*3.1. Centrifugation to Obtain CZTS Nanoparticle Ink*

Figure 4a–e shows TEM images of the CZTS particle distribution of the dispersion subjected to different centrifugation conditions. Figure 4a shows the distribution of CZTS particles for the CZTS dispersion without a centrifugal treatment. The small particles and large particles agglomerated together to form large clusters such that the boundaries between particles became unclear and it was not possible to tell the size of the particles; hence, the larger and smaller particles and nanoparticles were not separated. Figure 4b shows an TEM image of the CZTS ink centrifuged for 10 min at 1500 rpm. A portion of the large particles was removed, which reduced the agglomeration. The particle boundaries were clear; however, particles larger than several hundred nm remained. To further reduce the size of the particles, the dispersion was centrifuged at a high speed of 6000 rpm for 10, 20, and 30 min. The results are shown in Figure 4c–e, respectively. The sample shown in Figure 4c, had the largest particles (in the range of 100 to 200 nm) and almost no agglomeration was observed. In sample (d), particles remaining in the dispersion were smaller than 100 nm, indicating that nanoparticles were obtained. The particle size of sample (e) was in the range of 50 to 100 nm, which indicated that after the treatment to obtain sample (d), the particle size of the dispersion was no longer affected by centrifugation because of the limitations of final particle sizes generated by ball milling processes.

**Figure 4.** TEM images of CZTS dispersions with different centrifugal conditions. Distribution of CZTS particle size in inks with different centrifugal conditions (**a**) Without centrifugation; (**b**) 1500 rpm for 10 min; (**c**) 6000 rpm for 10 min; (**d**) 6000 rpm for 20 min; and (**e**) 6000 rpm for 30 min.

#### *3.2. Deposition of CZTS Precursors*

The CZTS nanoparticle inks were used to deposit the CZTS precursors on glass substrates by a spin-coating method. The speed of the substrate was approximately 2000 rpm and 5 μL of CZTS ink was dripped at the center of the substrate for each drop, which was repeated 10 times to obtain a

film with a thickness of 1–1.5 μm. Figure 5a–c shows the surface morphology of the CZTS film with different magnifications. The SEM image showed a compact morphology with grains smaller than 100 nm without cracks and no large particles were observed. The specific grain size could not be measured because of the small boundaries between grains. Because the precursor was only grown at room temperature, an additional high-temperature treatment was necessary to improve the grain size and crystallinity of the film.

**Figure 5.** SEM images of a CZTS film deposited from the as-fabricated CZTS nanoparticle inks under different magnifications: (**a**) 20000×; (**b**) 10000×; and (**c**) 5000×.

## *3.3. Annealing of the Precursor*

To induce grain growth and reduce the residual organic impurities, the CZTS precursor was annealed in an atmosphere with a high sulfur vapor pressure for 20 min at a temperature of 600 ◦C. Figure 6a,b shows the surface and cross-sectional SEM images of the CZTS films after annealing, respectively. Comparing the precursor morphology, as shown in Figure 5, the grain size increased markedly. The final grain size ranged from several hundred nm to several μm and cracks begin to appear between the grains, either because of grain growth or decomposition of the CZTS particles. According to the cross-sectional image (Figure 6b), the grains extended throughout the film in the thickness direction, which is expected for high-quality films. However, cracks stretching from the surface to the bottom of the film were also observed (marked by the red arrow), indicating the low density of the film. One explanation for this cracking was reported by Scragg, et al. owing to decomposition of CZTS film, as shown in following reactions (1) and (2) [17].

$$\rm Cu\_2ZnSnS\_4 \rightleftharpoons Cu\_2S(s) + ZnS(s) + SnS(s) + 1/2S\_2(g) \tag{1}$$

$$\text{SrS}(\text{s}) \rightleftharpoons \text{SrS}(\text{g}) \tag{2}$$

One solution to overcome this issue is to reduce the annealing temperature to prevent equilibrium (1) from shifting to the right and extending the annealing time to ensure maintain the crystallinity.

**Figure 6.** (**a**) Surface and (**b**) cross-section of an annealed CZTS film annealed at 600 ◦C in a S-rich atmosphere.

To make a comparison, CZTS film using centrifugation condition: 1500 rmp for 10 min was also annealed with the same annealing condition and completed solar cell structure (Please refer to the Supplementary Materials).

Table 1 shows the composition of the CZTS precursor and annealed film, as determined by energy-dispersive X-ray spectroscopy (EDX). The precursor had a sulfur composition less than 50% whereas the sulfur content increased to 50.5% after annealing, indicating that the film was converted from sulfur poor to sulfur-rich, which produces p-type CZTS films. It has been widely reported that Zn-rich (Zn/Sn > 1.0) films are required for fabricating high-performance CZTS solar cells [18,19], meaning that the composition of our CZTS films needed to be adjusted. One possible way to adjust the film to Zn-rich is to fabricate a thin layer of ZnS nanoparticles between the CZTS precursor and Mo back-contact, such that in the following annealing step, both Zn and S will be supplemented.

**Table 1.** Composition of precursor and annealed film as measured by energy-dispersive X-ray spectroscopy (EDX).


Figure 7 shows the XRD patterns of the precursor and annealed film of CZTS. The crystallinity was also improved by high-temperature annealing. The sulfurization process induced sharpening and strengthening of the peaks. All the peaks of the precursor and the annealed film were assigned to kesterite CZTS. No peaks of secondary phases, such as ZnS and Cu2S, which easily form at high temperatures [20], were detected by XRD. However, XRD alone is incapable of identifying small amounts of secondary phases because of its detection limits. To complement this method, we also performed Raman measurements to confirm the absence of secondary phases. Raman spectra of the precursor and annealed CZTS thin films are shown in Figure 8. The lower spectrum shows the annealed CZTS film with peak fitting by a Lorentzian curve. According to the figure, the precursor showed one peak at 330 cm<sup>−</sup>1, corresponding to the A mode of kesterite CZTS. The annealed film exhibited a typical Raman spectrum of kesterite CZTS films with three peaks at 285, 330, and 369 cm<sup>−</sup>1, corresponding to the two A symmetry modes and a B symmetry mode of the CZTS kesterite structure, respectively [21,22]. This result also indicated that no secondary phases are observed after the annealing process.

The annealed CZTS films were used to fabricate complete solar cell structures. Solar cell performance was evaluated under standard conditions. The conversion efficiency of three cells on the same sample was measured as shown in Table 2. The solar cell ranged from 2.5% to 6.2%, indicating ununiform solar cell performance due to the poor film quality as shown in Figure 7.


**Table 2.** Performance of CZTS solar cells.

**Figure 7.** XRD of precursor and annealed CZTS film.

**Figure 8.** Raman spectrums of precursor and annealed CZTS film with fitting of the peaks using Lorentzian curve.

Figure 9 shows dark and light I–V curves of solar cell using annealed CZTS film as the absorber layer with best solar cell performance. The photovoltaic device exhibited an efficiency of 6.2%, with *Voc* = 633.3 mV, *Jsc* = 17.6 mA/cm2, and FF = 55.8%, for an area of 0.20 cm2.

**Figure 9.** J–V curve of CZTS.

Figure 10 shows the external quantum efficiency (EQE) curve of the CZTS solar cell. Over the visible range of the solar spectrum, the maximum QE was less than 60%, indicating strong recombination. The QE curve decreased sharply in the infrared region at 770 nm, which is the CZTS absorption edge. Thus, the calculated bandgap of the CZTS films was approximately 1.61 eV. The features near 510 nm and 380 nm correspond to the absorption edges of the CdS and ZnO layers [23,24], which are commonly used CdS buffer and ZnO window layers.

**Figure 10.** External quantum efficiency (EQE) of CZTS solar cell.

On the basis of the EQE data of a solar cell, *Jsc* was calculated as [25]

$$J\_{\rm sc} = q \int\_0^\infty QE(E) b\_\rm s(E, T\_\rm s) dE \tag{3}$$

where, *q* is the elementary charge, *QE* is the quantum efficiency, and *bs* is solar flux or irradiation. For an air mass of 1.5, the data is available from Ref. [26]. On the basis of Equation (3), Figure 10, and the solar irradiation spectrum, *Jsc* of the CZTS solar cells was calculated to be 14.2 mA/cm2, because the J-V curve represents the real performance of a photovoltaic device. The slight deviation of *J*sc calculated from the QE curve can be explained by the fact that the QE measurement is performed at a single wavelength with a much lower intensity than one-sun irradiation.
