Effect of Cd1-xMnxSe Alloy Thickness on the Optical and Photovoltaic Properties of Quantum Dot-Sensitized Solar Cells

In this work, the Cd1-xMnxSe alloy was successfully prepared using successive ionic layer adsorption and reaction method to investigate the layers effect on the properties of devices while concentration dopant was optimized at 20% (molar concentrations between Mn2+ and Cd2+ ions in the Cd1-xMnxSe material). The layers of the Cd1-xMnxSe alloy play a role in the improving optical, photovoltaic and electrochemical properties of the solar cells. Hence, the efficiency performance of devices based on the Cd1-xMnxSe alloy reached approximately 3.8%. Besides, in order to explain this result, the experimental I-V curve was also used to determine the resistances at the interfaces and the resistances diffusion of the devices. This dynamic resistance can be compared with that of Electrochemical Impedance Spectra.


INTRODUCTION
Semiconductor quantum dots (QDs) have drawn great attention for application in a number of fields due to the optical properties of these materials (Chen et al., 2016;Liu et al., 2016Liu et al., , 2019aLi et al., 2019). Now, nanoparticles prominently become a dye sensitized for the third-generation solar cells because of low-cost fabrication technology, high photostability, the controlled sizes (Peng and Peng, 2001), higher absorption coefficient (Beard Matthew, 2011), and the multiple exciton generation (Sargent, 2005). However, QDSSCs (quantum dot sensitized solar cells) have reached ∼13% performance, which is lower than the theory limits (Jiao et al., 2017). Recently, plenty of QDs [CdS, CdSe, CdTe (Shen et al., 2015), and PbS (Jumabekov et al., 2014)] are widely applied in the QDSSCs because of their unique properties (Duan et al., 2014). It is noticeable that the CdS and CdSe QDs have attracted considerable interest due to their optical property stability (Lin et al., 2014), a higher conduction band than TiO 2 (Lee and Lo, 2009), low resistivity (Mendoza-Perez et al., 2009), and wide absorbed spectrum (Liji Sobhana et al., 2011). However, this result was still low compared with that of dye-sensitized solar cells (DSSCs). So, a CdS/CdSe system was widely investigated due to its wide absorption spectrum, the shift of the absorption peak toward in the visible region, and rising of the conduction band (CB) as the combined CdS and CdSe QDs compared with TiO 2 CB. However, the performance based on this system achieved 4% efficiency (Lee and Lo, 2009), and its performance was still lower than that of DSSCs due to much trapping and recombination at the TiO 2 /QDs/electrolyte triple interfaces (Abdellah et al., 2014).
In recent times, metal ions doped into the QDs can be replaced by the single QDs and co-sensitized system to reduce achievable recombination (Hodes et al., 1987;Fang et al., 1997Fang et al., , 2011Gratzel, 2001Gratzel, , 2003William Yu et al., 2003;Shen and Lee, 2008;Fan et al., 2009;Gimenez et al., 2009;Zhuge et al., 2009;Gonzalez-Pedro et al., 2010;Schmid, 2014;Tan Phat et al., 2018) because it can improve the charge collection and transfer process. In addition, metal ions are famous for their lowest resistance and large mobility. For example, Tan Phat et al. recorded a performance of 4.22% as Cu 2+ ion doped into CdSe QDs because its attractive optical and magnetic properties were more interesting than that of CdSe and PbS QDs (Tan Phat et al., 2018). Their improving properties can be archived by doping metal in QDs like those in Refs. (Hodes et al., 1987;Gratzel, 2001Gratzel, , 2003Fan et al., 2009;Gimenez et al., 2009;Zhuge et al., 2009;Gonzalez-Pedro et al., 2010) to make contributions to the more absorption photons of photoelectrodes.
Herein, Mn 2+ ions were doped on CdSe nanoparticles to study the optical and photovoltaic properties of the QDSSCs. We investigate how changing the thickness of Cd 1−x Mn x Se films affects efficiency performance. Besides, in order to explain this result, the experimental I-V curve was also used to determine the resistances at the interfaces and the resistances diffusion of the devices. This dynamic resistance can be compared with that of electrochemical impedance spectra.

TiO 2 Films
The TiO 2 paste was deposited onto transparent conducting substrates F-doped SnO 2 (FTO) with 7 cm −2 of the sheet resistance. The FTO/TiO 2 film was sintered in air at 500 • C for 30 min.

TiO 2 /CdS Films
The FTO/TiO 2 film immersed in 0.1 M Cd 2+ solution [2.66 g Cd(CH 3 COO) 2 .2H 2 O was mixed with 100 ml of de-ionized water] followed by 0.1 M S 2− solution (2.4 g Na 2 S.9H 2 O was dissolved in 100 ml of methanol). All processes were repeated from one to three times (denoted FTO/TiO 2 /CdS photoelectrode).
The Se powder was mixed with Na 2 SO 3 (0.6 M) and 100 ml of pure water at 70 • C for about 7 h. To accommodate the doping of Mn metal ion, relevant molar concentrations of 0.3 mM of Mn(CH 3 COO) 2 .2H 2 O were mixed with Cd(CH 3 COO) 2 .2H 2 O anion source. The SILAR process of CdSe and Mn-doped CdSe QDs was similar to that of CdS except that 15 min and 50 • C were required for dipping the TiO 2 /CdS film in the Se aqueous solution. Then, the FTO/TiO 2 /CdS film was dipped in the above solution for 1 min before dipping in Se 2− solution for 1 min at 80 • C (called 1 layer).
Polysulfide solution was made by dissolving 0.5 M Na 2 S.9H 2 O, 0.2 M S, and 0.2 M KCl in DI water/methanol (7:3 by volume). The Cu 2 S counter electrode was synthesized through chemical bath deposition according to a previous publication (Fan et al., 2009). Briefly, 0.24 g CuSO 4 was dissolved in 60 ml of DI in a glass bottle. N 2 was bubbled through the water for 10 min to remove the dissolved oxygen from the system. Then 0.37 g of Na 2 S 2 O 3 .5H 2 O was mixed in the solution, and the color turned to light green. Afterwards, a clean FTO glass was immersed in the solution, with its conductive surface facing down and had an angle against the wall. The system was then settled in the water bath of 90 • C and kept for 1 h. The Cu 2 S crystal would directly grow onto the conductive surface of FTO glass. Finally, the as-prepared Cu 2 S-coated FTO glass sample was rinsed with deionized water and dried in air. The post-heat treatment was carried out in an N 2 atmosphere at 200 • C for 30 min and a structure of device was shown in Figure 1.

Characterization
The scanning electron microscopy (SEM) with a JEOL 7500 F high-resolution scanning electron microscope was used to determine the morphology of films. The structure of materials were recorded by an X-ray diffraction pattern, Philips model, and the absorption spectrum was investigated by a JASCO V-670. The I-V curve was recorded using simulated AM 1.5 sunlight with an output power of 100 mW cm −2 . The resistances of QDSSCs were studied by electrochemical impedance spectroscopy (EIS) Series G750. and TiO 2 /CdS(3)/Cd 0.8 Mn 0.2 Se(3) photoanodes was determined to be ∼11.606, 11.750, and 12.056 µm from Figures 2B-D, respectively. Moreover, 0.5 µm in Figure 2C (0.563 µm in Figure 2D) and 11.006 µm are the thickness of FTO and the TiO 2 /CdS(3) film without FTO. The energy peaks related to Ti and O elements in the TiO 2 film and Cd, Se, and S elements of CdS and CdSe nanocrystal were clearly found in the EDX spectra of TiO 2 /Cd 1−x Mn x Se/CdSe photoanode. Si and C energy peaks had been originated from FTO and excessive organic solution remaining in the layer (since the electrodes were sintered in vacuum), respectively. Mn energy peaks came from the anion precursor solution. The EDX spectra confirmed that QDs had been assembled and crystallized on the TiO 2 layer ( Figure 2E). The optical properties of TiO 2 /CdS/Cd 1−x Mn x Se photoelectrodes were investigated by UV-Vis spectra with different thicknesses (Supplementary Table 1). The red shift is more pronounced with the increase of SILAR cycles due to the growth and thickness of film and attributed to the size quantization effect. This indicates that the high absorption coefficient of CdSe:Mn 2+ QDs is attributed to TiO 2 nanoparticles, which are extended to almost the whole visible region as corresponding to SILAR cycles from 1 to 3 ( Figure 3A). However, a decline in overall absorption was observed when SILAR cycle is higher than 3. This can be attributed to the aggregation of CdSe:Mn 2+ nanocrystal due to decreasing photocurrent and increase in dynamic resistances (Singh et al., 2008;Bhupendra et al., 2011;Cao et al., 2015;Muthalif et al., 2016). Furthermore, the Tauc plot and additional information on it are shown in Table 1, Supplementary Table 2 and Figure 3B. The bandgap of QDs decreased from 2.04 eV for Cd 0.8 Mn 0.2 Se (1) to 1.7 eV for Cd 0.8 Mn 0.2 Se (3) QDs. This result shows that there is a strong influence of the doped concentration and thickness on the energy band structure of the CdSe host material (Gopi et al., 2015). Figure 4A exhibits an alignment energy of photoanode, which includes a dopant energy in the bandgap of CdSe QDs caused by the shift peak, an increasing absorption intensity (shown in Figure 3A), and the (αhν) 2 vs. (hν) curves (shown in Figure 3B). The results are also confirmed by the timeresolved photoluminescence spectrum in Figure 4B and the data in Supplementary Table 3 and Table 1. In a similar manner,  (Thongpron and Kirtikara, 2006;Thanh et al., 2015) and our experimental I-V curves were used to calculate the external dynamic resistance (R D ) and the internal dynamic resistance (R d ), the series resistance (R s ), and the shunt resistance (R SH ) of cells. It is necessary and more important to obtain the reliable characterization in the QDSSCs when the dynamic parameters were determined. We can control and determine the amount of loss mechanism as accurately as possible to improve the efficiency performance in the next work (Sze and Ng, 1981).

RESULTS AND DISCUSSION
The photo current density (I ph ) and open voltage circuit (V OC ) of a solar cell is given by and R d = 1 α (I 2 − I 1 ) ln I ph + I o − I 1 I ph + I o − I 2 (4) R D and R d are the external dynamic resistance and internal dynamic resistance of the equivalent circuit of solar cells. The shunt resistance (R SH ) was obtained: where V o is the initial voltage.  In order to determine the performance, we recorded the I-V curves of QDSSCs with the different layers of Cd 1−x Mn x Se QDs, which is shown in Figure 5. In comparison, It is obvious that the optimized thickness of Cd 1−x Mn x Se (3 layers) QDs made contributions to boost the efficiency of QDSSCs (∼3.8%) (Supplementary Table 4). This result is suitable to that of UV-Vis, lifetime, and IES.
On the whole, our view is that resistances showed up as the increasing SILAR cycles of Cd 1−x Mn x Se films (Sze and Ng, 1981). The result agrees well with that of the I-V curve (3.8% of efficiency). Furthermore, R SH was calculated from Equation 5, and it depended on the technology process. The values of R SH are large, corresponding to a good QDSSC. Looking at Table 1, it reveals that the R SH of CdS/Cd 1−x Mn x Se cosensitized TiO 2 is the largest. This is also confirmed by the long lifetimes of charges with loading SILAR cycles more than 3. In brief, the dynamic resistances, saturated current intensity, lifetimes of charges, and bandgap depend on the thickness of TiO 2 /CdS(3)/Cd 0.8 Mn 0.2 Se(3) with the highest efficiency of 3.8%. Figure 6A gives information about the circuit, which corresponds to the QDSSCs. Figure 6B shows the experimental Nyquist plots of devices corresponding to the resistance at the surface of the polyelectrolyte/counter electrode (denoted as R ct1 ) and the diffuse resistance in the TiO 2 film and TiO 2 /QDs surface (denoted as R ct2 ) (Veerathangam et al., 2017). The lifetime of excited electron (τ n ) is determined from Figure 4B, and the capacitance (c µ ) can be determined by c µ = τ n R ct2 and listed in Supplementary Table 5 and Table 2. As a rule, the Cd 0.8 Mn 0.2 Se (1), Cd 0.8 Mn 0.2 Se (2), Cd 0.8 Mn 0.2 Se (3), Cd 0.8 Mn 0.2 Se (4), and Cd 0.8 Mn 0.2 Se (5) photoelectrodes have significantly changed in the photovoltaic because layers played a role in the recombination process. It is obvious that the thicker the film is, the larger resistance becomes. From Table 2, the resistances of four to five layers are larger than the resistance of three layers, while the excited electrons' lifetime and capacitances are much lower (Omid et al., 2015). Above all, the performance increased because of a rise in CB of the Cd 1−x Mn x Se QDs and a shift of the absorption peak after doping (shown in Figure 4A). Table 1 illustrates the value of dynamic resistances from one illuminated I-V curve and EIS with the same conditions. Looking at the graph, it is immediately obvious that they depend on the SILAR cycles of deposition of Cd 0.8 Mn 0.2 Se with the same rules. In this case, the results show that the R D , R d , R ct1 , and R ct2 are the smallest with loading at three SILAR cycles of deposition, but the value of R SH is the largest. We noted larger R SH indicates a better quality of QDSSCs. The trend of the recombination resistance (R ct2 ) of all devices can clearly be analyzed when the SILAR cycles of deposition are changed. The R D , R d , and R ct2 are characterized by the dynamic processes, dynamic resistances, and resistance transfer at surfaces of TiO 2 /QDs. With the  smallest R D , R d , and R ct2 , the optimum energy conversion efficiency was obtained ∼3.8% at three cycles of deposition (Supplementary Table 6). This is completely suitable with the results of UV-Vis and lifetimes.

CONCLUSIONS
To summarize, photoelectrodes such as TiO 2 /CdS/Cd 1−x Mn x Se have successfully been prepared using SILAR. The thickness of the Cd 1−x Mn x Se film affected the optical and photovoltaic properties of QDSSCs. The J-V curves show that the conversion efficiency is improved due to the optimized thickness at three cycles and Cd 1−x Mn x Se QDs. In addition, this result is also confirmed by the shift of absorption toward to the visible region, increasing lifetimes, and reducing charge recombination at the polyelectrolyte/counter electrode, TiO 2 /Cd 0.8 Mn 0.2 Se/polyelectrolyte interfaces, and diffusion resistance in TiO 2 films. As a result, QDSSCs exhibited a high conversion efficiency of 3.8%.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
HT and DP conceived and planned the experiments and carried out the experiments, contributed to sample preparation, took the lead in writing the manuscript. They also performed the experiments about the structural materials and contributed to the analysis of the new results of the manuscript.