In my pursuit of advancing renewable energy technologies, I have focused extensively on thin film solar panels, particularly those based on earth-abundant materials. Among these, copper zinc tin sulfur selenide (Cu2ZnSn(S,Se)4, or CZTSSe) thin film solar panels stand out due to their tunable bandgap, high absorption coefficient, and non-toxic constituents. The performance of these thin film solar panels hinges critically on the quality of the absorber layer, which is often fabricated via solution-based methods followed by a selenization heat treatment. This process transforms a precursor film into a crystalline CZTSSe layer, and its optimization—encompassing selenization steps, temperature, and time—is paramount for achieving high photovoltaic efficiency. In this comprehensive study, I delve into the intricacies of the selenization process, employing systematic experiments to elucidate its impact on absorber layer morphology, phase purity, and ultimately, the device performance of thin film solar panels.
The structure of a typical CZTSSe thin film solar panel is multilayered. It begins with a soda-lime glass substrate coated with a molybdenum (Mo) back contact, followed by the CZTSSe absorber layer, a cadmium sulfide (CdS) buffer layer deposited via chemical bath deposition, a window layer stack of intrinsic zinc oxide (i-ZnO) and indium tin oxide (ITO) sputtered thereon, and finally aluminum front grid electrodes. The heart of the device, the absorber, must exhibit large, columnar grains extending through the layer thickness to minimize carrier recombination. Solution processing, especially using dimethyl sulfoxide (DMSO)-based precursor inks, offers a low-cost, scalable route for absorber fabrication. However, the subsequent selenization step in a controlled atmosphere (often involving Se vapor and N2 carrier gas) is where the precursor film crystallizes into the desired kesterite CZTSSe phase. Inadequate selenization leads to small grains, secondary phases, and poor interfacial properties, whereas excessive selenization can cause decomposition, void formation, and overly thick MoSe2 layers at the back contact—all detrimental to thin film solar panel efficiency.
To systematically investigate the selenization process for CZTSSe thin film solar panels, I designed experiments varying three key parameters: the selenization program (single-step vs. two-step), the peak selenization temperature, and the dwell time at high temperature. All precursor films were prepared by spin-coating a DMSO-based solution containing copper acetate, zinc chloride, tin chloride, and thiourea onto Mo-coated glass, followed by pre-annealing to form a CZTS precursor film. Selenization was performed in a tubular furnace with a constant N2 flow, using elemental Se pellets as the Se source. The conditions are summarized in Table 1.
| Selenization Type | Step 1: Ramp to T1 / Dwell | Step 2: Ramp to T2 / Dwell | Label |
|---|---|---|---|
| Single-Step | Ramp to 560°C in 1 min / 15 min | N/A | S-560 |
| Two-Step | Ramp to 350°C in 0.5 min / 5 min | Ramp to 520°C in 0.5 min / 15 min | T-350/520 |
| Ramp to 350°C in 0.5 min / 5 min | Ramp to 540°C in 0.5 min / 15 min | T-350/540 | |
| Ramp to 350°C in 0.5 min / 5 min | Ramp to 560°C in 0.5 min / 15 min | T-350/560 | |
| Ramp to 320°C in 0.5 min / 5 min | Ramp to 540°C in 0.5 min / 15 min | T-320/540 | |
| Ramp to 380°C in 0.5 min / 5 min | Ramp to 540°C in 0.5 min / 15 min | T-380/540 |
Additionally, to probe the effect of dwell time, I modified the high-temperature step for selected conditions: T-350/540 with 10 min dwell (T-350/540-10) and T-350/520 with 20 min dwell (T-350/520-20). The resulting CZTSSe absorber layers were characterized by X-ray diffraction (XRD), Raman spectroscopy, and scanning electron microscopy (SEM). Complete thin film solar panel devices were then fabricated by depositing CdS, i-ZnO/ITO, and Al grids. Their current density-voltage (J-V) characteristics under AM1.5G illumination and external quantum efficiency (EQE) were measured to evaluate photovoltaic performance.
The phase evolution during selenization is complex. The general reaction pathway can be described by considering the formation of intermediate binary and ternary phases before final CZTSSe crystallization. In a simplified model, the precursor film containing Cu-Sn-S/Se and Zn-S/Se compounds reacts to form CZTSSe. The kinetics depend heavily on temperature and Se partial pressure. A two-step process, with an intermediate soak at a lower temperature (e.g., 350°C), allows for better Se incorporation and the formation of intermediate phases like Cu2SnSe3 and ZnSe, which subsequently react to form CZTSSe at higher temperature. In contrast, a single-step rapid thermal process might lead to incomplete reactions and persistence of secondary phases. The presence of secondary phases like ZnSe, Cu2SnSe3, or SnSe2 can act as recombination centers, degrading the performance of thin film solar panels.
My XRD analysis confirmed these postulations. The single-step selenized sample (S-560) showed diffraction peaks corresponding to CZTSSe, but also minor peaks attributable to CuSe2 and Sn(S,Se)2. All two-step selenized samples exhibited primarily the kesterite CZTSSe peaks (PDF #52-0868) with no detectable secondary phase peaks in XRD. However, due to the similarity in crystal structures, Raman spectroscopy with a 488 nm laser was employed for finer phase identification. The Raman spectra revealed the characteristic CZTSSe modes near 173, 197, 236, and 242 cm-1. For sample S-560, a strong peak at 251 cm-1, assigned to ZnSe, was evident, indicating residual secondary phases. In contrast, the two-step samples, particularly T-350/540, showed minimal ZnSe signal, suggesting higher phase purity. This underscores the advantage of a two-step selenization protocol for obtaining phase-pure absorbers in thin film solar panels.
The surface and cross-sectional morphology, as revealed by SEM, profoundly impacts the electrical properties of thin film solar panels. Single-step selenization (S-560) resulted in a relatively sparse surface with smaller grains and, critically, large voids at the interface between the CZTSSe layer and the Mo back contact. These voids likely arise from the volatilization of SnSe2 formed during rapid heating, and they severely compromise back-contact adhesion and increase series resistance. Two-step selenization generally yielded larger surface grains and a bilayer grain structure in cross-section: a layer of larger grains near the back contact and another near the top surface. Among the two-step conditions, T-350/540 produced the most compact surface and a well-defined bilayer with minimal voids and a relatively thin MoSe2 layer. The thickness of the MoSe2 layer, which forms inevitably during selenization, is crucial; while a thin layer is beneficial for ohmic contact, an overly thick layer increases series resistance. The MoSe2 thickness correlated with the second-step temperature: T-350/560 (560°C) showed the thickest MoSe2, while T-350/520 and T-350/540 showed thinner layers. The grain size can be estimated from SEM images, and its dependence on selenization temperature T2 follows an Arrhenius-type relationship for grain growth:
$$ G = G_0 \exp\left(-\frac{E_a}{k_B T_2}\right) $$
where \(G\) is the average grain diameter, \(G_0\) is a pre-exponential factor, \(E_a\) is the activation energy for grain growth, \(k_B\) is Boltzmann’s constant, and \(T_2\) is the absolute temperature. However, excessively high temperatures (e.g., 560°C) can lead to abnormal grain growth and decomposition, explaining the observed voids in S-560 and T-350/560.
The photovoltaic performance of the completed thin film solar panels directly reflects the absorber quality. Table 2 summarizes the average photovoltaic parameters from multiple devices for each selenization condition.
| Condition | Open-Circuit Voltage, \(V_{oc}\) (V) | Short-Circuit Current Density, \(J_{sc}\) (mA/cm2) | Fill Factor, \(FF\) (%) | Power Conversion Efficiency, \(\eta\) (%) |
|---|---|---|---|---|
| S-560 | 0.34 ± 0.01 | 31.76 ± 1.1 | 37.6 ± 1.0 | 4.05 ± 0.17 |
| T-350/520 | 0.408 ± 0.005 | 32.88 ± 1.3 | 31.29 ± 1.2 | 4.39 ± 0.28 |
| T-350/540 | 0.374 ± 0.008 | 35.64 ± 1.0 | 42.17 ± 1.5 | 5.72 ± 0.10 |
| T-350/560 | 0.351 ± 0.012 | 33.64 ± 1.4 | 39.47 ± 1.4 | 5.21 ± 0.22 |
| T-320/540 | 0.367 ± 0.011 | 34.45 ± 1.9 | 40.96 ± 2.0 | 5.58 ± 0.08 |
| T-380/540 | 0.373 ± 0.014 | 34.97 ± 0.5 | 38.49 ± 1.2 | 5.03 ± 0.30 |
The efficiency \(\eta\) is calculated from the standard formula:
$$ \eta = \frac{V_{oc} \times J_{sc} \times FF}{P_{in}} \times 100\% $$
where \(P_{in}\) is the incident power density (100 mW/cm2 for AM1.5G). The two-step selenized devices consistently outperformed the single-step device. The champion condition was T-350/540, yielding an average efficiency of 5.72%, with the best device reaching 5.91%. This represents a significant improvement over the single-step process (4.05%) and highlights the critical role of optimized thermal profiling. The higher \(J_{sc}\) for T-350/540 is attributed to its compact, large-grained absorber which enhances light absorption and carrier collection. The fill factor \(FF\) is strongly influenced by series resistance (\(R_s\)) and shunt resistance (\(R_{sh}\)). The improved morphology and thinner MoSe2 layer in T-350/540 likely reduce \(R_s\), contributing to its high \(FF\) of over 42%. The open-circuit voltage \(V_{oc}\) generally decreased with increasing second-step temperature, which can be linked to a reduction in the absorber bandgap due to increased Se incorporation at higher temperatures. This bandgap narrowing was confirmed by analysis of the EQE spectra.
The external quantum efficiency, which measures the fraction of incident photons converted to collected electrons at each wavelength, provides insight into the spectral response and optical bandgap (\(E_g\)) of the thin film solar panels. The EQE curves for devices from key conditions are shown conceptually in Figure 1 (the actual image link is inserted earlier). The T-350/540 device exhibited the highest EQE across the visible to near-infrared range (400-1100 nm), consistent with its superior \(J_{sc}\). The bandgap can be estimated from the inflection point of the EQE curve or more accurately by plotting \((EQE \times h\nu)^2\) versus \(h\nu\) (photon energy) and extrapolating the linear region to zero. The relationship is derived from the absorption coefficient \(\alpha\) for direct bandgap semiconductors like CZTSSe:
$$ \alpha h\nu = A (h\nu – E_g)^{1/2} $$
where \(A\) is a constant. Assuming EQE is proportional to \(\alpha\) near the band edge, a Tauc plot yields \(E_g\). The calculated bandgaps were approximately 1.15 eV for T-350/560, 1.20 eV for T-350/540, and 1.25 eV for T-350/520. This trend explains the \(V_{oc}\) variation, as \(V_{oc}\) is often limited by the bandgap according to the detailed balance theory:
$$ V_{oc} \approx \frac{E_g}{q} – \frac{k_B T}{q} \ln\left(\frac{J_{00}}{J_{ph}}\right) $$
where \(q\) is the electron charge, \(T\) is temperature, \(J_{00}\) is the reverse saturation current prefactor, and \(J_{ph}\) is the photocurrent density. Thus, a lower \(E_g\) reduces the maximum attainable \(V_{oc}\), but it can increase \(J_{sc}\) due to broader spectral absorption. The optimal condition T-350/540 strikes a balance, offering a moderate bandgap that maximizes the product \(V_{oc} \times J_{sc}\).
Investigating the dwell time effect further refined the process. For the T-350/520 condition, extending the dwell at 520°C from 15 min to 20 min (T-350/520-20) did not improve efficiency; instead, average \(\eta\) dropped to 2.76%. SEM revealed that prolonged heating at this relatively low temperature led to a porous, small-grained surface without eliminating the intermediate fine-grain layer, likely due to incomplete crystallization and possible Se loss. For the optimal T-350/540 condition, reducing the dwell to 10 min (T-350/540-10) resulted in slightly smaller grains and a lower average efficiency of 4.75%. This indicates that 15 min at 540°C is sufficient for complete grain growth and consumption of intermediate phases, whereas shorter times leave the film under-selenized. These findings emphasize that not only temperature but also time must be optimized to achieve the desired crystalline quality in thin film solar panel absorbers.
The selenization process also influences defect chemistry. CZTSSe is prone to point defects and defect complexes, such as CuZn antisites and [2CuZn + SnZn] clusters, which form deep-level recombination centers. The formation energies of these defects are sensitive to the chemical potential during growth, which is governed by selenization temperature and Se vapor pressure. A higher Se partial pressure (promoted by sufficient temperature and time) favors the formation of Se-rich conditions, which can suppress the formation of certain deleterious defects like SnZn. However, excessive Se can lead to Se vacancies or other disorders. The defect concentration \(N_d\) can be related to processing temperature through an Arrhenius equation:
$$ N_d = N_0 \exp\left(-\frac{E_f}{k_B T}\right) $$
where \(E_f\) is the defect formation energy. Optimizing selenization minimizes \(N_d\), thereby reducing non-radiative recombination and enhancing carrier lifetime \(\tau\), which directly impacts \(V_{oc}\) and \(J_{sc}\) via the relation for diffusion length \(L_d = \sqrt{D\tau}\), where \(D\) is the diffusion coefficient.
In conclusion, my systematic investigation into the selenization process for CZTSSe absorber layers demonstrates that a carefully designed two-step protocol with an intermediate soak at 350°C followed by a 15-minute anneal at 540°C yields the highest quality films for thin film solar panels. This condition produces phase-pure CZTSSe with large, compact grains, a favorable bilayer structure, a thin MoSe2 interlayer, and an optimal bandgap around 1.20 eV. The resultant thin film solar panels achieved an average power conversion efficiency of 5.72%, a substantial improvement over devices made with single-step or non-optimized two-step selenization. The study underscores that the selenization step is not merely a heat treatment but a critical phase evolution and crystal growth process that must be meticulously controlled. Parameters like ramp rate, step temperatures, dwell times, and Se vapor pressure interact complexly to determine the final absorber morphology, stoichiometry, and defect landscape. Future work could involve real-time monitoring techniques like in-situ XRD or Raman during selenization to better understand the kinetic pathways. Furthermore, integrating these optimized absorbers with advanced buffer layers and interface passivation schemes could push the efficiency of CZTSSe thin film solar panels closer to their theoretical limit. As the demand for low-cost, sustainable photovoltaics grows, such process optimizations are essential for making thin film solar panels based on earth-abundant materials commercially viable and impactful in the global energy landscape.
The journey of optimizing thin film solar panels is iterative and deeply rewarding. Each experiment brings new insights into the delicate balance between material synthesis and device physics. The selenization process, in particular, stands as a cornerstone for high-performance CZTSSe photovoltaics. By sharing these findings, I hope to contribute to the collective knowledge driving innovation in thin film solar panel technology, moving us closer to a future powered by clean, efficient, and accessible solar energy.