Preparation of a Dense Cu(In,Ga)Se₂ Film From (In,Se)/(Cu,Ga) Stacked Precursor for CIGS Solar Cells

Abstract

The $Cu(In,Ga)Se_2$ (CIGS) thin film obtained by two-step process (metal deposition and Se annealing) has a rough surface morphology and many voids at the CIGS/Mo interface. To solve the problem a precursor that contains Se was employer by depositing a (In,Se)/(Cu,Ga) stacked layer. We devised a two-step annealing (vacuum pre-annealing and Se annealing) for the precursor because direct annealing of the precursor in Se environment resulted in the small grains with unwanted demarcation between stacked layers. After vacuum pre-annealing up to $500^{\circ}C$ the CIGS film consisted of CIGS phase and secondary phases including $In_4Se_3$, InSe, and $Cu_9(In,Ga)_4$. The secondary phases were completely converted to CIGS phase by a subsequent Se annealing. A void-free CIGS/Mo interface was obtained by the two-step annealing process. Especially, the CIGS film prepared by vacuum annealing $450^{\circ}C$ and subsequent Se annealing $550^{\circ}C$ showed a densely-packed grains with smooth surface, well-aligned bamboo grains on the top of the film, little voids in the film, and also little voids at the CIGS/Mo interface. The smooth surface enhanced the cell performance due to the increase of shunt resistance.

1. Introduction

Thin film solar cells based on chalcopyrite compound Cu(In,Ga)Se₂ have been attracted much attention due to high efficiency and possible low-cost fabrication. These solar cells have achieved high conversion efficiencies nearly 22.6~22.9%for laboratory scale devices^1,2). CIGS thin film solar cells were fabricated using co-evaporation and two-step process. Two-step process has attracted much attention due to its scalability, large area, and high-efficiency devices^3,4). In the two-step process, the precursor metals is deposited first and then the precursor is reacted to form selenide by annealing in Se, H₂Se, and ditertbutyl diselenide environment^3-5)

However, the structural, morphological and electrical properties of the absorber films grown by two-step process strongly depend on the nature of the precursors (metallic or selenium containing), annealing temperature, annealing time, and annealing environment.

The films grown from two-step process with metal precursors having Cu, Ga and In elements exhibit large voids at the CIGS/Mo interface due volume expansion during the selenization, resulting in the increase of contact resistance and resulting in the poor adhesion and delamination in worst case⁶⁾. The films grown with metal precursor exhibits Ga depletion at the top of the CIGS film surface thereby reducing the bandgap of the film at the surface. The decrease of bandgap at the surface leads lower open-circuit voltage in devices^6,7).

To minimize voids at the CIGS/Mo interface Si-containing precursor can be utilized. Some studies have been reported on cigs film formation using Si-containing Cu-Ga-In-Se precursors⁸⁾, single compound CIGS sputtering^9-13), and (Cu,Ga)/In/Se precursor14). The efficiencies achieved with these methods were about 8-13%. The disadvantage of single compound CIGStarget is the difficulty of large grains. Based on the literature report there are still many rooms to understand the reaction mechanism and increase grain size with well packing in the theSe-containing precursor for large are applications

The objective of this study is to understand the reaction kinetics during the annealing process from the Se-containing precursors and obtain an optimal annealing condition to form high-quality CIGS absorbers with compact, dense, and less voids at the CIGS/Mo interface. To achieve this, we employed a precursor layer that consists of four (In,Se)/(Cu,Ga) stacks onMo by co-evaporation at low temperature and annealed in vacuum and selenium atmospheres.

The films pre-annealed in a vacuum at 450°C and subsequently annealed in Se environment at 550°C showed large grains with dense packing and less voids in the absorber and at the CIGS/Mo interface, while the Ga deficiency at the surface was observed in the films. The Ga deficiency at the CIGS surface can be solved by further annealing in a vacuum environment.

2. Experimental

2.1 Precursor deposition

The CIGS thin films were grown by a two-step process: deposition of precursor and annealing of the precursor. Fig. 1(a )shows the schematic diagram of the precursor layer consisting of offour alternatives (In,Se)/(Cu,Ga) stacks, which was deposited byco-evaporation described in the experimental section.

Fig. 1. Schematic cross-sectional diagram (a) and SEM cross-sectional image (b) of precursor consisting of four (In,Se)/ (Cu,Ga) stacks

The precursor layer was formed by repeated sequential co-evaporation of Cu+Ga and In+Se on a Mo-coated soda-lime glass (SLG) at room temperature. The fluxes of Cu, In, Ga, andSe were maintained at 1.0, 1.9, 0.8, and 25 Å/s, respectively, as determined by a quartz crystal microbalance. After the deposition, the precursor layer was somewhat intermixed as revealed from Fig.1(b). The brighter line and darker line represent (Cu,Ga)layer and (In,Se) layer respectively. A precursor with four stacks of (In,Se)/(Cu,Ga) was selected because the metal precursor employed for precious development in Avancis Korea consisted of In/(Cu,Ga) four stacks. We wanted to minimize the experimental variation.

The overall cation composition in the precursor was maintained with Cu/(In+Ga) = 0.8, Ga/(Ga+In) = 0.3, and Se/M =0.79 (44% atomic Se among all). With the Cu composition ofCu/(In+Ga) = 0.8 the stoichiometric Se/M ratio in is 1.06. Ourprecursor with Se/M = 0.79 is in selenium-poor compositioncompared to the stoichiometric value.

The precursors were annealed in a Se environment for 30min. However, the direct annealing of the precursor in a Seenvironment did not provide large grains and showed unwanted demarcation between stacked layers. Therefore, we introduced a pre-annealing in vacuum before the Se annealing. The sample was naturally cooled down to 350°C in a Se environment and cooled down further in a vacuum.

2.2 Film characterization

The crystal structure of the annealed precursors and CIGSfilms was investigated using X-ray diffraction (XRD; RIKAKUUltima IV x-ray diffractometer operated at 40 kV/40 mA withCuKα1 1.405 Å) and Raman spectroscopy (LabRAM HighResolution Visible-NIR) with laser excitation wavelength of532 nm. The morphologies of the films were analyzed using a scanning electron microscope (SEM, Nova230). The average composition of the precursor was analyzed using an energy dispersive spectroscopy (EDS) installed in the SEM. The atomic depth profile of the CIGS film was obtained by a time of flight secondary ion mass spectrometry (TOF-SIMS system from ION-TOF GmbH).

2.3 Cell fabrication

CIGS thin film solar cell devices were fabricated in anAl/ZnO:Al/i-ZnO/CdS/CIGS/Mo configuration structure. A50-nm thick CdS buffer layer was deposited by chemical bath deposition using CdSO₄ as a Cd source, thiourea as a sulfur oxide layer consisting of a 50-nm thick intrinsic ZnO layer and a 350-nm thick Al-doped ZnO layer was deposited by rf magnetron sputtering. The Al metal electrode was deposited by thermal evaporation through an aperture mask. The cell area(0.45 cm2) is defined with mechanical isolation. The illuminated current-voltage (J-V) properties of the CIGS solar cells were measured using a solar simulator at an AM 1.5 spectrum illumination with 100 mW/cm² intensity after calibration with a reference silicon solar cell.

3. Results and discussion

3.1 Direct annealing in Se environment

Fig. 2 shows the SEM cross-sectional images of CIGS film prepared by annealing in a Se environment at 450 and 550°C. Four layers and the demarcation of four (In,Se)/(Cu,Ga) stacks are clearly seen by the direct selenization. Large grains are observed on the surface layer and all the other layers contains small grains corresponding to the thickness of each layer. Also the surface is very rough with many crevices. The direct growthof CIGS film is not successful from the morphology point of view. We need to devise a new way to improve the surface roughness and reduce crevices.

Fig. 2. SEM cross-sectional images of CIGS film prepared by annealing in a Se environment at 450 and 550°C

3.2 Phase evolution in vacuum annealing

The precursor was annealed in vacuum to enhance the chemical reaction from precursor to CIGS and analyzed the phase formation at various temperatures.

Fig. 3 shows the SEM cross-sectional images of the CIGSfilms annealed at (a) 350°C, (b) 400°C, (c) 450°C, and (d)500°C for 10 min in vacuum. The stacked-layer demarcationnshown in the precursor disappeared at 350°C and compounds including CIGS and In4Se3 were formed. At 400°C, round-shaped grains were formed and the grain sizes were about 250nm. At 450°C, the film showed a dense film with increased grains sizes. A further increase in temperature 500°C the film showed a dense morphology with 1-μm size grains.

Fig. 3. SEM cross-sectional images of CIGS films annealed at various temperatures in vacuum for 10 min

It is noted that no void was observed at the CIGS/Mo interface for the films annealed at 450 and 500°C. This suggests that annealing temperatures of 450 to 500°C are sufficient for growing large grains. However, the film consists of secondary phases as revealed by XRD studies, indicating that films must be annealed in selenium atmosphere to convert there remaining second phases into CIGS phase with large grain sizes for device applications. Based on the above experiments the precursorsannealed at 400 and 450°C in vacuum were chosen aspre-annealing conditions to obtain CIGS films.

Fig. 4 shows the x-ray diffraction patterns of the precursorsannealed in vacuum for 10 min. The precursors annealed at300°C consist of In₂Se₃ and Cu₂-xSe and Cu4Ga9 phases, the 2θpeak position at 43.7° appears from Cu2-xSe and Cu9Ga4. Itclearly suggests that the (In,Se)/(Cu,Ga) stacks in the precursorwere intermixed at these temperatures to form binary phases. Asthe temperature increased to 350°C the precursors had an intermetallic Cu9(In,Ga)₄ compound along with CIGS phase.The peak of Cu9Ga₄ shifted to lower diffraction angles from 43.7to 43° due to incorporation of In into Cu₉Ga₄ phase.

Fig. 4. XRD patterns of CIGS film annealed at various temperatures for 10 min in vacuum: 1 Cu2-xSe, 2 In2Se₃, 3 Cu4(In,Ga), 4 CIGS, 5 In₄Se3, 6 InSe, 7 Cu₉(In,Ga)₄

As the temperature increased to 400°C the film showed Cu₉(In,Ga)₄, InSe, CIGS phases, and another binary In₄Se₃ compound. The CIGS phase was formed at this temperature due to the reaction of In₄Se₃ and Cu4(In,Ga)9 phases. This is in agreement with the reported phases in the Cu-In-Ga system and the reaction mechanism from Cu-In-Ga precursors with selenium6,^15-17).

With further increase in temperature to 450 and 500°C the films contain CIGS and Cu9(In,Ga)₄ phases. Since the precursors were Se poor Se/M ratio =0.79, the precursors annealed in vacuum are not converted into CIGS films completely due to selenium re-evaporation from the precursor at 450 to 500°C which leads to the formation of ternary metallic compounds. This suggests that the selenium in the precursor is not sufficient for complete conversion into CIGS films indicating the requirement of external selenium flux during the film formation.

3.3 Phases after vacuum pre-annealing and Seannealing

Fig. 5 shows the Raman spectra of the CIGS films after-annealing in vacuum for 10 min and annealed in a Seenvironment for 30 min at 300, 400, and 500°C. In Raman spectroscopy, the Cu2-xSe binary phase can be well detected because of its high structural sensitivity compared to quaternaryCIGS structure

Fig. 5. Raman spectra of CIGS films annealed in vacuum for 10min and then annealed in Se environment for 30 min at 300, 400, and 500°C

A Cu2-xSe second phase existed after 300°C, which is also seen in the vacuum annealing. It means the Cu2-xSe is originated by the reaction of Cu in (Cu,Ga) layer and Se in (In,Se) layer.The Cu2-xSe second phase completely disappeared at 400°Cannealing and the InSe phase which was distinctly observed in vacuum annealing disappeared after Se annealing. Only pureCu(In,Ga)Se2 phase is detected after Se annealing at 400 and500°C, indicating that a complete CIGS phase is possible by Seannealing down to 400°C.

Fig. 6 shows the SEM cross-sectional images of CIGS films obtained by pre-annealing in vacuum for 10 min and then annealing in a Se environment for 30 min, named as two-step annealing. The film prepared at 400°C pre-annealing and 400°CSe annealing (V400+Se400) shows small grains with a multilayer structure. The demarcation of stacked precursor was pronouncedly revealed when the Se annealing temperature increased to 450°C as shown in Fig. 6(b). This suggests that400°C pre-annealing is not high enough for large grains. It is necessary to increase the pre-annealing temperature higher than400°C.

Fig. 6. SEM cross-sectional images of CIGS films pre-annealed in vacuum for 10 min and annealed Se environment for 30 min at various temperatures: (a) V400+Se400, (b) V400+Se450 (c) V450+Se450, and (d) V450+Se550

With pre-annealing 450°C and Se annealing at 450°C (V450+Se450 in Fig. 6(c)) the outline of grains is more clearly seen and the demarcation of the precursor stack layer was weakened compared to that of V400+Se450, suggesting that the CIGS was grown across the demarcation at 450°C in vacuum annealing. Especially, the film grown by 450°C vacuum and 550°C Seannealing (marked as V450+Se550) revealed that the grains onthe top side of the film were well aligned as bamboos and were densely packed each other with no grove at the grain boundaries.Also the films also showed much less void in the film and on the bottom side of the film. The microstructure process with V450+Se550 is a very desirable morphology with a smooth surface, well-aligned grains on the surface, and densely packed grains in the bulk, little voids at the bottom side of film, and no void at theCIGS/Mo interface.

Fig. 7 shows the SIMS profile of Cu, In, Ga, and Se elements in the CIGS films pre-annealed in vacuum for 10 min and annealed in Se environment for 30 min at (a) 400°C (marked as V400+Se400) and (b) 450°C (marked as V450+Se450).The film grown at 400°C as shown in Fig.7(a) shows the Ga concentration is sharply depleted at the CIGS surface and is uniform through the bottom side of CIGS film. The In concentration is high at the surface and low at the bottom side ofCIGS film. The Se concentration is homogeneously distributed through the film depth. The Cu concentration is unevenly distributed through the film depth and accumulated at theMo/CIGS interface.

Fig. 7. SIMS depth profile of Cu, Ga, In, and Se in CIGS films pre-annealed in vacuum for 10 min and annealed in Se environment for 30 min at400(a) and 450oC (b)

The 400°C annealing led to the sharp depletion of Ga concentration at the surface and uneven distribution of Cu concentration within CIGS film. The Cu concentration is continuously and evenly distributed and the Ga at the surface is less depleted by annealing at 450°C as shown in Fig. 7(b).

The depletion of Ga at the surface is widely observed theCIGS films grown from a two-step process, regardless of the nature of precursor stacks such as sputtered (Cu,Ga)/In3),(Cu,In)/(Cu,Ga)6), electroplated9), oxide-based precursors10), Secontaining precursor11). The accumulation In and depletion of a at the surface are attributed to the preferential reaction of Inwith Se due to the difference in the electronegativity of In andGa6).

Fig. 8 shows the depth profiles of Ga/(Ga+In) ratio in theCIGS films, determined from the calibrated SIMS profiles with reference to EDAX data. The film annealed at 400°C shows aGa depletion down to 350-nm depth and a steep slope above600-nm depth of CIGS film, increasing from 0.3 to 0.7. The film annealed at 450°C shows a Ga depletion down to 300-nm depth and a slow grade, increasing from 0.3 to 0.4.

Fig. 8. Depth profiles of Ga/(Ga+In) ratio in CIGS films annealed in vacuum for 10 min and in Se environment for 30 min at 400 (marked as V400+Se400) and 450°C (marked as V450+Se450)​​​​​​​

The solid line in Fig. 8 is a reference GGI profile to obtain above 18% cell efficiency. The film we prepared has low GGI at the CIGS surface while Ga grading in the bulk can be adjusted by increasing the Ga composition in the precursor and with a more precise temperature control. However, the very low GGIat the CIGS surface is a big disadvantage, causing a low open circuit voltage (Voc).

3.4 Cell performance

Fig. 9 shows the illuminated current-voltage curves of theCIGS solar cells fabricated with CIGS films grown at three different conditions. The photovoltaic parameters such as cell efficiency (η), short-circuit current (Jsc), open-circuit voltage(Voc), fill factor (FF), series resistance (Rs), and shunt resistance(Rsh) are listed in Table 1. The cell efficiency prepared with theCIGS film processed at 400°C (marked as V400+Se400 in Fig.8) was 9.7%. The efficiency decreased to 8.3% as the annealing temperature increased to 450°C, marked as V450+Se450 in Fig.9, due to smaller Jsc than the cell with 400°C-processed CIGSfilm. The reason of smaller Jsc at 450°C is due to less grading ofGa depth profile in Fig. 8.

Fig. 9. Illuminated current-voltage curves of CIGS thin film solar cells fabricated with three different CIGS films preannealed in vacuum for 10 min and annealed in Se environment for 30 min​​​​​​​

Table 1. Photovoltaic parameters of CIGS solar cells fabricated with the CIGS films annealed at various conditions described in Fig. 8​​​​​​​​​​​​​​

The efficiency again increased 10.4% when the CIGS film was pre-annealed at 450°C in vacuum and annealed at 550C in Se, marked as V450+Se550 in Fig. 9, due to the recovery of Jsc and increase of both Voc and FF. The improved conversion efficiency with 550°C Se annealing is attributed with improved surface roughness of the CIGS films.

The Rs and Rsh for CIGS film processed by V450+Se450 are1.2 and 820 Ω ․ cm², respectively, and those for CIGS film processed by V450+Se550 are 0.7 and 1200 Ω ․ cm², respectively.The typical Rsh value of CIGS cells with a rough CIGS surface was in the range of 150~350 Ω ․ cm². The Rsh value suggests that the CdS buffer could be covered more uniformly on the smoothCIGS surface, resulting in the sturdier buffer to reduce tunneling recombination through the CdS buffer layer. It is noted that the decrease of Rs and increase of Rsh after 550°C Se annealing resulted in the increase of both Voc and FF.

In overall, the Voc is still too small due to low band gaps at the surface of the CIGS films, expected from Ga deficiency as shown in Fig. 8. To increase the bandgap at the CIGS surface we need an additional vacuum post-annealing process to push Gaout to the surface. For the purpose, annealing above 550°C and short annealing time of 10 min are required to obtain better Gahomogeneity in the films18,19). This is beyond the scope of this paper. Our paper reports a new a way to obtain a CIGS film with a smooth surface, well aligned-grain on surface, densely packed grains in the bulk, little voids in the bottom and at theCIGS/Mo interface.

4. Conclusions

CIGS absorbers were grown by a two-step process. A Secontaining precursor that consisted of four stacks of (In,Se)/(Cu,Ga) layer was prepared by co-evaporation process. Growth of CIGS film by direct Se annealing was unsuccessful because the film showed small grains with clear demarcation between each stack. We introduced vacuum pre-annealing process before Se annealing to improve the film morphology.

The films grown at 450 and 500°C in vacuum showed a densely packed grain with large gains and little voids. The reason of densely packed grains was due to the formation ofinter metallic phase during pre-annealing in vacuum. The precursor annealed at 300°C in vacuum had Cu2-xSe and In2Se3phases. An InSe phase appeared at 400 and 450°C, and a ternary metallic Cu9(In,Ga)4 appeared at 500°C.

The CIGS films obtained with 450°C pre-annealing and 550°C Se annealing a densely-packed CIGS film with a smooth surface, a well-aligned bamboo-like grains on top of the film,little voids in the bottom and at Mo/CIGS interface. The SIMSprofile shows the Ga was depleted at the CIGS surface

The film prepared with 450°C pre-annealing in vacuum and550°C Se annealing that has the best morphology showed best cell performance 10.4% because of the smooth surface with well-aligned grains. The next step is to increase the bandgap of the CIGS surface by alkaline post-deposition treatment and it is beyond the scope of this paper.