Open Access
Add to BibSonomyAbstract
Radial junction nanopillar Si solar cells are interesting for cost effective efficiency improvement. Here, we address a convenient top-down fabrication of Si nanopillar solar cells using spin-on doping and rapid thermal annealing (RTA) for conformal PN junction formation. Broadband suppressed reflection as low as an average of 5% in the 300-1100 nm wavelength range and un-optimized cell efficiency of 7.3% are achieved. The solar cell performance can be improved by optimization of spin-on-doping and suitable surface passivation. Overall, the all RTA processed, spin-on doped nanopillar radial junction solar cell shows a very promising route for low cost and high efficiency thin film solar cell perspectives.
© 2017 Optical Society of America
1. Introduction
For useful use of solar photovoltaic energy, it is important to increase sun light to electricity conversion efficiency in a cost effective manner. In this regard, some important aspect include, the use of available light, efficient photon to electron/hole conversion, better collection of generated charge carriers and lower thermal budget of processing can boost possibilities. In this regard, Radial PN junction nanopillar array solar cells have several advantages, like antireflective property, broadband light trapping and absorption, better carrier collection etc. compared to planar bulk solar cells [1–4]. These properties can improve solar cell’s photon to current conversion efficiency in lesser use of active material or in a cost effective manner. There are several reports where radial junction Si nanopillar or nanowires are used to maximize the use of available light and better photo generated carrier collection [5,6]. However, processing ease for this kind of cell and achieved efficiency require improvement. From processing perspective, there are different doping methods exist for the junction formation, including chemical vapor deposition (CVD), solid source doping (SSD), monolayer doping (MLD) and also spin-on doping (SOD) [7–9]. Normally all these processes need relatively high temperature, more than 800 o C for dopant diffusion and annealing (activation) and are carried out in conventional furnaces. High temperatures and long annealing times can affect doping profiles, hence PN junction location; additionally they can introduce defects. On the other hand, rapid thermal annealing (RTA) process is a very fast process and anneal time of 60 seconds and less are common, and can reduce thermal budget and processing time. Thus, RTA is extremely suitable for shallow junction formation, which is desirable in radial junction formation without introduction of defects [10–12]. Combination of SOD and RTA is also reported for doping in nanopillar-based devices [13,14]. In addition, to harvest all benefits of radial junction solar cell it is important to appropriately control the doping profile/ junction and better design of nanopillar geometry. In the case of thin film solar cell, apart from reflection loss, another major loss mechanism is incomplete absorption and that put a restriction on the minimum thickness of the cell. Appropriate design of nanopillar architecture in thin film solar cell can enhance light absorption by increasing light path and confining light in nanopillar and enable realization of efficient solar cell.
In this paper, we report on an all RTA processed, top down fabricated radial junction Si nanopillar solar cells and evaluate their antireflective and photovoltaic properties. Full vectorial electromagnetic (EM) simulation is performed to obtain optimal nanopillar array parameters for low reflection and higher absorption. The radial PN junction is formed by SOD with RTA diffusion. RTA also used for the Al back contact formation. The doping profile is characterized by secondary ion mass spectrometer (SIMS) measurement for different temperatures. For comparison, here we have fabricated SOD diffused and standard planar PN junction cells along with SOD radial junction cells and measured their current voltage characteristics. The antireflection property is verified by total reflection measurement and current- voltage measurement with solar AM 1.5 (1 Sun) are performed to characterize the solar cell.
2. Device design and simulation
For better utilization of antireflective properties and absorption enhancement with nanopillar solar cell, it is important to properly choose the pillar parameters, in particular, shape, diameter and period of the array [15–17]. Full vectorial electromagnetic (EM) simulation is very useful for this purpose to optimize the parameters for low reflection and better absorption. Here we have used commercial Lumerical Finite Difference Time Domain (FDTD) tool for the EM simulation [18]. For simulation, we used hexagonal unit cell of Si pillar array with periodic boundary condition in X and Y directions and perfectly match layer (PML, absorbing) above and below the reflection and transmission monitors respectively in Z direction. We have calculated solar AM1.5 weighted average absorption (WAA) in 300 nm to 1100 nm wavelength range, here after called absorption, in the Si nanopillar array plus 1500 nm in the Si substrate. We have used simulation interval of 1nm. AM 1.5 solar spectral irradiance data is sampled for 300- 1100 nm wavelength range with accuracy of 1nm in a .txt file format and the data is incorporated to calculate the WAA using Lumerical FDTD script. The WAA or simply absorption is calculated using Eq. (1), where IAM1.5 (λ) is the solar AM 1.5 reference irradiance (Fig. 1.b) and A(λ) is wavelength dependent absorption in the structure under test. A(λ) is calculated in percentage with respect to reflection, R(λ) and transmission, T(λ) data of the reflection monitor MR and transmission monitor MT respectively [Fig. 1(a)] and given by Eq. (2),
Figure 1(a) shows simulation data of 2D contour plot of absorption of AM1.5 solar radiation in the 300 nm to 1100 nm wavelength range with respect to nanopillar period and diameter for a 4500 nm total active thickness of solar cell. Here, the transmission monitor position is 1500 nm into the substrate, which means that the absorption is taken into account for 4500 nm thickness of Si including the pillar length of 3000 nm in our simulation. These parameters are chosen considering application in thin film solar cell with lesser use of material and further quantified with ref. to Fig. 1(b). Referring to the plot in Fig. 1(a), on the x-axis, we plot the spacing between nanopillars (equivalently, different periods) and the y-axis shows the average diameter of the nanopillar. The height of the nanopillars is fixed as at 3000 nm for practical reason, such as for ease of fabrication with decent aspect ratio between height and diameter of the pillars. The average diameter is the average of the top and bottom diameters of the truncated conical nanopillar and is varied between 200 to 600 nm. Here for all simulated average diameter of the pillars, the slant angle of the pillar is kept close to 2°. The period is varied from 500 nm to 1200 nm. As can be seen from the plot, the absorption can vary from 0.3 to 0.65 with different combination of pillar average diameter and period. Higher absorption implies that for those combinations of diameter and period, broadband suppression in reflection is achieved, and the light is well coupled to the nanopillars. Further, the highest absorption zone, as indicated by red colored areas follows a near linear relationship between pillar diameter and period and can server as worthy design rule. To understand the variation ofabsorption with total active thickness/height of the nanopillar solar cell, we have simulated WAA by varying the thickness of the substrate while keeping pillars height fixed at 1500 nm; and alternately by varying pillars height while keeping the thickness of the substrate fixed at 1500 nm. The result is shown in Fig. 1(b). In both cases the WAA followed similar trend with thickness/height increase, while increase in pillar height compared to increase in thickness gives little higher WAA. In both cases up to total thickness/height of 4500 nm, there is sharp increase in WAA beyond that the increase is rather sluggish. As further described in detail in next section our route is to use SiO2 colloidal particles as mask for nanopillars fabrication with some definite selection of diameters, like 500 nm and 1000nm. Therefore, we have selected viable 1000 nm diameter particles so that hexagonal close pack monolayer coverage will leads to period of 1000 nm and to restrict average pillar diameter to 575-600 nm for highest absorption with respect to total theoretical device thickness of 4500 nm.
Fig. 1 (a) 2D Contour plot of AM1.5 weighted average absorption for different average diameter and period combinations. The inset shows the reflection (MR) and transmission monitor (MT) positions; (b) Weighted average absorption (WAA) for different pillar height (or for different substrate thickness) at fixed substrate thickness (or at fixed pillar height).
3. Experimental procedures
To compare and evaluate performances we have prepared three different sets of samples: (i) standard planar Si cell, where we have used solar cell grade c-Si PN junction, (ii) RTA diffused PN junction planar cell and (iii) RTA diffused radial PN junction nanopillar cell. In the latter two types the PN junction is formed by phosphorus (P) diffusion in to boron (B) doped p-type Si substrate (a 1x1016 cm−3) using a spin-on phosphorous dopant and RTA. The schematic of the radial junction solar cell is shown in Fig. 2(a). The as formed PN junction can be viewed as being conformal. The geometry of the PN junction in the nanopillar is radial and the bottom junction with the substrate can also collect carriers generated in the substrate.
Fig. 2 (a) Schematic illustration of radial junction solar cell with top and bottom contacts; (b) 25° tilted scanning electron microscope (SEM) image of fabricated radial junction Si pillar with ITO contact.
The top and bottom electrode processing is same for all three samples. The bottom electrode on p side of the substrate is deposited by e-beam evaporation of Al and RTA treated at 4500 C in N2 atmosphere for 45 s. The top ITO contact is sputtered deposited on the samples by magnetron sputtering. The Si nanopillar arrays for radial junction solar cell are fabricated by colloidal lithography and plasma dry etching processes [19]. We used colloidal lithography with 1-micron SiO2 particle to form a hexagonally close packed monolayer on the Si surface. The diameter of SiO2 particles are subsequently size reduced to 700 nm by reactive ion etching to serve as the masks for etching Si nanopillars. The Si pillars are realized by inductive coupled plasma reactive ion etching (ICP-RIE) using a pseudo-Bosch process. The shape of the pillars are made truncated cone like (tapered at the top) to enhance the antireflection property of the pillar array. Here, the platen power of the ICP is optimized to make the pillar profiles tapered.
For phosphorus (P) diffusion, we use a spin-on dopant (SOD) source, Filmtronics- P500 series diffusant [20]. Before the diffusion process, it is kept for 15 hours at room temperature. The SOD solution is spin coated onto pre-cleaned surface of the samples (both planar and pillar samples) with the following recipe: first at 1000 rpm with 250 rpm/sec for 10s followed by at 3000 rpm with 1000 rpm/sec for 30s to get an evenly distributed film of the dopant solution. Then the SOD coated samples are prebaked on a hot plate at 200 °C for 20 minutes. Three different RTA diffusion (anneal) temperatures are used to prepare three batch of samples, namely, sample A at 950 °C; sample B at 1050 °C ; and sample C at 1150 °C. Other common parameters in RTA diffusion process are, time 60 second; environment: 4 sslm N2; ramp up: 55 °C/s; ramp down: 55 °C/s. The dopant oxide formed during the SOD process is removed by a 2 minute 50% HF dip and rinsed in DI water.
The representative 25° tilted SEM image of the pillar arrays is shown in Fig. 2(b). The fabricated pillars arrays have pillars of truncated cone shape with top and bottom diameters approximately 500 nm and 700 nm respectively, a height of 3000 nm and a period of 1000 nm. The slant angle (ϴ, in Fig. 2(b)) of the truncated conical pillar is approximately 2°. Comparing with the simulation data in Fig. 1a, the fabricated nanopillar arrays having an average pillar diameter of 600 nm and period of 1000 nm should have theoretical absorption of 0.60 to 0.65, which is in the maximum absorption range for 4500 nm total Si cell described earlier. However, in our fabricated nanopillar solar cells the substrate is much thicker than the simulated one so there is will be increase in overall absorption compared the simulated one, as also discussed with reference to Fig. 1(b). Nevertheless, the fabricated pillar parameters are appropriate from light absorption point of view.
4. Result and discussion
SIMS measurements
For profiling the phosphorus (P) dopant diffusion in RTA annealed SOD doped samples, Secondary Ion Mass Spectrometry (SIMS) measurements are performed with CAMECA, dynamic SIMS instrument, IMS4F. The dopant (P) profiles are investigated in planar samples and plotted with respect to depth from the surface for samples A, B and C (annealed at 950 °C, 1050 °C and 1150 °C respectively) as shown in Fig. 3(a). As can be seen in Fig. 3(a), the ‘P’ diffusion depth increases with temperature and are around are 70 nm, 150 nm and 500 nm in samples A, B and C respectively. Based on the above data, we have chosen the parameters for sample B (which is annealed at 1050 oC) for actual radial junction formation on our nanopillar solar cell fabrication. As the smallest dimension of the pillar is ~500 nm (top diameter), so P diffusion of 150 nm is a good choice for fabricating the radial PN junction in the nanopillars.
Fig. 3 (a) Secondary Ion Mass Spectrometry (SIMS) measurements of P diffusion profile for three different temperatures on planar cell; (b) Measured and simulated total reflectivity for Si pillar and measured total reflectivity for planar samples, with and without ITO coating
Total reflectivity measurements
One of the main motivations for the nanopillar solar cell is to suppress surface reflection and thereby increase the light absorption over a wide wavelength range where Si has significant absorption in the solar spectrum. We have performed total reflectivity measurements with a UV-Visible spectrometer fitted with an integrating sphere. Here we have measured planar silicon as a reference for other samples and the data are shown in Fig. 3(b). Compared to planar Silicon, which has well above 30% reflection for the measured wavelength range of 300 nm to 1100 nm, the planar Si sample with ITO has reflectivity close to average 20%. This reduction can be attributed to the nanoscale rough morphology of the sputtered ITO film and the refractive index step it creates between Si and air. The measurement on bare Si pillar array shows a large overall reduction in reflectivity compared to planar Si and ITO coated planar Si. The reduction is mainly due to better impedance matching of input light with the nanostructured surface. Further, the truncated conical structure of the nanopillar enhances antireflection in broader wavelength range. The typical Fabry-Perot oscillation in reflection is not distinct due to the conical shape of the pillar [21]. Overall, we have a very flat reflection spectrum with an average around 5% reflectivity for the wavelength range of 300 to 1100 nm and slight higher reflection (between 7 and 10%) in 300-400nm wavelength range. This is possibly due to the chosen period and diameter of the pillar arrays that is little more favorable for larger wavelength side of the spectrum as per antireflection is concerned. As seen in the measured spectrum, the ITO deposited pillar has further reduced reflectivity particularly in the lower wavelength range. Otherwise, it follows the same trends as the bare Si pillars. The suppressed reflection can be attributed to the effect of a lower index coating and together with possible absorption in ITO at the shorter wavelengths. The low broadband total reflection gives a good indicator for the nanopillar geometry for the solar cell with more available photon for electron hole pair generation.
Solar cell characterization
Figure 4 shows the measured current density vs voltage characteristics for the fabricated solar cells under AM1.5 normal solar illumination of 100 mw/cm2 at room temperature. As can be seen in, the solar cell with standard planar PN junction has short circuit current density (Jsc) 17.5 mA/cm2 and open circuit voltage (Voc) 0.607 V respectively. This cell returns an approximate efficiency of 6.28%. The calculated efficiency is much lower compared tocommercial c-Si solar cells. This can be due to few separate reasons, firstly the contacts are not optimized, there is no surface passivation as such and back surface field (BSF) and antireflection coatings are not used here. There is also appreciable series resistance due to un-optimized contacts. On the other hand, the planar cell with the PN junction made by SOD and RTA has short circuit current density (Jsc) 16.25 mA/cm2 and open circuit voltage (Voc) 0.55 V, respectively. This cell returns an approximate efficiency of 4.73%. The Voc is further dropped compared to standard planar cell. This is possibly due to defect generated during diffusion process. In addition, the unoptimized PN junction can be reason for lower Jsc [22]. However, there are possibilities for the improvements by reducing diffusion related defect around the junction and surface passivation [23].
Fig. 4 Current –voltage measurement under solar AM 1.5 illumination conditions for SOD planar cell, standard planar cell and SOD radial junction nanopillar cell
Compared to the previous two planar cells, the current voltage characteristics of the nanopillar solar AM1.5 illumination condition shows that the Jsc has increased from 16.25 to 28 mA, clearly indicating the effect of antireflection due to nanopillar architecture where by more photons are absorbed in the cell. However, further increase in Jsc is anticipated by improving photo generated carriers collection probability for nanopillar cell with reduced surface or defect related recombination. On the other hand, the Voc decreases to 0.52 volt and the overall efficiency slightly increases to 7.29%. The etched structure in the form of nanopillar has contributed in increased the photocurrent possibly due to reduced reflection or due to better carrier collection with the radial junction. However, it has adversely effected the Voc. Reduction in Voc has to do with more open surface due to higher surface to volume ratio of the nanopillar geometry which increases number of recombination sites or surface states and also possibly due to etch related defects [24,25]. The fill factor (FF) is also significantly low, due to possible defect related recombination and due to parasitic resistance (particularly due to higher series resistance) effect as obvious from the shape of I-V curve. We further have observed that both shunt and series resistances have effected all three solar cell I-V curves, however, in the nanopillar cell the effect is more pronounced. We believe that nanopillar radial junction cell can be significantly improved with proper surface passivation while addressing defect related issues and resistive effect [26]. Also spectrally resolved I-V measurement would be very useful to get better insight on the solar cell performance and to optimize nanopillar design and junction. In Table 1, we summarized the solar cell parameters for three of our fabricated solar cells.
Table 1. Solar cell efficiency and J-V parameters of three different fabricated Si solar cells
5. Conclusion
In conclusion, a top-down fabrication of Si nanopillar solar cells using a spin-on doping and rapid thermal annealing for conformal PN junction formation is investigated. The fabricated radial junction Si nanopillar array solar cells are compared with planar cells in terms of anti-reflection and solar cell characteristics. From an optical point of view, it clearly shows reduced total reflection up to average 5% on 300-1100nm wavelength range. The nanopillar cell returns 7.29% efficiency. Efficiency improvement is clearly seen on the radial junction cell, although it is not yet that at the level of commercial crystalline Si cell. We anticipate further improvements in efficiency by better front and rear electrical contact, reduction in series, addressing defect related issues and with optimized radial doping of the pillar. A combination of Nanopillar architecture of solar cell with RTA treated SOD doped process can be anticipated as a route for efficient and cost effective solution for photovoltaic power generation with thin film Si solar cell.
Funding
Swedish Research Council (grant 349-2007-8664) ; Swedish Energy Agency (grant 42028-1).
Acknowledgements
The authors acknowledge the support from the Linné center for advanced optics and photonics (ADOPT) funded by the Swedish Research Council.
References and links
1. B. M. Kayes, H. A. Atwater, and N. S. Lewis, “Comparison of the device physics principles of planar and radial p-n junction nanorod solar cells,” J. Appl. Phys. 97(11), 114302 (2005). [CrossRef]
2. M. D. Kelzenberg, S. W. Boettcher, J. A. Petykiewicz, D. B. Turner-Evans, M. C. Putnam, E. L. Warren, J. M. Spurgeon, R. M. Briggs, N. S. Lewis, and H. A. Atwater, “Enhanced absorption and carrier collection in Si wire arrays for photovoltaic applications,” Nat. Mater. 9(3), 239–244 (2010). [PubMed]
3. K. R. Catchpole, S. Mokkapati, and F. J. Beck, “Comparing nanowire, multijunction, and single junction solar cells in the presence of light trapping,” J. Appl. Phys. 109(8), 084519 (2011). [CrossRef]
4. L. Hu and G. Chen, “Analysis of optical absorption in silicon nanowire arrays for photovoltaic applications,” Nano Lett. 7(11), 3249–3252 (2007). [CrossRef] [PubMed]
5. E. C. Garnett, M. L. Brongersma, Y. Cui, and M. D. McGehee, “Nanowire solar cells,” Annu. Rev. Mater. Res. 41(1), 269–295 (2011). [CrossRef]
6. E. Garnett and P. Yang, “Light trapping in silicon nanowire solar cells,” Nano Lett. 10(3), 1082–1087 (2010). [CrossRef] [PubMed]
7. R. Elbersen, W. Vijselaar, R. M. Tiggelaar, H. Gardeniers, and J. Huskens, “Fabrication and doping methods for silicon nano and micropillar Arrays for solar cell applications: A Review,” Adv. Mater. 27(43), 6781–6796 (2015). [CrossRef] [PubMed]
8. J. F. Nijs, J. Szlufcik, J. Poortmans, S. Sivoththaman, and R. P. Mertens, “Advanced manufacturing concepts for crystalline silicon Solar Cells,” IEEE Trans. Electron Dev. 46(10), 1948–1969 (1999). [CrossRef]
9. J. C. Ho, R. Yerushalmi, Z. A. Jacobson, Z. Fan, R. L. Alley, and A. Javey, “Controlled nanoscale doping of semiconductors via molecular monolayers,” Nat. Mater. 7(1), 62–67 (2008). [CrossRef] [PubMed]
10. B. Hartiti, A. Slaoui, J. C. Muller, R. Stuck, and P. Siffert, “Phosphorus diffusion into silicon from a spin-on source using rapid thermal processing,” J. Appl. Phys. 71(11), 5474–5478 (1992). [CrossRef]
11. A. Usami, M. Ando, M. Tsunekane, and T. Wada, “Shallow junction formation on silicon by rapid thermal diffusion of impurities from a spin-on source,” IEEE Trans. Electron Dev. 39(1), 105–110 (1992). [CrossRef]
12. Z. T. Zhu, E. Menard, K. Hurley, R. G. Nuzzo, and J. A. Rogers, “Spin on dopants for high-performance single-crystal silicon transistors on flexible plastic substrates,” Appl. Phys. Lett. 86(13), 133507 (2005). [CrossRef]
13. C. Lin and M. L. Povinelli, “Optical absorption enhancement in silicon nanowire arrays with a large lattice constant for photovoltaic applications,” Opt. Express 17(22), 19371–19381 (2009). [CrossRef] [PubMed]
14. D. Y. Lee, H. Kim, H. M. Li, A. R. Jang, Y. D. Lim, S. N. Cha, Y. J. Park, D. J. Kang, and W. J. Yoo, “Hybrid energy harvester based on nanopillar solar cells and PVDF nanogenerator,” Nanotechnology 24(17), 175402 (2013). [CrossRef] [PubMed]
15. J. S. Sadhu, H. Tian, T. Spila, J. Kim, B. Azeredo, P. Ferreira, and S. Sinha, “Controllable doping and wrap-around contacts to electrolessly etched silicon nanowire arrays,” Nanotechnology 25(37), 375701 (2014). [CrossRef] [PubMed]
16. B. Wang and P. W. Leu, “Enhanced absorption in silicon nanocone arrays for photovoltaics,” Nanotechnology 23(19), 194003 (2012). [CrossRef] [PubMed]
17. S. A. Boden and D. M. Bagnall, “Tunable reflection minima of nanostructured antireflective surfaces,” Appl. Phys. Lett. 93(13), 133108 (2008). [CrossRef]
18. https://www.lumerical.com/tcad-products/fdtd/
19. B. D. Choudhury, R. Casquel, M. J. Bañuls, F. J. Sanza, M. F. Laguna, M. Holgado, R. Puchades, A. Maquieira, C. A. Barrios, and S. Anand, “Silicon nanopillar arrays with SiO2 overlayer for biosensing application,” Opt. Mater. Express 4(7), 1345 (2014). [CrossRef]
20. http://www.filmtronics.com/capabilities/products/diffusants
21. Y. Lu and A. Lal, “High-Efficiency ordered silicon nano-conical-frustum array solar cells by self-powered parallel electron lithography,” Nano Lett. 10(11), 4651–4656 (2010). [CrossRef] [PubMed]
22. O. Breitenstein, “Understanding the current-voltage characteristics of industrial crystalline silicon solar cells by considering inhomogeneous current distributions,” Opto-Electro. Rev. 21(3), 259–282 (2013).
23. Y. Dan, K. Seo, K. Takei, J. H. Meza, A. Javey, and K. B. Crozier, “Dramatic reduction of surface recombination by in situ surface passivation of silicon nanowires,” Nano Lett. 11(6), 2527–2532 (2011). [CrossRef] [PubMed]
24. M. M. Adachi, M. P. Anantram, and K. S. Karim, “Core-shell silicon nanowire solar cells,” Sci. Rep. 3, 1546 (2013). [CrossRef] [PubMed]
25. D. Kumar, S. K. Srivastava, P. K. Singh, M. Husain, and V. Kumar, “Fabrication of Silicon nanowire arrays based solar cell with improved performance,” Sol. Energy Mater. Sol. Cells 95(1), 215–218 (2011). [CrossRef]
26. J. Oh, H. C. Yuan, and H. M. Branz, “An 18.2%-efficient black-silicon solar cell achieved through control of carrier recombination in nanostructures,” Nat. Nanotechnol. 7(11), 743–748 (2012). [CrossRef] [PubMed]
