Subscriber access provided by University of Winnipeg Library
Article
Complexation Chemistry in DMF-Based Molecular Inks for Chalcogenide Semiconductors and Photovoltaic Devices James A. Clark, Anna Murray, Jung-min Lee, Tom S. Autrey, Andrew D Collord, and Hugh W. Hillhouse J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.8b09966 • Publication Date (Web): 09 Dec 2018 Downloaded from http://pubs.acs.org on December 10, 2018
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 12 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Journal of the American Chemical Society
Complexation Chemistry in DMF-Based Molecular Inks for Chalcogenide Semiconductors and Photovoltaic Devices James A. Clark,† Anna Murray,† Jung-min Lee,† Tom S. Autrey,‡ Andrew D. Collord,† and Hugh W. Hillhouse† † Department of Chemical Engineering, Clean Energy Institute, Molecular Engineering & Sciences Institute, University of Washington, Seattle, Washington 98195-1750 ‡ Pacific Northwest National Laboratory, 902 Battelle Boulevard, P.O. Box 999, Richland, Washington 99352 United States Supporting Information Placeholder ABSTRACT: Molecular inks based on dimethyl sulfoxide, thiourea, and metal salts have been used to form high optoelectronic quality semiconductors and have led to high power conversion efficiencies for solution-processed photovoltaic devices for Cu2ZnSn(S,Se)4 (CZTS), Cu2Zn(Ge,Sn)(S,Se)4 (CZGTS), CuIn(S,Se)2 (CIS) and Cu(In,Ga)(S,Se)2 (CIGS). However, several metal species of interest including Ag(I), In(III), Ge(II), and Ge(IV) have either low solubility (requiring dilute inks) or lead to precipitation or gelation. Here, we demonstrate that the combination of N,N-dimethylformamide (DMF) and thiourea (TU) has the remarkable ability to form intermediatestability acid-base complexes with a wide number of metal chloride Lewis acids (CuCl, AgCl, ZnCl2, InCl3, GaCl3, SnCl4, GeCl4, and SeCl4) to form high-concentration stable molecular inks. Using calorimetry, Raman spectroscopy, and solubility experiments, we reveal the important role of chloride transfer and thiourea to stabilize metal cations in DMF. Methylation of thiourea is used to vary the strength of the Lewis basicity and demonstrate that the strength of the thiourea-metal chloride complex formed after DMF evaporation is critical to prevent volatilization of metal containing species. Further, we formulated a sulfur-free molecular ink which was used to deposit crystalline CuInSe2 without selenization that sustains high quasi-Fermi level splitting under constant illumination. Finally, we demonstrate the ability of the DMF-TU molecular ink chemistry to lead to high photovoltaic power conversion efficiencies and high open-circuit voltages for solution-processed CIS and CZGTS with PCE’s of 13.4% and 11.0% and Voc/Voc,SQ of 67% and 63%, respectively.
1. Introduction Rapid deployment of clean energy solutions to mitigate the effects of climate change is among the biggest challenges facing humanity. The abundance of the solar resource makes photovoltaics (PV) an attractive technological solution to this challenge. As of 2017, the worldwide electricity generation from PV reached 1.7% of total electricity generation, and PV manufacturing capacity increased to 94.6 GW/year.1 However, given that global energy demand is currently 20 TW and is expected to increase to rise above 25 TW by 2040,2 disruptively inexpensive PV manufacturing technologies are needed to displace existing fossil fuel technologies and fulfill new demand. One of the biggest barriers to increasing investment in PV is the high capital expenditure (CAPEX) to build PV manufacturing plants.3,4 It has been estimated that fully roll-to-roll processed PV can substantially reduce the required CAPEX and cost of PV manufacturing.5,6 PV technologies utilizing solution-processed chalcogenide-based absorber layers are attractive because of their high stability and power conversion efficiency. A variety of chalcogenide solution processes have been developed including: nanocrystal inks,7 electrodeposition,8,9 colloidal suspensions,10,11 and molecular ink processes,12–25 but molecular ink processes have consistently led to devices with the highest power conversion efficiency (PCE). The solvent systems for these molecular ink processes can be divided into four major categories: (1) hydrazine,12,13 (2) polar protic (H2O, CH3OH, methoxyethanol),14,15 (3) amine-thiol,16–18 and (4) polar aprotic (DMSO19–21 and DMF22–25). Our group first developed DMSObased molecular inks (using thiourea (TU) as both complexing ligand and sulfur source) for chalcogenide semiconductor film
formation since the solvent and sulfur source combination: (i) lacks a reactive oxygen site and thus avoids the formation of metal-oxygen-metal bonds (which can be hard to fully exchange with sulfur or selenium), (ii) has high dipole moment and can solubilize many inexpensive metal salt precursors for the main elements and dopants,20,26 (iii) contains three different character Lewis base sites (the O in DMSO along with the S and N in TU) which can help coordinate different size and hardness metal cations in solution, and (iv) is much safer and more environmentally friendly than hydrazine. This chemistry has led to high quality films and high PV device PCEs for Cu2ZnSn(S,Se)4 (CZTS), CuIn(S,Se)2 (CIS) and Cu(In,Ga)(S,Se)2 (CIGS) as reported by several research groups.20,21,27–29 Previously, we reported an 11.8% PCE CZTS device using a Lidoped DMSO-TU based ink.20 More recently, researchers at Empa reported a new PCE record of 12.3% for hydrazine-free CZTS (compared to the hydrazine-based record of 12.6%) using this Li-doped DMSO-TU based ink chemistry.29 Also of note, DMSO-TU based molecular inks for CIS and CIGS have led to device PCEs of 13.1% and 14.7%,21 respectively, which is approaching the hydrazine-based record of 17.3% for CIGS.13 While there are numerous benefits of the DMSO-based molecular ink chemistry for chalcogenide semiconductor thin films, some important metal species have low solubility or low stability in DMSO, including Cu(I), Ag(I), In(III), Ge(II), and Ge(IV) species.21,22 Cu(I) species can be effectively solubilized to form stable inks with DMSO by using thiourea as a ligand.19 However, the same DMSO-TU-based stabilization does not significantly increase the solubility of Ag(I) and In(III) compounds, and forms a gel with Ge(II) and Ge(IV)
ACS Paragon Plus Environment
Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
compounds.22 Ag(I) and Ge(IV) are of interest because alloys of these elements with CZTS have been increasingly studied in attempt to reduce cation site disorder in the kesterite lattice.30–35 To overcome these issues with stability and solubility, we investigated numerous other solvents and ligands and found that N,N-Dimethylformamide (DMF) and TU stabilized these species and allowed for the creation of high concentration stable molecular inks. In addition to the increased stability and concentration (relative to DMSO-based inks), the new DMFbased ink chemistry was used to add all the metal species necessary for CZTS and Cu2Zn(Sn,Ge)(S,Se)4 (CZTGS) in their desired oxidation state, Cu(I), Zn(II), Sn(IV), and Ge(IV), without needing to undergo redox reactions.19,22 A subsequent report focused on CIS used a 50% DMF and 50% isopropanol-based ink to achieve a highly-concentrated stable ink that was used to deposit 1 µm thick CIS films after 4 layers of spin-coated ink (compared to 13 layers from a the more dilute DMSO-TU ink). The resulting PV devices had a PCE of 3.4%.23 One contributing factor to the low PCE is the high concentration of IPA, which may lead to the formation of M-O-R and M-O-M linkages. In a recent report, we used a DMF-based ink (without so-solvents) to deposit compositionally graded precursor films which lead to a new world record for a solution-processed CIS PV device with a 13.8% PCE.25 The combination of DMF and TU contains four different character Lewis base sites (the O and N in DMF along with the S and N in TU) which appear to be unique in their ability to stabilize a wide number of (Lewis acidic) metal chlorides (particularly high valence In(III), Ga(III), Sn(IV), and Ge(IV)) at high concentration. However there is little fundamental understanding of the mechanisms through which these high concentrations are achieved. In this work, we examine the complexes formed in DMF solutions of Cu, Ag, Zn, In, Ga, Sn, and Ge metal chlorides, along with chalcogen sources (thiourea, N-N’-dimethylthiourea, tetramethylthiourea, selenium tetrachloride, and elemental selenium). Using solubility experiments, Raman spectroscopy, and mixing calorimetry to directly measure the enthalpy of complexation, this work reveals new insights about complexes involving these species. Specifically, these data illustrate the ways in which chloride transfer drastically increases the solubility of a number of metal chlorides in DMF and the importance of strong complexes to prevent metal chloride volatilization during annealing before metal chalcogen bond formation. These results allowed us to develop a novel sulfur-free molecular ink composed of Cu, In, and Se chlorides in DMF. In contrast, thiourea, thioacetamide, and thiol based molecular inks initially form a sulfide upon annealing (i.e. Cu2ZnSnS4 or CuInS2) and subsequently require selenization in a second step to form the generally higher optoelectronic quality selenide (i.e. Cu2ZnSnSe4 or CuInSe2).14,15,21 This novel ink formulation is shown to lead to crystalline CuInSe2 without selenization which makes it an excellent candidate as a low-cost solution-based deposition method to directly deposit CISe, CZTSe, and potentially other selenium-based functional chalcogenide materials. Finally, we present PV device data utilizing CZTGS and CIS absorbers deposited from DMF-TU inks, which respectively have the highest Voc and one of the highest PCEs for any solution-processed device using these absorber materials.
2. Experimental 2.1 Materials, Formulation
Solubility
Experiments,
and
Ink
Page 2 of 12
All solubility experiments were performed at room temperature in a N2-filled glovebox with O2 and H2O below 20 ppm. Solutions were mixed and stirred overnight to determine solubility limits for various precursors and complexing agents. All precursors were purchased from Sigma-Aldrich with the following purity specifications: anhydrous DMF (99.98%), CuCl (99.995%), AgCl (99.999%), ZnCl2 (99.999%), InCl3 (99.999%), GaCl3 (99.999%), GeCl4 (99.99%), SnCl4 (99.995%), SeCl4 (unlisted purity, product number: 323527), Se (99.999%), tetraethylammonium chloride (TEAC) (98%), thiourea (TU) (99%), dimethyl-thiourea (DMTU) (99%), and tetramethyl-thiourea TMTU (98%). All precursors were used without further purification with the exception of TU which was twice recrystallized from deionized water. Note SnCl4 and GeCl4 are fuming liquids and require caution when handling. The poly selenium chloride (PSC) CISe ink was formulated from 0.225 M CuCl (66.8 mg), 0.35 M InCl3 (232.2 mg), 1.0 M SeCl4 (662.3 mg), 0.5 M Se (118.4 mg), and 3.0 M urea (360.4 mg) in 3 mL DMF. Elemental ratios were Cu/In=0.64, Se/SeCl4=0.5, and Se/(Cu+In)=2.61. The CIS DMF-TU ink was prepared by sequentially dissolving 4.737 M recrystallized TU (7590 mg), 0.789 M CuCl (1563 mg), and 0.877 M InCl3 (3880 mg) in 20 mL of anhydrous DMF. Elemental ratios were Cu/In=0.9 and S/(Cu+In)=3. The CZGTS ink was mixed via a spray coating gradient method reported previously.22 The nominal composition of this ink is 0.385 M CuCl (153 mg), 0.262 M ZnCl2 (232.2 mg), 0.075 M GeCl4 (64.4 mg), 0.175 M SnCl4 (182.5 mg), and 2.0 M urea (610 mg) in 4 mL DMF. All molarities are based on pure solvent volume.
2.2 Film Deposition and Device Fabrication Before deposition of films, molybdenum-coated soda lime glass (MSLG) substrates were cleaned by sequentially sonication in H2O, acetone, IPA, hexane, and chloroform for 10 minutes each. Substrates were dried then placed in an argon plasma cleaner for 5 minutes before a final sonication in water for 10 minutes. CISe films from the PSC ink were deposited by drop casting 100 µL of ink onto a substrate and annealing for 20 minutes under a petri dish in the glovebox. No further selenization or annealing was performed. The CIS DMF-TU ink was filtered using a 0.45 um PTFE filter and spin coated onto MSLG substrates at 3000 rpm for 1 min. Films were annealed on a hot plate at 270 °C for 90 sec, then allowed to cool. This process was repeated 6 times to form a film ~2200 nm thick, measured with profilometry. The CZGTS films were deposited via a spray coating gradient method reported previously.22 CISe (from the DMF-TU ink) and CZGTS films were separately placed in a graphite box in a tube furnace with ~380 mg of selenium pellets and annealed at 540 °C for 20 min under continuous argon purge, then allowed to cool naturally. Devices were completed via deposition of 40 nm of CdS via chemical bath deposition. Window layers of 50 nm i-ZnO and 200 nm ITO were deposited using RF-sputtering. Ni/Al contacts of 50 and 1000 nm were then deposited using thermal evaporation. In the case of the CISe devices, 110 nm of MgF was added by e-beam evaporation as an anti-reflective coating. Individual devices were manually scribed to an average area of 0.105 cm2.
2.3 Calorimetry Calorimetric measurements were performed utilizing a Setaram C80 Calvet calorimeter. Measurements were conducted in Hastealloy reversal mixing cells with solutions separated by a cup/annulus geometry as shown in Figure 3a. The instrument was operated in isothermal mode and allowed to come to thermal
2 ACS Paragon Plus Environment
Page 3 of 12 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Journal of the American Chemical Society
equilibrium with a reference cell for at least 30 minutes before mixing.
2.4 Film Characterization Scanning electron microscopy (SEM) micrographs were collected with an FEI Sirion XL30 microscope using a 5 kV accelerating voltage. Energy-dispersive X-ray spectroscopy (EDX) data were collected with an Oxford EDX detector using an accelerating voltage of 20 kV. Raman analysis was performed with a Renishaw inVia system equipped with the Leica DMIRBE inverted optical microscope using 514.5 and 785 nm excitation sources. X-ray diffraction (XRD) was performed on the films with a Bruker D8 Discover. The source was a Cu Kα X-ray with an incident wavelength of 1.54059 Å and montel collimating optics to remove Kβ. A round collimator of diameter 0.5 mm was used. The detector was a 2D Pilatus 100k detector with radius 15.4 cm and an opening of 2.6 degrees. Photoluminescence (PL) spectra were collected using a modified Horiba LabRAM HR-800 using a 785 nm laser as an excitation source. The adjustable confocal hole before the monochromator was set at 800 μm and a 10x objective was used. A 150 gr/mm Czerny-Turner monochromator blazed at 1200 nm was used, and the emitted light after was passed to a InGaAs array detector cooled with liquid nitrogen. The system was continually purged with dry air to remove unwanted effects from the water in the produced spectra. Calibration of the PL to absolute photon flux was accomplished via blackbody measurements as reported previously.63
2.5 Device Characterization Current-voltage (JV) and external quantum efficiency (EQE) measurements were both performed in ambient conditions using a Keithley 2400 source-measure unit. JV measurements used a Newport AAA Oriel Sol3A solar simulator emitting 1000 W/m2 at standard conditions (25 °C and AM1.5G), calibrated with a Newport Si reference cell. EQE measurements were performed with a chopped monochromatic light beam and lock-in amplifier, calibrated with silicon and germanium reference diodes.
3. Results and Discussion 3.1 Chloride Complexes N,N-Dimethylformamide (DMF) is a Lewis-basic aprotic solvent which is known to solvate bivalent and trivalent cations, but is generally unable to solvate monovalent cations.36 CuCl has exceptionally low solubility in DMF alone and disproportionates according to 2Cu+ ↔ Cu0 + Cu2+. This disproportionation reaction is well-known, and equilibrium constants (Keq=[Cu0][Cu2+]/[Cu+]2) have been measured for a number of solvents, with water, DMF, DMSO, and acetonitrile being ~107, ~104, ~100, and ~10-21 respectively.37 However, this report also showed that ligands which preferentially bind Cu2+ can increase these equilibrium constants in DMF and other solvents. Analogously, species that stabilize the Cu+ oxidation state are
Figure 1. Solubility studies with various chloride species in DMF. Red X indicates that all solids did not dissolve. (a) CuCl mixed with tetraethylammonium chloride (TEAC) in DMF (with constant [CuCl] = 0.025 M). When TEAC is added to CuCl/DMF mixtures with TEAC:CuCl ≥ 1 the disproportionation is reversed and [CuCl2]- is stabilized in solution. (b) GeCl4 mixed with CuCl in DMF (with constant GeCl4 = 0.25 M). Individually, GeCl4 and CuCl have low solubility in DMF, but chloride transfer from germanium to copper drastically increases the solubility of both species. (c) SeCl4 mixed with Se metal in DMF (with constant SeCl4 = 0.5 M). Se is solvated up to a Se:SeCl4 ratio of 1.5 via equilibrium reactions with SeCl2 and Se2Cl2.
expected to decrease the disproportionation equilibrium constant. Figure 1a shows that the addition of tetraethyl ammonium chloride (TEAC) reverses the CuCl disproportionation reaction in DMF. When TEAC is added to mixtures of DMF and CuCl in ratios of TEAC:CuCl ≥ 1, all Cu0 dissolves and a clear solution forms. Since TEAC is known to almost completely dissociate into TEA+ and Cl- in DMF,38 this stoichiometry suggests that CuCl and Cl- react to form some combination of [CuCl2]-, [Cu2Cl4]2-, or [Cu3Cl6]3-, which are soluble in DMF and shifts the disproportionation reaction to the left. Indeed, Zhao et al. have shown that all three of these species exist in aqueous solutions of CuCl and NaCl and that the polynuclear species become more prominent for higher CuCl concentrations.39 Higher concentration solutions of CuCl for TEAC:CuCl = 1 appear yellow in color (but still stable) as shown in SI Figure 1a. This may indicate higher concentrations of polynuclear (Cu2Cl42-, or Cu3Cl63-) copper complexes. A number of the other metal chloride species can donate chloride to stabilize the Cu+ oxidation state in solution via the formation of anionic Cu species (CuCl + Cl- → [CuCl2]-). As shown in SI figures 1c and 1d, GaCl3 and InCl3 can stabilize CuCl in solution with ratios of GaCl3:CuCl ≥ 0.33 and InCl3:CuCl ≥ 1 respectively. In DMF, GaCl3 and InCl3 have been shown to exist in equilibrium between freely dissolved Cl- and different [MCln](3-n)+ configurations, where n=0,1,2,3,4 for indium and n=0,1,4 for gallium.40 Given that 1:1 CuCl:Cl- is required for the formation of stable CuCl2-, this indicates that each Ga can donate 3 Cl- per metal, while each In can donate only 1 Cl- per metal.
Table 1. Solubilities of chloride species in DMF alone, with excess chloride and with excess thiourea. Starting Species
Solubility in DMF
Effect of Tetraethylammonium Chloride Donates Free Clon Solubility to Stabilize CuCl
Effect of Thiourea on Solubility
CuCl
760 Torr,60 ~550 Torr,60 ~10 Torr,61 and >760 Torr,62 respectively. These data highlight an important role of TU in aprotic molecular inks that was previously unreported: TU is a vital complexing agent which prevents metal chloride evaporation during annealing. Thiourea complexes with Cu, Zn, In, and Sn chlorides all exhibit structures where Lewis-acidic metal interacts with the Lewis-basic sulfur on TU.44,47,48 Some species may not complex with TU in solution, but these complexes form upon ink
6 ACS Paragon Plus Environment
Page 7 of 12 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Journal of the American Chemical Society
Table 2. EDX Summary of Thin Films from CIS and CZTS Inks with Multiple Sulfur Sources Sulfur Source
Target Material
TU DMTU
CuInS2
TMTU Sulfur Source
Target Material
TU DMTU TMTU
Cu2ZnSnS4
Metal Loss
Cu (at%)
Ga (at%)
In (at%)
Cu/III
Ga/III
Cu/In
None
9.54
3.33
8.22
0.83
0.29
1.16
Ga (some In)
14.57
0.11
10.9
1.32
0.01
1.34
Ga and In
7.1
0
0.08
88.75
0.00
88.75
Metal Loss
Cu (at%)
Zn (at%)
Sn (at%)
Cu/(Zn+Sn)
Zn/Sn
Cu/Sn
Some Sn
10.78
8.05
4.09
0.89
1.97
2.64
Sn
17.44
7.9
2.58
1.66
3.06
6.76
Sn and Zn
16.85
0
0
N/A
N/A
N/A
drying. For example, the S in TU does not directly complex with InCl3 in solution as shown by the Raman data above. However, when DMF is driven off of solutions of InCl3 and TU, Raman spectra of the dried film exhibits a substantial shift in the C=S peak at 743 cm-1 as shown in SI Figure 5. This suggests InCl3 complexes with TU via In-S interaction upon ink drying, which inhibits InCl3 vaporization. Since CuCl has been shown to coordinate with TU via interaction with sulfur in DMF, mixing calorimetry experiments with CuCl and the R-TU compounds were used as a model system to assess the metal-sulfur interaction strength for these species. As shown by Figure 3b, TU, DMTU, and TMTU respectively exhibit enthalpy of complex formation of -20.03± 1.1, -12.83± 0.71, and -5.95± 0.33 kJ/mol with CuCl. The decreasing enthalpy of complex formation suggests decreasing interaction strength between the Lewis-acidic metal and the Lewis-basic sulfur, which is consistent with the trend in metal loss shown in Table 2.
3.5 Selenides from Solution Processing without PostSelenization The selenides of CZTS, CIS, and CIGS almost universally lead to higher PCE devices than the sulfide versions of these compounds. This has been attributed to favorable defect chemistry and better band alignment with CdS.64,65 Further, the low bandgap of CuInSe2 (CISe) (1.0 eV) make an excellent candidate for the bottom absorber in tandem PV cell applications.21,25 Our discovery that free Cl- stabilizes CuCl in DMF motivated us to explore alternative molecular inks free from sulfur (TU). As
shown in SI Figure 6a, SeCl4 readily donates chloride to CuCl, which stabilizes it in solution. Inks consisting of CuCl, InCl3, and SeCl4 hereafter referred to as poly-selenium chloride (PSC) CISe inks are also stable; however, inks of very high concentration will gel, as shown SI Figure 6b. Using SeCl4 as a chalcogen source has some key advantages over TU: (1) the direct incorporation of Se into a molecular ink could eliminate the need for selenization, which is one of the biggest barriers to scale-up for chalcogenidemolecular-ink-based solar cells, and (2) SeCl4 is carbon and nitrogen free, which has been shown to limit device performance.66 However, substituting SeCl4 for TU also has some challenges. (1) the chalcogen is in the wrong oxidation state and must be reduced (from Se4+ to Se2-) and (2) the absence of TU as a binding complex will allow metal chlorides to evaporate during annealing. Similar to the DMTU and TMTU inks in the previous section, there is significant InCl3 and SeCl4 loss when no binding agent is present in PSC CISe inks, as shown in SI Figure 7. As a first attempt to overcome the two challenges with PSC CISe ink, urea was added to act as both a complexing and reducing agent. This ink was drop cast onto a substrate, annealed under a petri dish (550 °C for 20 mins), and characterized with powder XRD, SEM, EDX and absolute-intensity photoluminescence (AIPL). As shown in Figure 5a the XRD pattern from this film is an excellent match with CuInSe2 and shows some evidence of CuSe2-x. SEM of the film reveals heterogeneous morphology typical of drop-cast films; however, some regions showed highly faceted morphology as shown in Figure 5b. Further, EDX revealed a roughly constant Cu/In and
Figure 5. Characterization of a drop-cast metal chloride PSC ink annealed at 550°C for 20 min. (a) Powder XRD shows peaks characteristic of CuInSe2 with some β-Cu1.75Se (PDF 04-014-3323). *The peak near 40 2θ is from the molybdenum-coated substrate. (b) SEM of annealed film. The inset shows the average metal ratios measured with EDX for this film (c) Absolute intensity confocal photoluminescence (AIPL) of annealed film (excitation intensity greater than one sun). The PL peak at 1.0 eV is typical of CuInSe2 PL. QFLS was extracted from AIPL by a peak-fitting method we have developed and previously reported.63
7 ACS Paragon Plus Environment
Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 8 of 12
Figure 6. Device characterization of champion CuIn(S,Se)2 (CISe) and Cu2Zn(Ge,Sn)(S,Se)4 (CZGTS) PV devices. Note that the CIS cell had a MgF2 AR coating while the CZGTS device did not. (a) JV of a CISe device. The active area efficiencies of 13.4% are among the highest reported PCEs for solutionprocessed CISSe. (b) JV of a CZTGS device. This is the highest reported Voc/Voc,SQ for any kesterite device. (c,d) The integrated current from the EQE of the CIS (c) and CZTGS (d) devices show excellent matching to the Jsc values.
Se/(Cu+In) across the film. However, these values are Cu-poor and Se-poor when compared to pure CISe. EDX also revealed small amounts of Cl and N remaining in the film (4.1 ± 2.1 atomic % and 3.2 ± 2.0 atomic %, respectively). Given XRD only shows CISe and CuSe2-x, this suggest that an indium rich amorphous phase is present in the film. AIPL of the CISe film was measurable with a characteristic peak position at 1.00 eV as shown in Figure 5c. AIPL was somewhat heterogeneous across the substrate, and regions near corners and edges of the substrate showed the most intense PL, which may be due to increased Naflux near the edges of the substrate. The slight difference between the data and full-peak fit from 0.8-0.9 eV can be attributed to CuSe2-x plasmon resonance,67 and was found to be much more intense with other substrates with higher Cu/In ratios. These data represent the first example of photoluminescent, hydrazine-free solution-processed CISe and may represent a novel ultra-low CAPEX pathway to deposit chalcogenide materials.
3.6 Photovoltaic Devices from DMF Molecular Inks Utilizing the high solubility of CuCl and InCl3 in the DMF-TU molecular ink, allowed for the deposition of a ~2.2 µm CuInS2 precursor films in 6 spin coating cycles, (compared to 13 layers required for a 1.5 µm precursor film for the DMSO-TU CIS ink).21 The champion devices from this absorber layer exhibited an exceptionally high PCE of 13.4% as shown in Figure 6a. The Voc of 512 mV of this device is equivalent to 67% of the Schottky-Queisser (SQ) limit Voc based on the 1.00 eV bandgap extracted from EQE, which is among the highest value reported for any absorber with a 1.00 eV bandgap. The EQE data in Figure 6c reveal collection efficiencies
