N型表面钝化晶体硅太阳电池Al2O3_SiNx减反射叠层的优化研究

Vol. 32, No. 9 Journal of Semiconductors September 2011 Optimization of Al2O3/SiNx stacked antireflection structures for N-type surface- passivated crystalline silicon solar cells� WuDawei(吴大卫)Ž, Jia Rui(贾锐), DingWuchang(丁武昌), Chen Chen(陈晨), Wu Deqi(武德起), Chen Wei(陈伟), Li Haofeng(李昊峰), Yue Huihui(岳会会), and Liu Xinyu(刘新宇) Institute of Microelectronics, Chinese Academy of Sciences, Beijing 100029, China Abstract: In the case of N-type solar cells, the anti-reflection property, as one of the important factors to further improve the energy-conversion efficiency, has been optimized using a stacked Al2O3/SiNx layer. The effect of SiNx layer thickness on the surface reflection property was systematically studied in terms of both experimental and theoretical measurement. In the stacked Al2O3/SiNx layers, results demonstrated that the surface reflection property can be effectively optimized by adding a SiNx layer, leading to the improvement in the final photovoltaic characteristic of the N-type solar cells. Key words: antireflection coatings; aluminum oxide; silicon nitride; simulation; solar cells DOI: 10.1088/1674-4926/32/9/094008 EEACC: 2520 1. Introduction The N-type based solar cell, as one of the promising alter- natives to conventional solar cells, has attracted extensive re- search attention due to its several advantages. Compared with p-type silicon, n-type silicon is more tolerant to common metal contaminationsŒ1. Also, n-type silicon shows no light-induced degradation, which is known as a defect for p-type CZ sili- con due to boron–oxygen pairsŒ2. With conversion efficiencies above 23%, the potential of n-type silicon has been demon- strated at the device level in recent yearsŒ3; 4. In the case of N-type based solar cells, a boron diffused (p-type) emitter was commonly formed as the pn junction. In such a solar cell structure, the antireflection coating as well as the surface passivation is of crucial importance to optimize the final photovoltaic performance of a p-type emitter. However, common dielectric layers for the passivation of an n+ emitter, for example SiNx , SiO2, do not show the same performance as the p+ emitterŒ5. Thin films of aluminum oxide (Al2O3/ grown by atomic layer deposition (ALD) are known to provide an ex- cellent level of surface passivation on lowlyŒ6 and highlyŒ7 boron-doped p+ surfaces. Although the single Al2O3 layer can passivate the p-type emitter, the growth of such a layer suf- fers from the slow growth rate using the ALD process, and the anti-reflection property is not satisfactory. Hence, an extra an- tireflection coating (ARC) layer, e.g. SiNx , has to be added to form the stacked surface passivation. The surface passivation property of an Al2O3/SiNx stacked layer has been studied systematically thus farŒ8. Nev- ertheless, for the Al2O3/SiNx stacked layer, the effect of SiNx thickness on the surface anti-reflection property is still in its infancy. In this paper, we report the optimized SiNx thickness value, and verify the simulation results in experiments based on the theoretical computational simulation analysis. The ef- fect of emitter passivation and the theoretical highest efficiency of p+nn+ type solar cells extracted from Afors-Het simulation were also studied. 2. Theory The optical matrix approach is usually employed for cal- culation of the reflection coefficient. The main idea of this method matches the E- and H-fields of the incident light on the interfaces of the two layer optical coatings. The matrix re- lation defining the two layer antireflection coating problem is given in Ref. [9],� B C � D ( qY rD1 � cos �r .i sin �r /=�r i�r sin �r cos �r �)� 1 �m � ; (1) where �r D .2�Nd cos �r /=� and �m is substrate admittance. The tilted optical admittance � is given by �p D Nycos � ; p-waves; (2) �s D Ny cos �; s-waves: (3) B and C are total electric and magnetic field amplitudes of the light propagating in the medium. Thus the optical admit- tance is given by the ratio Y D C B : (4) The following relations give reflectance, transmittance and ab- sorbance, respectively. * Project supported by the State KeyDevelopment Program for Basic Research of China (Nos. 2006CB604904, 2009CB939703), the National Natural Science Foundation of China (Nos. 60706023, 60676001, 90401002, 90607022), the Chinese Academy of Sciences (No. YZ0635), and the Chinese Academy of Solar Energy Action Plan. Ž Corresponding author. Email: wudawei@ime.ac.cn Received 2 April 2011, revised manuscript received 28 April 2011 c 2011 Chinese Institute of Electronics 094008-1 J. Semicond. 2011, 32(9) Wu Dawei et al. R D � �0 � Y �0 C Y �� �0 � Y �0 C Y �� ; (5) T D 4�0Re.�m/ .�0B C C/.�0B C C/� ; (6) A D 1 � T �R D .1 �R/ � 1 � Re.�m/ Re.BC �/ � : (7) Thus, each layer is represented by a 2 � 2 matrix M , of the form M r D � cos �r .i sin �r /=�r i�r sin �r cos �r � : (8) For solar cells, it is important to have aminimum reflection over the entire visible spectrum (300–1100 nm). The cell per- formance is influenced by other parameters, such as the photon flux F (�/. Since the reflection coefficient needs to be mini- mized where F (�/ have their maximum values, the weighted reflectance Rw is calculated from Ref. [10], Rw D R �1 �2 Fi.�/R.�/d�R �1 �2 Fi.�/d� : (9) The F (�/ value is extracted according to the AM 1.5 solar spectrum, and �1 D 300 nm, �2 D 1100 nm. 3. Experiments In experiment, a 300-�m thick boron-doped p-type (100) FZ-Si wafer with a resistivity of 3.5 �cm was used, in which both sides were polished and used to eliminate the effects of surface roughness. All wafers were thoroughly cleaned us- ing 5% diluted hydrofluoric acid to remove native oxide. The Al2O3 films were deposited by thermal ALD in an Beneq TSF200 reactor at a substrate temperature of 200 �C. The cy- cle times were � 5 s and the growth-per-cycle was 1.0 s. The cycles were repeated until the target film thickness 10 nm was finished. Then the SiNx layers with different layer thicknesses were deposited by an Unaxis Plasma Therm790+ PECVD sys- tem at a substrate temperature of 350 �C. The flow rates of silane and ammonia gases to deposit SiNx films were 8.5 and 200 sccm, respectively. The thicknesses of SiNx were changed from 35 nm to 95 nm, with a deviation of 15 nm. The deposition rate of SiNx films was 30 nm/min. The refractive indexes of Al2O3 and SiNx weremeasured to be 1.64 and 1.96 using spec- troscopic ellipsometry, respectively. The reflectance of wafers coated with Al2O3/SiNx stacked layers wasmeasured for com- parison. The reflectancewas characterized at room temperature by a measurement system using a xenon lamp as a light source. As shown in Fig. 1, the surface reflection spectra of the single Al2O3 and SiNx layers was demonstrated for a p-type emitter via aMatlab simulation. The theory utilized in this sim- ulation is elaborated in the theory section. In the simulation, given by Ref. [11], the refractive index of Al2O3 and SiNx is adopted as 1.64 and 1.96 at 2.0 eV, respectively. To achieve the lowest surface reflectance index, the simulation process veri- fied that the optimized thickness of the Al2O3 single layer and Fig. 1. Surface reflection spectra of the optimized Al2O3 single layer (square line), and the reflection spectra of the optimized SiNx single layer for a p-type emitter via a Matlab simulation. the SiNx single layer is determined to be 95 nm and 75 nm, respectively. As shown in Fig. 1, the reflectance of the SiNx single layer seems to be slightly lower than that of the Al2O3 layer within the whole wavelength range from 300 nm to 1100 nm. The weighted reflectance of the SiNx and Al2O3 layers is 12.27% and 10.52%, respectively. Compared with the re- flectance value of the SiNx layer, the deviation was increased to this maximum around the wavelength of 600 nm (the high- est density of photon flux) within the incident sunlight spectra, resulting in the lower weighted reflectance value of the single Al2O3 layer. As shown in Fig. 1, the reflection spectra result shows that within the whole incident spectra, the reflectance value of the Al2O3 single layer seems to be higher than that of the SiNx single layer, leading to a higher weighted reflectance value. It could be inferred that only using the Al2O3 single layer can passivate the p-type emitter surface. However, the anti- reflection property is not optimized. Taking account of the anti- reflection property as well as the surface passivation, the SiNx layer has to be added to form the stacked Al2O3/SiNx layers in order to achieve a better anti-reflection property for a p-type emitter. Furthermore, as the SiNx antireflection layer can be grown by the PECVD process with a high yielding rate, the incorporation of the SiNx layer enables the shortened yielding process for the stacked Al2O3/SiNx layers. And, the thermal stability of the ultrathin Al2O3 is significantly improved by depositing a capping layer of SiNx onto the Al2O3, which is probably due to the very high (10–15 at.%) hydrogen content in the PECVD-deposited SiNx film. Therefore, it is worthwhile further analyzing the property of the stacked Al2O3/SiNx lay- ers as an effective anti-reflection coating for a p-type emitter. In Fig. 2, we examined the anti-reflection property of the stacked Al2O3/SiNx layers with different SiNx layers thick- nesses by fixing the Al2O3 layer thickness as 10 nm. As shown in Fig. 2, with an increase in the SiNx layer thickness from 35 to 95 nm, the weighted reflectance of the stacked Al2O3/SiNx layers was first decreased and then increased. The best result with the lowest weighted reflectance appears at the thickness of 65 nm of SiNx capping layer, with a weighted reflectance of 10.50%.And the reflectance of the 65 nmSiNx capping layer is 9.39% in experimental measurement. This dependence of the 094008-2 J. Semicond. 2011, 32(9) Wu Dawei et al. Fig. 2. Theoretical and experimental results of the weighted re- flectance as a function of thickness of silicon nitride in an Al2O3/SiNx stacked layers. Fig. 3. Optimized surface reflection property using the stacked Al2O3/SiNx layers both experimentally and theoretically. thickness in the stacked Al2O3/SiNx layers can be explained by the weighted reflectance theory. When the thickness of the SiNx layer increased from 35 to 65 nm, the minimum value of the reflectance spectra shifted within the weighted region, re- sulting in a decrease in the weighted reflectance value. On the other hand, when the thickness of the layer further increased above 65 nm, the minimum value shifted out of the weighted region, leading to an increase in the weighted reflectance. The measured curve, as shown in Fig. 2, shows very good agree- ment between the theoretical and experimental results. The dif- ference between the two main attributes to the refractive index of the silicon substrate used in the simulation. And the refrac- tive index of Al2O3 may be increasing for the higher temper- ature during the deposition process of SiNx , and also the sur- face coverage would be improved by the increase in thickness of SiNx layer. Figure 3 shows the optimized surface reflection property using the stacked Al2O3/SiNx layers both experimentally and theoretically. The two curves fit well within an interesting wavelength region. The difference between the two curves within the short-wavelength response was mainly due to the different refractive indexes of silicon substrate used in simula- tion and in experiment. The intensity of the xenon lamp used Fig. 4.C–V curves of the different antireflection and passivation con- trolled solar cells under standard test conditions (AM 1.5 irradiation, T D 300 K and P D 100 mW/cm2/. in the measurement system was not strong in such a spectrum region. In the long wavelength region from 700 to 1050 nm, the incident light could be reflected by the rear surface of the Al2O3/SiNx layers, and then collected by the CCD. However, this portion of reflected light was not taken into consideration during the simulation process, which is attributed to the de- viation between the simulation and experimental results. We added a simple interface reflection film and achieved a more consistent result. Next we used Afors-Het software to study the effect of the antireflection and passivation property on a p+nn+ silicon solar cell. Figure 4 plots the current–voltage (C–V ) curves of the different antireflection and passivation controlled solar cells under standard test conditions (AM 1.5 irradiation, T D 300 K and P D 100 mW/cm2/. As shown in Fig. 4, compared to the cell using the same antireflection data from the previous sim- ulation with a higher SRV (surface recombination velocity) of 5000 cm/s, the cell with a SRV of 5 cm/s cell exhibits a 3.2% increase in Voc due to the better passivation. With a fixed SRV of 5 cm/s, the ideal reflectance (i.e. surface reflectance equals 0) cell exhibits a 10.14% higher Jsc than the cell using the pre- vious reflection data. And the ideal cell presented here can get a high efficiency of 24.85%. We can infer that the antireflec- tion is more important than passivation in a p+nn+ solar cell. Through other methods like texturing, a better solar cell per- formance can be expected. 4. Conclusion The anti-reflection property of N-type solar cells has been optimized using a stacked Al2O3/SiNx layer consisting of 10 nm Al2O3 and 65 nm SiNx . This further improves the energy- conversion efficiency. The effect of SiNx layer thickness on the surface reflection property was systematically studied both theoretically and experimentally. The optimized reflectance of theoretical and experimental measurement is 10.50% and 9.39%, respectively. In the stacked Al2O3/SiNx layers, the re- sults demonstrated that the surface reflection property can be effectively optimized by adding a SiNx layer, which would im- prove the final photovoltaic characteristic of N-type solar cells with the conversion efficiency as high as 24.85%. 094008-3 J. Semicond. 2011, 32(9) Wu Dawei et al. References [1] Macdonald D, Geerligs L J. Recombination activity of intersti- tial iron and other transition metal point defects in p- and n-type crystalline silicon. Appl Phys Lett, 2004, 85: 4061 [2] Glunz SW, Rein S, Lee J Y, et al. Minority carrier lifetime degra- dation in boron-doped Czochralski silicon. J Appl Phys, 2001, 90: 2397 [3] Benick J, Hoex B, van de Sanden M C M, et al. High efficiency n-type Si solar cells on Al2O3-passivated boron emitters. Appl Phys Lett, 2008, 92: 253504 [4] Taguchi M, Tsunomura Y, Inoue H, et al. High-efficiency HIT solar cell on thin (< 100 �m) silicon wafer. 24th European Pho- tovoltaic Solar Energy Conference Hamburg, Germany, 2009: 1690 [5] Altermatt P P, Plagwitz H, Bock R, et al. The surface recom- bination velocity at boron-doped emitters: comparison between various passivation techniques. 21st European Photovoltaic Solar Energy Conference Dresden, Germany, 2006: 647 [6] Hoex B, Heil S B S, Langereis E, et al. Ultralow surface recom- bination of c-Si substrates passivated by plasma-assisted atomic layer deposited Al2O3. Appl Phys Lett, 2006, 89: 042112 [7] Hoex B, Schmidt J, Bock R, et al. Excellent passivation of highly doped p-type Si surfaces by the negative-charge-dielectric Al2O3. Appl Phys Lett, 2007, 91: 112107 [8] Dingemans G, Engelhart P, Seguin R, et al. Stability of Al2O3 and Al2O3/a-SiNx :H stacks for surface passivation of crystalline silicon. J Appl Phys, 2009, 106: 114907 [9] Macleod H A. Thin-film optical filters. Bristol: Adam Hilger, 1986 [10] Bouhafs D, Moussi A, Chikouche A, et al. Design and simulation of antireflection coating systems for optoelectronic devices: ap- plication to silicon solar cells. Solar Energy Materials and Solar Cells, 1998, 52: 79 [11] Dingemans G, Kessels W M M. Recent progress in the devel- opment and understanding of silicon surface passivation by alu- minum oxide for photovoltaics. 25th European Photovoltaic So- lar Energy Conference Valencia, Spain, 2010: 1083 094008-4