Sulfide Nanocrystal Inks for Dense Cu(In1−xGax) - ACS Publications

Jun 11, 2009 - Qijie Guo, Grayson M. Ford, Hugh W. Hillhouse,* and Rakesh Agrawal*. School of Chemical Engineering and the Energy Center, Purdue ...
0 downloads 0 Views 241KB Size
NANO LETTERS

Sulfide Nanocrystal Inks for Dense Cu(In1-xGax)(S1-ySey)2 Absorber Films and Their Photovoltaic Performance

2009 Vol. 9, No. 8 3060-3065

Qijie Guo, Grayson M. Ford, Hugh W. Hillhouse,* and Rakesh Agrawal* School of Chemical Engineering and the Energy Center, Purdue UniVersity, West Lafayette, Indiana 47907 Received May 13, 2009; Revised Manuscript Received June 4, 2009

ABSTRACT Recent developments in the colloidal synthesis of high quality nanocrystals have opened up new routes for the fabrication of low-cost efficient photovoltaic devices. Previously, we demonstrated the utility of CuInSe2 nanocrystals in the fabrication of CuInSe2 thin film solar cells. In those devices, sintering the nanocrystal film yields a relatively dense CuInSe2 film with some void space inclusions. Here, we present a general approach toward eliminating void space in sintered nanocrystal films by utilizing reactions that yield a controlled volume expansion of the film. This is demonstrated by first synthesizing a nanocrystal ink composed of Cu(In1-xGax)S2 (CIGS). After nanocrystal film formation, the nanocrystals are exposed to selenium vapor during which most of the sulfur is replaced by selenium. Full replacement produces a ∼14.6% volume expansion and reproducibly leads to good dense device-quality CIGSSe absorber films with reduced inclusion of void space. Solar cells made using the CIGSSe absorber films fabricated by this method showed a power conversion efficiency of 4.76% (5.55% based on the active nonshadowed area) under standard AM1.5 illumination.

The development of low-cost efficient solar cells is one of the ultimate solutions for the renewable and clean energy challenge. Advances from colloidal science including the synthesis of high-quality nanocrystals have opened up new routes to address this challenge and have recently been reviewed.1 In one route, quantum confined semiconductor nanocrystals exhibiting multiexciton generation are being explored for next generation high-efficiency solar cells. In another approach, semiconductor nanocrystals are used as precursors to yield dense bulk semiconductor films via an ultra-low-cost fabrication route to solar cells by taking the advantage of solution-based processes. Recently, photovoltaic devices based on various semiconductor nanocrystals, such as CdTe,2 Cu2S,3 Pb(Sx,Se1-x),4,5 and CuInSe26,7 have been demonstrated. Among the many materials investigated for photovoltaic applications, copper indium gallium diselenide (CIGSe) and related materials are some of the most promising candidates for the development of low-cost thin-film solar cells.8-10 Numerous approaches for the fabrication of CIGSe solar cells based on nonvacuum methods by selenization of various precursor materials for the formation of a device quality CIGSe absorber layer have been demonstrated.6,11-16 However, the ability to control the composition of the CIGSe thin film remains to be a key challenge because the optical and electrical properties in the CIGSe absorber depends greatly on the composition.10 * Corresponding authors, [email protected] and [email protected]. 10.1021/nl901538w CCC: $40.75 Published on Web 06/11/2009

 2009 American Chemical Society

An alternative approach is to use CIGSe nanocrystals with the desired composition and crystal structure as the building blocks for the fabrication of CIGSe absorber thin films. Previously, we have demonstrated the utilization of CuInSe2 nanocrystals for the fabrication of functional CuInSe2 thin film solar cells.6 However, thermal treatment of selenide precursors does not expand in volume and thus may leave some voids in the final film. Here, we address this issue using a nanocrystal film of Cu(In1-xGax)S2 (CIGS) as the source material while maintaining the advantage of composition control at all length scales. The substitution of Se into the CIGS matrix creates volume expansion, which reduces void space and reproducibly leads to a better device quality dense CIGSe absorber layers. The lattice volume expansion assuming complete replacement of S with Se is ∼14.6%, calculated based on the lattice parameters of CuInS2 (a ) 5.52 Å, c ) 11.12 Å) and CuInSe2 (a ) 5.78 Å, c ) 11.62 Å). Initial devices made using such fabricated Cu(In, Ga)(S,Se)2 (CIGSSe) absorber films showed a photon to electricity conversion efficiency of 4.76% under AM1.5G illumination (5.5% based on active nonshadowed area). Several methods have been reported in the literature for the synthesis of ternary chalcogenide nanocrystals.6,7,17-22 Here we adapted our synthesis of the CuInSe2 nanocrystal ink for the synthesis of various CuInS2 nanocrystals by replacing Se with S in the synthesis procedure; please see the experimental section in Supporting Information for details in the preparation of the CuInS2 nanocrystals.6 Figure 1a

Figure 1. PXRD patterns of the (a) CuInS2 and Cu(In1-xGax)S2 nanocrystal films with various Ga concentrations (x) deposited on molybdenumcoated soda lime glass and (b) expanded view of the (112) peak showing the peak shift due to incorporation of Ga.

shows the powder X-ray diffraction (PXRD) pattern of the as-synthesized CuInS2 nanocrystals on molybdenum-coated soda lime glass. The observed X-ray diffraction peaks at 28.04, 32.56, 46.6, and 55.28 can be indexed to the (112), (200), (204), and (312) reflections of the CuInS2 crystal structure. The observed peaks match very well with the reference JCPD data (PDF #38-0777). The crystalline size of the CuInS2 nanoparticles calculated estimated using the Scherrer equation is ∼15 nm. A transmission electron microscopy image of the as-synthesized CuInS2 nanocrystals is shown in Figure S1 of the Supporting Information, and the average particle size shows good agreement with the estimated crystalline size from PXRD. The average composition of the sulfide nanoparticles determined by energy dispersive X-ray (EDX) is Cu1.00(0.02In0.95(0.03S2, which is nearly stoichiometric within the errors of EDX. Previously, we have demonstrated the capability to expand the synthesis of CuInSe2 for the preparation of Cu(In1-x,Gax)Se2 and CuGaSe2 nanocrystals by replacing a fraction or all the indium precursors with corresponding gallium precursors.23 For sulfur-free materials, alloying with gallium is desired because it increases the band gap energy of the semiconductor leading to a higher achievable open circuit voltage in the final device. Thus, here we synthesize Cu(In1-x,Gax)S2 and CuGaS2 nanocrystals by adapting the same approach. Figure 1 shows PXRD patterns of the assynthesized Cu(In1-x,Gax)S2 nanocrystal inks with varied Ga content. The Ga/(In+Ga) ratio, denoted as x, of the resulting Cu(In1-x,Gax)S2 nanocrystals was determined by EDX. Figure 1b shows the expanded view of the (112) diffraction peak of the various Cu(In1-x,Gax)S2 nanocrystals. The (112) peak of the Cu(In1-x,Gax)S2 nanocrystals showed a systematic shift toward higher degree 2θ with increasing amount of Ga, i.e., higher x. The shift in the diffraction peak is expected due to decrease in the unit cell lattice parameters with the incorporation of Ga, which has smaller atomic size relative to In. In addition, UV-vis absorption spectra of the Cu(In1-x,Gax)S2 nanocrystals with various x are shown in Nano Lett., Vol. 9, No. 8, 2009

Figure S3 of the Supporting Information. The spectra show a blue shift with increasing x indicating an increase of band gap energy with increasing Ga. The as-prepared nanocrystal inks may be dispersed in various organic solvents such as toluene to form a stable ink solution, where typical concentration of the ink is ∼10 mg/mL. The final ink solution may be applied directly on substrates to form a thin film coating. For photovoltaic demonstrations, molybdenum-coated soda glass was used as the substrate. Nanocrystal thin films were prepared by drop casting a desired amount of ink on Mo coated soda lime glass directly and letting the solvent evaporate slowly. Large area (1 square inch) crack-free thin films were prepared in this manner. Multiple coatings may be applied to fabricate thin film with desired thickness. Figure 2a is a cross section image of the nanocrystal film deposited on a Mo-coated soda lime glass substrate showing a very dense packing of the CuInS2 nanocrystals. The CuInS2 nanocrystal film can be easily converted to a predominantly selenide material, CuIn(SySe1-y)2, by annealing the film in a small graphite enclosure containing the film and elemental Se pellets. From thermodynamic calculations, the free energy of reaction for CuInS2(s) + Se2(g) T CuInSe2(s) + S2(g) at 800 K is ∼0.3 kJ/mol, using thermodynamic constants of individual species from literature.24,25 Therefore, the reaction may be driven in either direction by adjusting the vapor pressure of Se and S. In our case, there is no added sulfur vapor, and the driving mechanism for the replacement of S in CuInS2 by Se is the elevated partial pressure of Se around the film during selenization. The graphite enclosure is not gastight but is used to help provide a uniform thermal environment and increase the partial pressure of Se above the film and allow sulfur released from CuInS2 to escape. Field emission scanning electron microscopy (FE-SEM) images of the CuInS2 nanocrystal film selenized at 500 and 450 °C are shown in parts b and c of Figure 2, respectively. At 500 °C (Figure 2b), the selenized film exhibits large and densely 3061

Figure 2. FE-SEM images of the nanocrystal derived absorber film at various stages of the process: (a) as-casted CuInS2 nanocrystal film on Mo-coated soda lime glass substrate, (b) after selenization at 500 °C for 45 min showing complete recrystallization into large and densely packed grains and (c) after selenization under Se vapor at 450 °C for 45 min showing the unique bilayer film formation of a dense large grain layer on top of a particle-like layer, and (d) CuInS2 nanocrystal film annealed under a sulfur atmosphere at 500 °C for 45 min which do not show much recrystallization of the nanocrystals.

packed grains, an important feature for device quality absorbers. Significantly different and interesting morphology is observed for films selenized at 450 °C, Figure 2c. At 450 °C, the films typically exist as a bilayer structure after selenization, with a ∼700 nm thick top layer consisting of densely packed large grains and an amorphous and particlelike bottom layer. Secondary ion mass spectroscopy (SIMS) depth profile (data not shown) of the matrix elements in the selenized film shows that the composition of the film is uniform throughout the thickness of the film, indicating the particle-like bottom layer has also been converted into corresponding selenides. In comparison, similar CuInS2 nanocrystal precursor films subject to annealing under sulfur atmosphere do not show much sign of crystal regrowth as shown in Figure 2d, further signifying the role of selenium in the growth of the nanocrystals. PXRD patterns of films with varied Ga content (x) after selenization are shown in Figure 3. The diffraction peaks of the selenized CuInS2 (Figure 3b, x ) 0) film shifted to the left as compared to the CuInS2 nanocrystals (Figure 1b, x ) 0), which is expected due to the increase in the lattice parameters after selenization. Furthermore, the diffraction peaks sharpened after selenization indicating a growth in the crystalline domain of the grains in the final film, in agreement with the observations from FE-SEM as shown in Figure 2. Moreover, minor peaks characteristic of the desired chalcopyrite CISSe for photovoltaic applications are very prominent in the final film, for example the (110), (103), and (220) peaks. Similar sintering behaviors are found with the CIGS nanocrystal films after selenization. The diffraction peaks of the selenized CIGSSe films shift to higher degree 2θ with increasing Ga concentration as expected. Prior to device fabrication, a short etching using bromine in methanol solution is performed to reduce the overall 3062

roughness in the film formed after selenization. A brief KCN etching is typically followed to remove any residual copper selenides that might exist in the film after selenization and bromine etching. The final composition of the film was slightly copper poor, with Cu/In of 0.97, as determined by EDX. It is possible that a small amount of sulfur is remaining in the film. However, the exact amount of S could not be determined from EDX because the KR peak of sulfur (2.29 keV) overlaps with the LR peak of the molybdenum substrate (2.30 keV). Thus, a parallel sample prepared on bare soda lime glass was used to determine the degree of selenization, in which 5-6% of S is typically found in the film after selenization as determined by EDX. As a result, we refer our selenized absorber films either as CISSe or CIGSSe films. After chemical treatments, the films are fabricated into devices by chemical bath deposition of CdS (∼50 nm), rf sputtering of i-ZnO (∼ 50 nm) and rf sputtering of ITO (∼250 nm). The final devices are scribed into small areas (∼0.1 cm2) with a dab of silver paint as the front contact for performance evaluation (Figure S3a in Supporting Information). Currentvoltage (I-V) characteristic of the prototype solar cells fabricated using the selenized CISSe absorber is shown in Figure 4a. The CISSe device has a full area efficiency of 4.17% under standard AM1.5G illumination calibrated with an Oriel/VLSI certified c-Si reference solar cell. By exclusion of the shadowed area under the top contact (silver paint), the active area efficiency is calculated to be 5.14% (Voc ) 393 mV, Jsc ) 29.7 mA/cm2, FF ) 44%). The Voc and Jsc of the prototype device are comparable to other non-vacuumbased techniques. However, lower FF due to high series resistance causes inferior performance of the present device. Typical small area CISSe solar cells obtained through this method show consistent active area efficiencies in the range of 4-5% (Figure S4 in Supporting Information). CIGSSe solar cells can be made using the Cu(In1-xGax)S2 nanocrystals inks in a similar manner. The CIGSSe device has a full area efficiency of 4.76% (5.55% based on active nonshadowed area). Typical I-V characteristic of the selenized CIGSSe solar cells are shown in Figure 4b. The CIGSSe devices typically show a higher Voc (455 mV) due to increased band gap energy with Ga addition and slightly lower short circuit current (23.7 mA/cm2) than the CISSe devices. The fill factor for the CIGSSe device is 51.5%, slightly higher than that of the CISSe device. Overall, the fill factors of the CISSe and CIGSSe devices are lower than those typically reported for high-efficiency CIGSSe solar cells, partly due to the large amount of dark areas (∼15-20% of whole device area) from the silver contacts and shadow from the probe. Since no laboratory light source can precisely replicate the AM1.5G spectra, the measured Jsc needs to be corrected to reflect this discrepancy. In order to make this correction the spectral mismatch factor (M-Factor)26 was calculated to determine how the measured Jsc compared to a Jsc measured under true AM1.5G illumination. The M-Factor was calculated using the AM1.5G filtered Xe arc-lamp source spectrum, the ASTM G-173 standard AM1.5GT spectrum, the CIGSSe test cell spectral responsivity, and c-Si reference Nano Lett., Vol. 9, No. 8, 2009

Figure 3. PXRD patterns of (a) selenized CuInS2 and Cu(In1-xGax)S2 nanocrystal films with various Ga concentrations (x) and (b) expanded view of the (112) peak showing the peak shift due to increasing Ga concentration.

Figure 4. I-V characteristics of the completed (a) CISSe and (b) CIGSSe devices, respectively. The devices were scribed into small area devices (∼0.1 cm2) with a small dap of silver paint as the top contact. The efficiency of the cells was calculated per active area by excluding the shaded areas under standard AM1.5 illumination.

cell spectral responsivity and found to be 1.04. The M-Factor was not used to correct the Jsc or efficiency but was calculated to ensure that it is only a small correction. External quantum efficiencies (EQE) without light bias of both the CISSe and CIGSSe devices are shown in Figure 5. The CIGSSe devices are more efficient at the short wavelengths but inferior at longer wavelengths as compare to the CISSe device. The differences in the band gap energy due to incorporation of Ga are very apparent from the EQE curves. The estimated band gaps for CISSe and CIGSSe are 1.0 and 1.1 eV, respectively. The overall EQE for both the CISSe and CIGSSe devices are significantly lower than those typically reported for high-efficiency CIGSe devices. Also, the external quantum efficiencies for both the CISSe and CIGSSe drop significantly at longer wavelength as well. The inefficient collection of carriers at longer wavelength photons suggests a high recombination loss in the bulk of the film before reaching the depletion region. Furthermore, open circuit voltages obtained in the current CISSe (393 mV) and CIGSSe (455 mV) devices are significantly lower than Nano Lett., Vol. 9, No. 8, 2009

that of the reported high-efficiency devices (508 mV for CISe27 and 689 mV for CIGSe8), another indication of the high level of recombination losses in the absorber films. High recombination loss is likely a result of high majority carrier concentrations, directly proportional to nonradiative recombination, resulting from a high concentration of impurities. Figure 6 shows a plot of 1/C2 versus applied voltage of CISSe and CIGSSe devices, which are used to determine the free carrier concentration based on capacitance-voltage analysis. There is a range of values for chemical bath deposited CdS carrier concentrations (ND) reported in the literature. However, there is growing evidence that the carrier concentration in the CdS layer is about ND ) 1018 cm-3.28 Therefore a value of 1018 was assumed in our calculations for the carrier concentrations in the CISSe and CIGSSe thin films. If ND is 1019, the calculated NA values are virtually the same. If ND is 1017, the calculated NA values would be higher by a factor of 2. The free carrier concentrations calculated for the CISSe and CIGSSe devices are on the order of 1017 cm-3. Since the carrier concentration is inversely 3063

Conclusions. A novel and simple process for the fabrication of CIGSSe absorber layers using nonvacuum techniques based on the selenization of ternary (CuInS2) and quaternary sulfide (Cu(In1-xGax)S2) nanocrystal thin films is presented. The conversion of the nanocrystal precursor films into a device quality absorber layer is reproducibly achieved by mild thermal annealing under selenium vapor at temperatures as low as 450 °C. Active area cell efficiencies up to 5.55% (Voc ) 455 mV, Jsc ) 23.7 mA/cm2, FF ) 51.5%) have been demonstrated using the CIGSSe absorber layer fabricated. The initial results are exciting and a promising alternative for low cost fabrication of CIGSSe thin film solar cells. Optimization focused on reducing or eliminating the impurities in the film is underway and will be presented in future publications.

Figure 5. Plot of EQE without light for the CISSe and CIGSSe devices fabricated from the selenized sulfide nanocrystal precursors films. EQE of the device are lower than that of reported highefficiency devices due to high recombination losses.

Figure 6. Plot of inverse capacitance squared versus applied voltage for the CISSe and CIGSSe devices fabricated from the selenized sulfide nanocrystal precursor films. The calculated carrier concentrations are determined by a linear fit of the reverse and forward bias region.

related to the magnitude of the slope in Figure 6, the data also indicate that the carrier concentration decreases toward the junction. The carrier concentration in our films is an order of magnitude higher as compared to high efficiency devices reported in the literature.29 This translates into a space charge region width (at zero bias) of ∼150 nm for both of the CISSe and CIGSSe devices, as compared to ∼500 nm for high efficiency CIGSe devices. High carrier concentration is likely rooted in the precursor sources used in the syntheses of the nanocrystals and the techniques used in the fabrication of the devices. Detail analysis on the concentration of foreign impurities and their roles in the optical and electrical properties of the absorber film are underway for further improvements in device performance. 3064

Supporting Information Available: Experimental details, TEM image of CuInS2 nanocrystals, UV-vis absorption spectrum of various Cu(In1-xGax)S2 nanocrystals, and power conversion efficiency contour mapping for CISSe solar cells. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) Hillhouse, H. W.; Beard, M. C. Curr. Opin. Colloid Interface Sci. 2009 (in press). (2) Gur, I.; Fromer, N. A.; Geier, M. L.; Alivisatos, A. P. Science 2005, 310 (5747), 462–465. (3) Wu, Y.; Wadia, C.; Ma, W. L.; Sadtler, B.; Alivisatos, A. P. Nano Lett. 2008, 8 (8), 2551–2555. (4) Luther, J. M.; Law, M.; Beard, M. C.; Song, Q.; Reese, M. O.; Ellingson, R. J.; Nozik, A. J. Nano Lett. 2008, 8 (10), 3488–3492. (5) Ma, W.; Luther, J. M.; Zheng, H. M.; Wu, Y.; Alivisatos, A. P. Nano Lett. 2009, 9 (4), 1699–1703. (6) Guo, Q.; Kim, S. J.; Kar, M.; Shafarman, W. M.; Birkmire, R. W.; Stach, E. A.; Agrawal, R.; Hillhouse, H. W. Nano Lett. 2008, 8 (9), 2982–2987. (7) Panthani, M. G.; Akhavan, V.; Goodfellow, B.; Schmidtke, J. P.; Dunn, L.; Dodabalapur, A.; Barbara, P. F.; Korgel, B. A. J. Am. Chem. Soc. 2008, 130 (49), 16770–16777. (8) Ramanathan, K.; Contreras, M. A.; Perkins, C. L.; Asher, S.; Hasoon, F. S.; Keane, J.; Young, D.; Romero, M.; Metzger, W.; Noufi, R.; Ward, J.; Duda, A. Prog. PhotoVoltaics 2003, 11 (4), 225–230. (9) Siebentritt, S. Thin Solid Films 2002, (403-404), 1–8. (10) Stanbery, B. J. Crit. ReV. Solid State Mater. Sci. 2002, 27 (2), 73– 117. (11) Adurodija, F. O.; Song, J.; Kim, S. D.; Kim, S. K.; Yoon, K. H. Jpn. J. Appl. Phys., Part 1 1998, 37 (8), 4248–4253. (12) Eberspacher, C.; Fredric, C.; Pauls, K.; Serra, J. Thin Solid Films 2001, 387 (1-2), 18–22. (13) Ginley, D. S.; Curtis, C. J.; Ribelin, R.; Alleman, J. L.; Mason, A.; Jones, K. M.; Matson, R. J.; Khaselev, O.; Schulz, D. L., Nanoparticle precursors for electronic materials. In Microcrystalline and Nanocrystalline Semiconductorss1998; Materials Research Society: Warrendale, PA, 1999; Vol. 536, pp 237-244. (14) Kaelin, M.; Rudmann, D.; Kurdesau, F.; Meyer, T.; Zogg, H.; Tiwari, A. N. Thin Solid Films 2003, 431, 58–62. (15) Kapur, V. K.; Bansal, A.; Le, P.; Asensio, O. I. Thin Solid Films 2003, 431, 53–57. (16) Sheppard, C. J.; Alberts, V. J. Phys. D: Appl. Phys. 2006, 39 (17), 3760–3763. (17) Castro, S. L.; Bailey, S. G.; Raffaelle, R. P.; Banger, K. K.; Hepp, A. F. J. Phys. Chem. B 2004, 108 (33), 12429–12435. (18) Choi, S. H.; Kim, E. G.; Hyeon, T. J. Am. Chem. Soc. 2006, 128 (8), 2520–2521. (19) Du, W. M.; Qian, X. F.; Yin, J.; Gong, Q. Chem.sEur. J. 2007, 13, 8840–8846. (20) Gardner, J. S.; Shurdha, E.; Wang, C. M.; Lau, L. D.; Rodriguez, R. G.; Pak, J. J. J. Nanopart. Res. 2008, 10 (4), 633–641. (21) Jiang, Y.; Wu, Y.; Mo, X.; Yu, W. C.; Xie, Y.; Qian, Y. T. Inorg. Chem. 2000, 39 (14), 2964. (22) Nairn, J. J.; Shapiro, P. J.; Twamley, B.; Pounds, T.; von Wandruszka, Nano Lett., Vol. 9, No. 8, 2009

(23)

(24) (25) (26)

R.; Fletcher, T. R.; Williams, M.; Wang, C. M.; Norton, M. G. Nano Lett. 2006, 6 (6), 1218–1223. Guo, Q.; Agrawal, R.; Hillhouse, H. Rapid synthesis of ternary, binary and multinary chalcogenide nanoparticles. International Patent Applicaion No. PCT/US2007/069349. Mills, K., Thermodynamic Data for Inorganic Sulphides, Selenides and Tellurides. Butterworths: London, 1974. Cahen, D.; Noufi, R. J. Phys. Chem. Solids 1992, 53 (8), 991–1005. Shrotriya, V.; Li, G.; Yao, Y.; Moriarty, T.; Emery, K.; Yang, Y. AdV. Funct. Mater. 2006, 16 (15), 2016–2023.

Nano Lett., Vol. 9, No. 8, 2009

(27) Mitchell, K. W.; Chesarek, W.; Willett, D. R.; Eberspacher, C.; Ermer, J. H.; Gay, R. R. Sol. Cells 1991, 30 (1-4), 131–136. (28) Mauk, P. H.; Tavakolian, H.; Sites, J. R. IEEE Trans. Electron DeVices 1990, 37 (2), 422–427. (29) Repins, I. L.; Stanbery, B. J.; Young, D. L.; Li, S. S.; Metzger, W. K.; Perkins, C. L.; Shafarman, W. N.; Beck, M. E.; Chen, L.; Kapur, V. K.; Tarrant, D.; Gonzalez, M. D.; Jensen, D. G.; Anderson, T. J.; Wang, X.; Kerr, L. L.; Keyes, B.; Asher, S.; Delahoy, A.; Von Roedern, B. Prog. PhotoVoltaics 2006, 14 (1), 25–43.

NL901538W

3065