depicted in Figure 6. WA 98195, United States.
2628 dx.doi.org/10.1021/cm2002836 |Chem. Mater. 2011, 23, 2626–2629
Chemistry of Materials ARTICLE

’ ACKNOWLEDGMENT from the sum of ZnS and Cu2GeS3, which have peaks centered at 28.56
We are thankful for financial support from the NSF IGERT and 29.26° 2θ. However, this would disagree with the UV
vis data in
Figure 4.
program (award number 0903670-DGE). We also acknowledge (26) See Experimental Details section of the Supporting Information.
Samantha Hahn for her assistance in performing nanocrystal (27) Persson, C. J. Appl. Phys. 2010, 107, 053710.
syntheses. (28) Leon, M.; Levcenko, S.; Serna, R.; Gurieva, G.; Nateprov, A.;
Merino, J. M.; Friedrich, E. J.; Fillat, U.; Schorr, S.; Arushanov, E. J. Appl.
Phys. 2010, 108, 093502.
’ REFERENCES (29) Matsushita, H.; Ichikawa, T.; Katsui, A. J. Mater. Sci. 2005,
(1) Wadia, C.; Alivisatos, A. P.; Kammen, D. M. Environ. Sci. Technol. 40, 2003.
2009, 43, 2072. (30) Collaboration: Authors and editors of the volumes I.II/17H-
(2) Todorov, T. K.; Reuter, K. B.; Mitzi, D. B. Adv. Mater. 2010, 17I-41E: Cu2GeS3 crystal structure, lattice parameters, physical proper-
22, E156. ties. SpringerMaterials—The Landolt
B€ornstein Database; Madelung,
(3) Guo, Q.; Ford, G. M.; Yang, W.; Walker, B.; Agrawal, R.; O., R€ossler, U., Schulz, M., Eds.; Springer: New York; http://www.
Hillhouse, H. W. J. Am. Chem. Soc. 2010, 132, 17384–17386. springermaterials.com, DOI: 10.1007/10717201_135.
(4) It is interesting to note that the optimum Ga concentration for (31) Nikolic, P. M.; Popovic, Z. V. J. Phys. C: Solid State Phys. 1979,
Cu(In,Ga)Se2 solar cells coincides with a minimization of tetragonal 12, 1151.
distortion (c/a = 2).5,6 The tetragonal distortion of Cu2Zn(Sn,Ge) (32) Guo, Q.; Hillhouse, H. W.; Agrawal, R. J. Am. Chem. Soc. 2009,
(S,Se)4 for kesterite is c/a < 2, whereas for stannite, c/a > 2.14 131, 11672.
(5) Abou-Ras, D.; Caballero, R.; Kaufmann, C; Nichterwitz, M; (33) Guo, Q.; Ford, G. M.; Hillhouse, H. W.; Agrawal, R. Nano Lett.
Sakurai, K.; Schorr, S.; Unold, T.; Schock, H. W. Phys. Status Solidi 2008, 2009, 9, 3060.
2, 135. (34) Schleich, D. M.; Wold, A. Mater. Res. Bull. 1977, 12, 111.
(6) Hanna, G.; Jasenek, A.; Rau, U.; Schock, H. W. Thin Solid Films (35) Lee, C.; Kim, C. J. Korean Phys. Soc. 2009, 37, 364.
2001, 387, 71. (36) See the Supporting Information, Figure S7 and S8.
(7) In CRC Handbook of Chemistry and Physics, 91st ed. (Internet (37) Katagiri, H.; Jimbo, K.; Yamada, S.; Kamimura, T.; Maw, W.;
version); Haynes, W. M., Ed.; CRC Press/Taylor and Francis, Boca Fukano, T.; Ito, T.; Motohiro, T. Appl. Phys. Express 2008, 1, 041201.
Raton, FL, 2011. (38) Altered CZTGS nanocrystal synthesis procedure details dis-
(8) Holl, M.; Kling, M.; Schroll, E. Ore Geol. Rev. 2007, 30, 145–180. cussed in the Supporting Information, Figures S2 and S5.
(9) Dai, P.; Shen, X.; Lin, Z.; Feng, Z.; Xu, H.; Zhan, J. Chem. (39) Zoppi, G.; Forbes, I.; Miles, R. W.; Dale, P. J.; Scragg, J. J.; Peter,
Commun. 2010, 46, 5749–5751. L. M. Prog. Photovoltaics: Res. Appl. 2008, 17, 315–319.
(10) Liu, Q.; Zhao, Z.; Lin, Y.; Guo, P.; Li, S.; Pan, D.; Ji, X. Chem.
Commun. 2011, 47, 964–966.
(11) Guo, Q.; Hillhouse, H. W.; Agrawal, R. J. Am. Chem. Soc. 2009,
131, 11672.
(12) Riha, S. C.; Parkinson, B. A.; Prieto, A. L. J. Am. Chem. Soc.
2009, 131, 12054.
(13) Steinhagen, C.; Panthani, M. G.; Akhavan, V.; Goodfellow, B.;
Koo, B.; Korgel, B. A. J. Am. Chem. Soc. 2009, 131, 12554.
(14) In a typical Cu2ZnGeS4 synthesis, 1 mmol of Cu(acac)2,
0.5 mmol of Zn(acac)2 3 xH20, 0.5 mmol of GeCl4, and 15 mL of
oleylamine are added to a reaction vessel using standard air-free
techniques. While increasing the temperature to 280 °C, 2 mmol of
sulfur is added at 160 °C. The reaction is held at 280 °C for 1 h before
cooling, washing, and collecting the product. See the Supporting
Information for further detail on the synthesis procedure.
(15) Sulfur injections at 130 or 225 °C result in Zn deficient CZGS
nanocrystals, even when up to 20% excess Zn is used (see Figures S1 and
S2 in the Supporting Information).
(16) The injections at 225 °C result in nanocrystals with an upredicted
(112) peak location (see the Supporting Information, Figure S5).
(17) Nakamura, S.; Maeda, T.; Wada, T. Jpn. J. Appl. Phys. 2010,
49, 121203.
(18) Chen, S.; Gong, X. G.; Walsh, A.; Wei, S. Phys. Rev. B 2009,
79, 165211.
(19) Parasyuk, O. V.; Piskach, L. V.; Romanyuk, Y. E.; Olekseyuk,
I. D.; Zaremba, V. I.; Pekhnyo, V. I. J. Alloys Cmpd. 2005, 397, 85.
(20) Moodie, A. F.; Whitfield, H. J. Acta Crystallogr., Sect. B 1986,
42, 236–247.
(21) See the Supporting Information, Figure S4, for a comparison of
the expected PXRD peak positions of various impurity phases in the
CZGS system.
(22) Guo, Q.; Kim, S. J.; Kar, M.; Shafarman, W. N.; Birkmire, R. W.;
Stach, E. A.; Agrawal, R.; Hillhouse, H. W. Nano Lett. 2008, 8, 2982–2987.
(23) Doverspike, K.; Dwight, K.; Wold, A. Chem. Mater. 1990, 2, 194.
(24) Chen, S.; Walsh, A.; Luo, Y.; Yang, J.; Gong, X. G.; Wei, S. Phys.
Rev. B 2010, 82, 195203.
(25) This synthesis could form ZnS and Cu2GeS3 instead of CZGS;
the peak at 29.06° 2θ that is assigned to the CZGS (112) could result

2629 dx.doi.org/10.1021/cm2002836 |Chem. Mater. 2011, 23, 2626–2629