Improving Carrier-Transport Properties of CZTS by Mg Incorporation

Research Article www.acsami.org

Cite This: ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Improving Carrier-Transport Properties of CZTS by Mg Incorporation with Spray Pyrolysis Stener Lie,† Shin Woei Leow,† Douglas M. Bishop,‡ Maxim Guc,§ Victor Izquierdo-Roca,§ Oki Gunawan,‡ and Lydia Helena Wong*,†,∥ †

School of Materials Science & Engineering, Nanyang Technological University, 50 Nanyang Avenue, 639798 Singapore IBM T.J. Watson Research Center, 1101 Kitchawan Road, Yorktown Heights, New York 10598, United States § Catalonia Institute for Energy Research (IREC), Jardins de les Dones de Negre, 1, 2a pl., 08930 Sant Adrià de Besòs, Barcelona, Spain ∥ Singapore-HUJ Alliance for Research and Enterprise (SHARE), Nanomaterials for Energy and Energy-Water Nexus (NEW), Campus for Research Excellence and Technological Enterprise (CREATE), 138602 Singapore

Downloaded via UNIV OF SOUTHERN INDIANA on July 20, 2019 at 06:59:10 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

‡

S Supporting Information *

ABSTRACT: High nonradiative recombination, low diffusion length and band tailing are often associated with a large open circuit voltage deficit, which results in low efficiency of Cu2ZnSnS4 (CZTS) solar cells. Recently, cation substitution in CZTS has gained interest as a plausible solution to suppress these issues. However, the common substitutes, Ag and Cd, are not ideal due to their scarcity and toxicity. Other transition-metal candidates (e.g., Mn, Fe, Co, or Ni) are multivalent, which may form harmful deep-level defects. Magnesium, as one of the viable substitutes, does not have these issues, as it is very stable in +2 oxidation state, abundant, and nontoxic. In this study, we investigate the effect of Mg incorporation in sulfur-based Cu2ZnSnS4 to form Cu2MgxZn1−xSnS4 by varying x from 0.0 to 1.0. These films were fabricated by chemical spray pyrolysis and the subsequent sulfurization process. At a high Mg content, it is found that Mg does not replace Zn to form a quaternary compound, which leads to the appearance of the secondary phases in the sample. However, a low Mg content (Cu2Mg0.05Zn0.95SnS4) improves the power conversion efficiency from 5.10% (CZTS) to 6.73%. The improvement is correlated to the better carrier-transport properties, as shown by a lesser amount of the ZnS secondary phase, higher carrier mobility, and shallower acceptor defects level. In addition, the Cu2Mg0.05Zn0.95SnS4 device also shows better charge-collection property based on the higher fill factor and quantum efficiency despite having lower depletion width. Therefore, we believe that the addition of a small amount of Mg is another viable route to improve the performance of the CZTS solar cell. KEYWORDS: thin-film solar cells, kesterite, cation substitution, spray pyrolysis, CZTS potential fluctuation.10,13,14 Recent studies suggested that Cu or Zn substitution by other elements could suppress these defects.15 However, the common substitutes, Ag and Cd,16−18 are rare or toxic, which digress from the benefit of CZTS. Other transition-metal candidates (e.g., Mn, Fe, Ni, or Co) are multivalent, which may form harmful deep-level defects.19,20 Magnesium is a viable candidate, which does not have these issues, as it is very stable in +2 oxidation states, abundant, and nontoxic. Another advantage of Mg is that Zn-based secondary phases such as ZnSe and ZnS are well known to be present in CZTSSe, and their presence leads to a reduction in photovoltaic performance.21 However, MgS and MgSe are

1. INTRODUCTION Cu2ZnSnS4 (CZTS) has been one of the promising candidates for low-cost and environmental friendly thin-film solar cells owing to its large absorption coefficient (∼104 cm−12) and direct band gap energy of 1.4−1.6 eV.3 The highest recorded efficiency is achieved at 12.6%,4 which is still considerably lower than the theoretical Shockley−Queisser limit of 32.4% for CZTS5 or Cu(In,Ga)Se2 solar cells at 22.6%.6 The main issue for CZTS solar cells is the open circuit voltage (Voc) deficit (Eg/q − Voc),7−9 where Eg is the bandgap and q is electron’s charge. Several reasons have been proposed, such as severe band tail effects,10 nonoptimal quality of the kesterite absorber, the presence of secondary phases,11 and a nonideal interface between CdS and CZTS (“cliff”-like).12 Particularly, the CuZn antisite defect issue is proposed as one of the main reasons for the Voc deficit due to the creation of electrostatic © XXXX American Chemical Society

Received: March 24, 2019 Accepted: June 28, 2019 Published: June 28, 2019 A

DOI: 10.1021/acsami.9b05244 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces Table 1. Elemental Composition of Cu2MgxZn1−xSnS4 Thin Films composition (atom %) x 0.0 0.05 0.10 0.20 0.40 0.60 0.80 1.0

Cu 18.51 18.64 18.53 20.52 21.46 22.78 21.17 22.59

± ± ± ± ± ± ± ±

0.14 0.69 0.45 0.18 0.52 0.79 1.37 1.49

Mg

Zn

0 1.04 ± 0.11 1.72 ± 0.1 3.73 ± 0.48 9.62 ± 0.46 13.52 ± 1.78 15.15 ± 1.21 19.12 ± 0.96

12.78 ± 0.17 11.30 ± 0.20 11.39 ± 0.37 10.68 ± 0.41 9.91 ± 1.13 5.94 ± 1.39 2.64 ± 0.33 0

Sn 11.53 11.04 10.99 11.16 15.12 15.15 14.47 15.31

S

± 0.32 ± 0.23 ± 0.55 ± 0.66 ± 0.42 ± 2.58 ±2.04 ± 1.07

57.19 58.00 57.38 53.91 43.8 42.6 47.24 42.98

± ± ± ± ± ± ± ±

Cu/(Zn + Mg + Sn) 0.49 0.97 0.92 0.58 0.3 2.37 1.50 1.35

0.76 0.80 0.77 0.82 0.62 0.66 0.66 0.66

± ± ± ± ± ± ± ±

0.01 0.02 0.02 0.03 0.03 0.09 0.10 0.07

(Mg + Zn)/Sn 1.11 1.12 1.19 1.29 1.28 1.29 1.23 1.25

± ± ± ± ± ± ± ±

0.03 0.03 0.07 0.08 0.07 0.21 0.14 0.08

Mg/(Mg + Zn) 0 0.08 0.13 0.26 0.49 0.69 0.85 1

± ± ± ± ± ±

0.01 0.01 0.03 0.04 0.07 0.02

ment is supported by the positive finding in photoluminescence (PL) and capacitance−voltage (CV) measurements. From these results, we found that incorporation of magnesium in a small amount helps to improve the absorber quality of CZTS such as higher carrier mobility, shallower acceptor defect level, less amount of ZnS, and better charge collection.

unstable in the solution condition;22 therefore, the existence of secondary phases should be reduced. Moreover, the content of Mg in the earth’s crust is naturally higher than the Zn content.23 Several studies on Mg incorporation into Cu2ZnSn(S,Se)4 have been reported.22,24−28 Partial substitution of Cu with Mg in Cu2ZnSnSe4 results in a significant increase in hole mobility (36.5−120 cm2 V−1 s−1) and reduction in carrier concentration due to donor defect properties of Mg2+ when substituting Cu+.25 Partial substitution of Mg with Zn in Cu2ZnSn(S,Se)426,27 has been shown to improve the grain sizes and electrical properties of the films at a small amount of Mg substitution. None of these groups reported a photovoltaic device except for Caballero et al., who recently reported the device improvement for Cu2Zn1−xMgxSn(S,Se)4 at x = 0.04. The improvement is attributed to reduced structural defects based on Raman measurements although phase separation was observed at a high Mg content. On the other hand, Mg incorporation in Cu2ZnSnS4 is still less explored. From a theoretical perspective, there have been conflicting results on the formation of pure Cu2MgSnS4; one study predicted complete phase separation,29 whereas another study predicted the formation of a stable phase.15 As for the experimental studies, a few groups reported the formation of pure Cu2MgSnS4 nanoparticles and thin films by solution methods with band gap ranging from 1.63 to 1.76 eV.22,24 Partial substitution of sulfur-based Cu2ZnSnS4 by the pulse laser deposition technique shows the possibility of forming p-type Cu2MgxZn1−xSnS4 thin films with bandgap ranging from 1.3 to 1.5 eV.28 To the best of the authors’ knowledge, the effect of Mg incorporation in Cu2ZnSnS4 photovoltaic devices has not been reported before. In this study, we investigated the effect of Mg incorporation in CZTS using a facile spray pyrolysis method. Spray pyrolysis is used due to its promising methods in terms of procedure, operation, cost, and scalability. Many studies in spray pyrolysisbased CZTS have been done to understand how to avoid the formation of secondary phases and control the desired stoichiometry ratio and morphology.30−34 CZTSSe deposited by spray pyrolysis was reported to exhibit 5−8% efficiency, which is comparable to other solution methods.20,30,35−38 Following the film formation, the morphology, crystal structure, and phases of the thin films were investigated and correlated with photovoltaic device characteristics. A combination of multiwavelength Raman and X-ray diffraction (XRD) analyses reveals the crystal quality and phase formation in thin films with different Mg contents. Optimal device performance is achieved for films with a small amount of Mg (Cu2Mg0.05Zn0.95SnS4), whereas large Mg amount was found to be detrimental for the device performance. This improve-

2. EXPERIMENTAL METHOD 2.1. Cu2MgxZn1−xSnS4 Thin-Film Preparation. As a clarification, the term “Cu2MgxZn1−xSnS4” or “Cu2Mg0.05Zn0.95SnS4” in this study refers to the as-synthesized stochiometric ratio based on the precursor solution and does not reflect the composition and phases in the film. The precursor solution consists of copper chloride dihydrates (CuCl2·2H2O, 0.06 M), magnesium chloride hexahydrate/zinc chloride (MgCl2·6H2O/ZnCl2·2H2O, 0.044 M), tin(II) chloride dihydrates (SnCl2·2H2O, 0.036 M), and thiourea (CH4N2S, 0.64 M) into 50 ml of deionized water. These solutes were obtained from Sigma-Aldrich (99.99% purity). A Cu-poor (Cu/(Zn + Sn) = 0.75) and MgxZn1−x-rich (MgxZn1−x/Sn = 1.2) composition is used, as it exhibits optimum performance for CZTS based on this method.20,30,36 The amount of thiourea is higher than the required amount to compensate for its loss to heat during spray pyrolysis. Afterward, the solution was sprayed on to the Mo-coated glass on the 450 °C hotplate using N2 as a carrier gas. The as-deposited Cu2MgxZn1−xSnS4 samples were sulfurized at 600 °C for 30 min (ramp rate ∼10 °C/ min) to enhance the crystalline quality of Cu2MgxZn1−xSnS4 thin films. 2.2. Cu2MgxZn1−xSnS4 Device Fabrication. To make the solar cell device, a CdS buffer layer was deposited by chemical bath deposition at 80 °C for 9 min. For the transparent conducting oxide, indium tin oxide (ITO) layers were deposited by the sputtering technique. The devices were manually scribed to form grids with an active area of 0.15 cm2 (total area = 0.16 cm2). Finally, the Ag paste was placed as the top electrode. 2.3. Cu2MgxZn1−xSnS4 Characterizations. X-ray diffraction (XRD) and morphological analyses were carried out by a Bruker D8 Advance and field emission scanning electron microscopy (FESEM) (JEOL, JSM-7600F), respectively. Raman scattering spectra were measured using a Horiba Jobin-Yvon FHR-640 spectrometer coupled with a CCD detector and excited by a solid-state YAG:Nd laser (laser line 532 nm). The measurements were performed in a backscattering configuration through a Olympus objective coupled with a special probe designed in IREC. As for the photoluminescence (PL) setup, a 15 kHz-pulsed 532 nm laser is used as the excitation source at room temperature with a monochromator and an InGaAs photomultiplier as the tube detector. The exfoliating technique from a previous study39 was performed on the thin films as preparation for the Hall measurement in the recently developed rotating parallel dipole line AC Hall by IBM.40,41 Regarding the device characterization, current density−voltage (J−V) data were obtained using an in-house Xe-based solar simulator, which is calibrated with a Si reference cell. External quantum efficiency (EQE) was measured by PVE300 from Bentham. IBM-PVX system and HP 4192 impedance B


Research Article

ACS Applied Materials & Interfaces analyzer were used to obtain and analyze the data for both current− voltage−temperature and capacitance−voltage (CV) measurements, respectively.

addition, the ease of secondary phase formation at the high Mg content may contribute to an increase in the thickness.42 The largest grain is observed at x = 0.05 in conjunction with the best device performance due to the relation between fewer grain boundaries and better charge transport.43 At x ≤ 0.2, the grains are larger, denser, and more compact in comparison with smaller grains and layered structure, which are observed at x ≥ 0.4. In addition, voids are also apparent in these films, especially at x ≥ 0.8. Similar observations with decreasing grain size with Mg content are clearly shown in the plan-view SEM images (Figure S2). To present a better comparison among the grain sizes in these thin films, quantitative analysis on the grain size distribution was also performed. Figure S3 shows the grain size distribution based on the area and diameter. Figure 2 shows the diffraction patterns for Cu2MgxZn1−xSnS4 films at x = 0.0−1.0. Peaks at 2θ = 40.5°

3. RESULTS AND DISCUSSION The composition of these samples from the surface energy dispersive spectroscopy (EDS) measurement is shown in Table 1. To provide a better statistical representation of the composition, several batches of samples and measurement on different regions within the sample were performed. The values in the table represent the mean results with their standard deviation. The maximum elemental standard deviation of ≈2% absolute generally supports the conclusion reached by other characterization techniques that the films produced by spray pyrolysis are broadly homogenous. These results also justify the incorporation of Mg into the films after sulfurization, as shown by the Mg/(Mg + Zn) ratio. A notable difference occurs at x ≥ 0.4, which shows a very Cu-poor ratio and a less sulfur content compared to the other ratios. In addition, EDS analysis was also performed across the thickness of these films to understand the lateral elemental distribution (see Figure S1). It is found that all of the elements are distributed evenly throughout the thickness of the films. Figure 1 exhibits the cross-sectional FESEM images of the Cu2MgxZn1−xSnS4 thin films with different Mg concentrations.

Figure 2. X-ray diffraction pattern of the Cu2MgxZn1−xSnS4 thin film (x = 0−1.0).

are related to the molybdenum (Mo) signature peak. For CZTS (x = 0), the dominant peaks are consistent with diffraction peaks of the kesterite CZTS crystal structure (JPCDS No. 26-0575).44 From x = 0.0 up to x = 0.6, there is a decrease in intensity of the reflexes at 23.1°, which is identified as the (110) plane of the kesterite CZTS. A clear increase of the full width at half-maximum (FWHM) of this peak is observed, suggesting a degradation of the crystalline quality of the kesterite CZTS phase.43 This occurrence is consistent with the smaller grain size observed in Cu2MgxZn1−xSnS4 at x ≥ 0.4 via SEM. Furthermore, at x = 0.4−1.0, there is an increase of intensity of reflexes at 16.0, 17.6, 10.8, and 32.1°, which suggest the presence of the Cu2SnS3 monoclinic phase.45 In addition, the peak at 29.2° might be related to the presence of texturized SnS phase. For all compositions, the most dominant peak at around 28.5°, which is related to the (112) plane in the CZTS structure, does not show a clear shift. This suggests the independency of the cell lattice parameters of Zn substitution with the Mg or insignificant Mg incorporation in the CZTS structure. The detailed analysis based on the ICDD database on the possible presence of different phases at different Mg ratios is presented in Figure S4. It is possible that MgS and MgO are present at a higher Mg content, as shown by the small peaks at 34 and 43°. However, a strong overlap of different possible phases from Cu2(Zn,Mg)SnS4, Cu−Sn−S ternaries, and sulfide binaries in the XRD diffractograms does not allow their strict identification, therefore Raman scattering analysis was performed to distinguish these phases. Raman scattering spectra are measured on a bare absorber surface in, at least, four points for each sample. A Raman spectrum for each composition is presented in Figure 3a. For the sample with x = 0, a typical spectrum of the kesterite CZTS

Figure 1. FESEM cross-sectional images of Cu2MgxZn1−xSnS4 thin films with (a) x = 0, (b) x = 0.05, (c) x = 0.1, (d) x = 0.2, (e) x = 0.4, (f) x = 0.6, (g) x = 0.8, and (h) x = 1.0.

At x ≤ 0.2, the average thickness ranges from 700 to 1000 nm. As the amount of magnesium increases (x ≥ 0.4), the thickness of the films increases significantly (>2000 nm), which is caused by the evident degradation of the film compactness and increase in porosity, which leads to voids in between grains. In C


Research Article

ACS Applied Materials & Interfaces

Figure 3. (a) Raman scattering spectra of Cu2MgxZn1−xSnS4 thin films. At the bottom, the fingerprint spectra of Cu2SnS3,45 Cu3SnS4,46 and MoS247 are presented; (b) distribution of the phases in Cu2MgxZn1−xSnS4 thin films according to the analysis of the Raman scattering spectra.

compound is obtained,48,49 which denotes a good crystalline quality of the samples and low full width at half maximum (FWHM), in agreement with FESEM images. With the increasing Mg content, no significant changes are observed in the spectra up to x = 0.2. Fitting of the most intense Raman peak at 338 cm−1 with the Lorentzian curve shows an insignificant change of the peak position and FWHM (Figure S5). When Zn is substituted with a lighter cation (Mg), this peak is expected to shift to a higher wavenumber.50,51 Additionally, the creation of a solid solution of quaternary compounds leads to a noticeable increase of the FWHM at concentrations ∼10% of the substituted cation,50,51 which is not the case in the Cu2MgxZn1−xSnS4 system up to x = 0.2 (the dark yellow spheres in Figure S5). These findings suggest that only a small amount of Mg may be incorporated in the Cu2MgxZn1−xSnS4 solid solution for x up to 0.2. A further increase of Mg content (x ≥ 0.4) leads to a strong change in the Raman spectra of Cu2MgxZn1−xSnS4 thin films. The most intense peak, which is related to the A symmetry peak in the kesterite-type compound,48 exhibited a red shift, followed by a blue shift and accompanied with a significant increase in FHWM, and final disappearance at x = 1.0. This was observed together with the appearance of new Raman peaks, which start to be detected at x = 0.4. These new peaks are related to two ternary phases, a monoclinic polymorph of Cu2SnS3,45 an orthorhombic polymorph of Cu3SnS4,46 and to the MoS247 compound. These ternary phases may be formed because of the reduced amount of Zn for samples with x ≥ 0.4, and incapability of Mg to form the quaternary compound results in excessive Cu, Sn, and S atoms. In contrast, the MoS2 phase is not expected because of the low penetration depth of the 532 nm excitation wavelength (∼50 nm52) in comparison with the film thickness. A possible explanation could be the presence of holes, which is observed in Figure S2, allows the detection of the Raman signal from the MoS2 compound at the CMgZTS/Mo interface.53 It should be noted that in a Zn-free sample, only the peaks of ternaries and MoS2 phases were detected and hence confirming that the Cu 2 MgSnS 4

compound is not formed in the present study. This is in line with the previous results published in ref 27, where a mixture of ternary and binary phases is found for the Cu2MgxZn1−xSn(S,Se)4 thin films with a high Mg content and with the theoretical predictions of complete phase separation in the case of a pure Mg-based quaternary compound.29 It should be noted that quaternary Cu2MgSnS428 and Cu2MgSnSe454 compounds were reportedly formed. However, a detailed analysis of their data could also be interpreted as a mixture of a ternary phase with a Mg-based binary instead of the quaternary phase proposed by the authors.28 Furthermore, no sufficient evidence of the quaternary phase formation is presented in ref 54, except for XRD patterns, where ternary and quaternary phases are hardly resolved. To further investigate the secondary phase formation in Cu2MgxZn1−xSnS4 thin films, a complementary measurement of Raman spectra under 785 and 325 nm excitation wavelengths was performed (see Figure S6). The use of a 785 nm excitation laser line shows additional ternary phases (Cu2Sn3S7 and Cu4SnS445) in the thin films with a high Mg content. Highly intense peaks of the ZnS secondary phase were detected using the 325 nm excitation wavelength in all samples with x ≤ 0.2. The relative distribution of the phases as a function of the Mg content in Cu2MgxZn1−xSnS4 thin films is shown in Figure 3b. In the top figure, a relative amount of Cu2SnS3 and Cu3SnS4 is presented, as measured under 532 nm excitation wavelength, whereas the bottom figure exhibits the absolute intensity of the LO peak of the ZnS phase, as measured under 325 nm excitation wavelength. These observations are in agreement with XRD analysis. However, we do not detect the MgS and MgO binary phases obtained from XRD analysis (see Figure S4). As these are wide-bandgap and transparent semiconductors (for MgS Eg > 4.5 eV,55 for MgO Eg = 7.8 eV55), they are hardly detectable even with a 325 nm excitation laser line. Additionally, there is a possibility that the excess of Mg forms the segregation of metallic Mg in the interstitial positions, in the grain boundaries, or at the back of the absorber layer. It is also possible that the presence of Mg D


Research Article


Figure 4. (a) Carrier concentration (black) and hole mobility (blue) for Cu2MgxZn1−xSnS4 as a function of Mg/(Mg + Zn); (b) photoluminescence (PL) spectra (normalized) of Cu2MgxZn1−xSnS4 at x = 0−0.2.

out the magnesium effect in the CZTS device. Figure 5 shows the device performance parameters including power con-

results in more Zn atoms occupying Cu sites and/or Mg replacing Cu. However, a careful neutron diffraction study is needed to investigate the real structure of Mg-substituted CZTS. F ig u re 4 a s hows the e lectrical properties o f Cu2MgxZn1−xSnS4, as a function of Mg/(Mg + Zn). The carrier concentration and hole mobility are obtained by rotating the parallel dipole line magnet AC Hall setup.41,56 The values for CZTS are comparable with reported findings on high-efficiency CZTS using the same technique.57,58 For all magnesium concentrations, p-type conductivity is observed. At x = 0.0−0.2, the carrier concentration is around 1015 cm−3, whereas an increase in the carrier concentration is observed when x ≥ 0.4. As for the hole mobility, there is a small increase at x = 0.05, which is related to the carrier-transport improvement. The sudden increase at x = 0.4 is consistent with the significant changes observed in SEM, XRD, EDS, and Raman analyses. Furthermore, at x ≥ 0.4, there is an indication of a high number of defects or impurities based on the high carrier concentration and small hole mobility introduced by the higher amount of magnesium. Figure 4b shows the photoluminescence response of the samples. To increase the PL yield, the samples were airannealed in a dry box with low humidity for 10 min at 280 °C and then slowly cooled to room temperature. This method has also been shown to reduce the cation disorder in the CZTS absorber and beneficial to improve the open circuit voltage.59 The PL signal can only be detected for samples with x = 0.0− 0.2, whereas the other ratios do not yield any response. The PL signals are quenched at x ≥ 0.4, which suggest nonradiative recombination or lower carrier lifetime problem.60 This phenomenon also correlates with a high carrier concentration and the appearance of ternary phases, which are observed in AC Hall and Raman measurements, respectively. Another metric we can observe to assess the bulk absorber quality and depth of acceptor defects is the difference between the PL peak position and band gap (ΔE).61 The band gap is taken from the peak of d(EQE)/d(λ) from Figure S7. From Figure 4b, a slight reduction in the ΔE, i.e., 0.21−0.18 eV, is observed at x = 0.05. This phenomenon is usually related to the improvement in the absorber quality and possible reduction in the depth of acceptor defects.16 Solar cell devices (Mo/Cu2MgxZn1−xSnS4/CdS/i-ZnO/ ITO/Ag paste) were fabricated and measured to further figure

Figure 5. Device parameters (η, Jsc, Voc, and FF) as a function of Mg/ (Mg + Zn) for Cu2MgxZn1−xSnS4 thin-film solar cells.

version efficiency (η), fill factor (FF), short circuit current (Jsc), and open circuit voltage (Voc) as a function of the Mg content. The results are average of 32 cells to provide a statistical consistency. The best-performing cell in this study is achieved at x = 0.05 (Cu2Mg0.05Zn0.95SnS4). In comparison with the pure CZTS, there is an increase in average efficiency by 1%, which is accompanied by the increase in the rest of the parameters. The optimal Jsc and FF in this study are achieved at this composition, whereas the optimal Voc is achieved at x = 0.1. This result indicates improvement of the absorber bulk properties, which is observed in the optoelectronic measurement with incorporation of a small amount of Mg, with the assumption that other variables such as process parameters and other layers are unaffecting. An increase the Mg content leads to a notable decrease in the ZnS phase, which results in the overall improvement of the solar cell performance. On the other hand, Mg may also segregate as binary phases or metallic Mg, reduce the compactness of the absorber layer, and, thus, reduce the Jsc and FF. Meanwhile, at x ≥ 0.4, all device parameters abruptly decrease due to the appearance of the E


Research Article


Figure 6. (a) J−V curves for the highest CZTS and CMg0.05Z0.95TS; (b) spatial charge density vs depletion width from CV measurement for CZTS and CMg0.05Z0.95TS. We use the relative dielectric constant εr = 6.7.1 The dashed vertical line indicates the zero-bias voltage.

Table 2. Device Parameters That Are Extracted from the Diode Model for the Highest CZTS and CMg0.05Zn0.95TS devices CZTS Cu2Mg0.05Zn0.95SnS4

efficiency (%) 5.10 6.73

Jsc (mA/cm2) 17.35 17.19

Voc (mV)

fill factor (%)

613 665

47.97 59.37

n1 2.91 2.6

J01 (A/cm2) −7

7.61 × 10 1.81 × 10−7

Rs (Ω cm2)

Gsh (mS/cm2)

5.20 4.90

7.18 2.95

magnesium improves the absorber quality of the sample. Furthermore, as Voc increases by ∼50 mV despite the small reduction in bandgap (∼15 mV difference), a small amount of Mg may also contribute to reducing the Voc deficit.19 Capacitance−voltage (CV) measurements were carried out on the highest-performing and control samples with a voltage bias of +0.5 to −4 V. Figure 6b shows the space-charge density vs depletion width for both samples from the CV sweep based on the relation in the previous literature.65 The depletion width at zero bias for Cu2Mg0.05Zn0.95SnS4 is lower than that of CZTS (i.e., ≈106 and 170 nm, respectively). Depletion width corresponds with the charge separation quality between the electron−hole pairs. However, in this study, despite having a smaller depletion width, Cu2Mg0.05Zn0.95SnS4 performs better than CZTS. This suggests that the collection depth is better due to a better diffusion length of the carriers.4 A better diffusion length is also related to a Cu2Mg0.05Zn0.95SnS4 higher Hall mobility. Even though the minority carrier diffusion length is directly affected by minority carrier mobility, the high majority carrier mobility may also indirectly translate into high minority carrier mobility.66 The charge density value is also consistent with the carrier concentration value from the Hall measurement. In addition, both samples show a similar carrier density profile in a higher reverse bias condition, which suggests that despite having a similar charge density, a small amount of Mg improves the film quality intrinsically. The Voc vs temperature (T) and Jsc vs temperature (T) plots of the champion Cu2Mg0.05Zn0.95SnS4 are shown in Figure S8. From the plot, the activation energy (EA) is extracted for both CMg0.05Zn0.95TS (≈1.55 eV) and CZTS (≈1.5 eV). The activation energy for Cu2Mg0.05Zn0.95SnS4 is closer to the bandgap in comparison with CZTS, which suggests that the dominant recombination mechanism is closer to space-charge recombination (EA ≈ Eg),67 whereas for CZTS, other limiting factors such as interface recombination or band tailing may be more dominant.68 The improvement for Cu2Mg0.05Zn0.95SnS4 indicates a similar trend to that observed in alkali-metal doping studies (such as Na or K).69−71 Enhancement in grain size, electronic

ternary secondary phases and a high increase in the carrier concentration. Such a large increase in hole carrier density may lead to a very narrow depletion width and very short lifetime which result in a lower minority carrier diffusion length. Finally, the small photoconversion efficiency in the Zn-free (x = 1.0) device is most probably due to the photovoltaic effect of the Cu2SnS3 compound.45,62 EQE measurements conducted on these devices (see Figure S5) reveal the highest charge collection in Cu2Mg0.05Zn0.95SnS4, which is much improved from CZTS (x = 0). At x > 0.05, the EQE shows very small charge collection. This is consistent with the Hall study that indicates very high carrier density which result in very narrow depletion width and low collection efficiency. These EQE results also clarify the abrupt decrease in Jsc for x > 0.2. In addition, the influence of magnesium incorporation on the bandgap is also calculated. The calculation is only done for dEQE/dE curves at x ≤ 0.2 because of its inaccuracy for low EQE data (x ≥ 0.2). Minimal bandgap changes (1.54 eV ± 0.01) were observed at x = 0.0−0.2. This is commonly found in group-II cation substitution for the CZTS system, and theoretical calculation justifies this finding because the group-II cation does not affect both conduction and valence band formation.29 As for x = 1.0, there is a low charge collection which is due to the presence of low-band-gap Cu2SnS3.62 Next, we also compared the “champion” Cu2Mg0.05Zn0.95SnS4 device with the control-CZTS device as shown in Figure 6a with their diode model device parameters in Table 2. In addition to the four common solar cell parameters, extra diode model parameters such as ideality factor n1, reverse saturation current J01, series resistance (Rs), and shunt conductance (Gsh) are extracted using the Sites fitting method on the light IV curve and dark IV curve.63 The champion device’s fill factor (FF) of ≈59% is still relatively low, however it is comparable with the high-efficiency CZTS based on spray pyrolysis from other previous studies.30,35,36 The high diode ideality factor of >2 suggest a coupled defectlevel recombination mechanism.64 The n1, J01, and Rs are lower in comparison with the control samples, which indicates that F


Research Article

ACS Applied Materials & Interfaces Notes

properties improvement, and enhanced photovoltaic performance are observed from those studies. In addition, the small amount (

Improving Carrier-Transport Properties of CZTS by Mg Incorporation

Recommend Documents