Black silicon solar cell: analysis optimization and evolution towards a thinner and flexible future

Black silicon (b-Si)—as the name implies—absorbs almost every incident photon in the visible spectrum and therefore appears black to the naked eye. Reduction of reflection being a major criterion for the design of efficient solar cells, such black silicon surfaces (nanostructured silicon surface) have a tremendous potential for photovoltaic application (wafer based silicon substrates ∼180 μm thick). Nano-structuring the light-absorber layer with subwavelength features by a variety of techniques have been reported by several authors without following any design guidelines or masked etching procedure [1–6]. Although such techniques will essentially reduce the reflection from the silicon surface drastically, it may not offer ultra-low reflection characteristics for a wide range of incidence angle and unified photon–electron harvesting due to the increased charge carrier recombination offered by the large surface area of the nanostructures. Therefore, choice of the nanostructure geometry is very crucial for optimum photon–electron harvesting. Among several other nanostructure geometries, nanocones have shown to exhibit superior light trapping capabilities [7–9]. But even for nanocones, a thorough optimization of the dimensions (base diameter and height) aimed at achieving ultra-low reflection for wide angles of incidence along with reduced non-radiative carrier recombination is needed as the schemes for achieving optimum optical and electrical performance are mutually exclusive. Moreover, as the nanostructure geometry is commonly realized by dry/wet etching techniques the material loss incurred also needs to be taken into account. As the primary scaling factor in photovoltaics is the reduction of the active absorber thickness, research groups have initiated the use of nanostructures for ultrathin (3–40 μm) monocrystalline silicon substrates [8–13]. However the design of the nanostructures suffers from the same critique as mentioned earlier. Further, the key question of the ultrathin absorber layer thickness, for which the optimized nanostructure geometry will indeed be beneficial, remains unanswered.

In this paper we have first provided a thorough opto-electrical investigation on nanocones having different aspect ratio (AR) for unified photon electron harvesting, while maintaining (i) ultra-low reflection for wide angles of incidence, (ii) reduced surface area enhancement for low carrier recombination and (iii) low material wastage. It is seen that unique micro-nanostructures (base diameter in nanoscale and height in microscale) can provide a balance between the conflicting abovementioned requirements in unison. Secondly, the effect of micro-nanostructures on the efficiency for wafer based silicon solar cells (∼180 μm thick) has been thoroughly done to clearly bring out on how wafer based b-Si solar cells must gradually evolve towards ultra-thin flexible crystalline silicon substrates as a thick silicon wafer is brittle and therefore must be supported on a rigid support adding cost and limiting applications in the cell to module transition. Moreover, the work brings out the optimal substrate thickness on which the designed micro-nanostructures may be etched while maintaining photon absorption as that obtained with wafer based mono-crystalline silicon solar cells. As the micro-nanostructure geometry behaves as a lossy type photonic crystal in the visible region of the electromagnetic solar spectra, it will have sufficient absorption in the desired wavelength range (300–1100 nm) and an optimized substrate thickness etched with the designed micro-nanostructures is found capable of absorbing the entire solar spectrum in contrary to a bare substrate of the same thickness. Finally, we demonstrate the fabrication of the micro-nanostructures by a simple masked etching procedure using nanosphere lithography and MacEtch technique followed by the realization of a solar cell (p-type crystalline silicon substrate with n-type amorphous silicon hetero junction) on an ultra-thin silicon absorber. It is seen that b-Si based mono-crystalline ultra-thin flexible solar cells can adapt to various emerging technological environments due to its ability to be fabricated on flexible supports allowing roll-to-roll processing thereby making it technologically flexible.

To perform the opto-electrical optimization we have simulated a nanocone structure (height 'h', base diameter 'd') with different AR (given by 'h/d') for different angles of incidence. The nanocone geometry has been optimized with respect to three major criteria: (i) reduced reflection for wide angles of incidence, (ii) reduced surface area and (iii) reduced material wastage. The electrical analysis of the optimized nanocone geometry has been carried out subsequently to estimate the efficiency of the solar cell having a physical thickness of ∼180 μm.

2.1. Optical simulation

Before carrying out the simulation using commercially available electromagnetic solvers, a hand-waving analysis of the nanocone geometry has been done using ray optics approximation. Figure 1 shows the schematic wherein light gets trapped due to multiple bounces of light for wide angles of incidence. The number of bounces offered by the structures where ray-optics approximation is valid can be calculated by a simple algorithm for various geometries under consideration (supplementary information). From the information it is clearly seen that high AR (≥3) and high base diameter (≥400 nm) structures offer significant number of bounces of the incident light thereby significantly reducing the reflection. Further, it is seen that for a given AR, structures having different base diameter(s) will have different height(s) with certain portions of the height offering different number of bounces. This analogy is more appropriate when the dimension of the structures are such that the ray-optics approximation holds good.

Zoom In Zoom Out Reset image size

Simulations have therefore been carried out using Comsol Mutiphysics and Lumerical FDTD Solutions to better understand the behavior of the geometry due to its subwavelength/wavelength scale dimensions. Figure 2 shows the simulation model taken up for performing the optical analysis. Periodic boundary conditions are applied on either side of nanocones in a suitable way to resemble a hexagonal closed packed geometry. As shown in figure 2, ports 1 and 2 are used to obtain the reflection and transmission characteristic respectively (to obtain the absorption within the active layer) from the scattering parameters (S-parameters) which are given by

$$\begin{eqnarray}&&S11=\sqrt{\displaystyle \frac{{\rm{Power}}\_{\rm{Reflected}}\_{\rm{Port}}1}{{\rm{Power}}\_{\rm{Incident}}\_{\rm{Port}}1}}\mathrm{...},\end{eqnarray} \tag{ 1a }$$

$$\begin{eqnarray}&&{S}_{21}=\sqrt{\displaystyle \frac{{\rm{Power}}\_{\rm{delivered}}\_{\rm{Port}}2}{{\rm{Power}}\_{\rm{Incident}}\_{\rm{Port}}1}}\mathrm{....}\end{eqnarray} \tag{ 1b }$$

Zoom In Zoom Out Reset image size

For obtaining the reflection characteristics for bulk wafers (∼180 μm) perfectly matched layer boundary conditions are used. For finding the absorption in thin/ultra-thin silicon substrates (1–15 μm); perfect electrical contact (PEC) boundary conditions are used which takes into account the effect of reflection from the rear metal contact/reflector. The optical excitation (300–1100 nm) is given at port 1 and the refractive index of silicon is taken from Palik [14]. The optimization of the nanocone for the three (3) conflicting regions of interest is detailed below.

• (i) Region I: Low reflectance: Figure 3(a) shows the integrated reflection (R_Int) given as per equation (2) over the spectral region (300–1100 nm), offered by the geometry for normal incidence of light having different 'd' and AR.

$$\begin{eqnarray}&&{R}_{{\rm{Int}}}=\displaystyle \frac{\displaystyle {\int }_{\lambda =300\;{\rm{nm}}}^{1100\;{\rm{nm}}}R(\lambda ){N}_{0}(\lambda ){\rm{d}}\lambda }{\displaystyle {\int }_{\lambda =300\;{\rm{nm}}}^{1100\;{\rm{nm}}}{N}_{0}(\lambda ){\rm{d}}\lambda }\mathrm{...},\end{eqnarray} \tag{ 2 }$$

where R(λ) is the wavelength dependent reflectance and N₀(λ) is the photon flux corresponding to AM1.5 G solar spectrum in the wavelength regime of interest. One may note here that the AR = 0 represents a flat silicon surface. It is clearly seen from the figure that for AR ≥ 3 and base diameter ≥400 nm there is no significant change in reflection characteristics (uniform blue area). This is because light, which is incident on such high AR structures with the above mentioned base diameter(s), will be trapped due to the multiple bounces as absolute height of the nanostructures in the uniform blue region is above 1 μm. Again, the reflection characteristic significantly deteriorates for base diameters ≤200 nm. A wireframe in figure 3(a) is used to select the geometries satisfying the low reflection requirement of the solar cells. It is very important to point out here that figure 3(a) is not indicative of the upper limit of the AR as well as the base diameter. It only gives us the information that the AR should be greater than or equal to three (3) and the base diameter should be greater than or equal to 400 nm.

• (ii) Region II: Low surface area enhancement: From a solar cell perspective, high AR (≥3) and high base diameter (≥400 nm) structures will have a higher surface area enhancement over planar silicon. The surface area enhancement ratio may be expressed as ${A}^{F}/A{\rm{proj}}$ where A^F is nanostructured surface area and A_proj is the planar surface area. Structures having AR > 4 are seen to have higher ${A}^{F}/A{\rm{proj}}$ as seen in figure 3(b). Such structures will therefore lead to enhancement in the surface area resulting in increased surface recombination. A wireframe in figure 3(b) is used to select the geometries satisfying the low surface area enhancement which is also a major requirement for solar cells. It is interesting to note that figure 3(b) helps us to set the upper limit of the AR but is incapable of setting the upper limit of the base diameter.
• (iii) Region III: Low material wastage: Such nanostructures are normally obtained by etching of silicon where material losses will have a significant role to play especially for next generation solar cells. Figure 3(c) shows the material wastage as a function of different AR and base diameter. The material loss has been calculated by subtracting the silicon volume which needs to be etched to obtain the desired geometry from a silicon block. It is seen that realization of low AR < 5 and low base diameter <700 nm structures will lead to least material wastage. A wireframe in figure 3(c) is used to select the geometries satisfying the criteria of least material wastage and finally helps us to define the upper limit of the base diameter.
• (iv) Overlap of three regions: Choice of proper AR and base diameter can only be obtained by mapping the mutually exclusively requirements of reduced reflection for wide angles of incidence, reduced surface area enhancement and reduced material wastage. In figure 3(d), the wireframes from figures 3(a) to (c) have been overlapped. It shows that when the three (3) conflicting interests are put together structures having a base diameter between of 400 and 600 nm and AR between 2 and 4 emerges as a sweet-spot (Region d). One may note here that a nanocone having a base diameter of 500 nm and an AR of four (4) will have an absolute height of 2 μm. Such structures may be termed as micro-nanostructures due to their unique dimensions and are of significant importance for photovoltaic applications. This is in contrast to true microstructures (both base diameter and height in microns: textured surface) and true nanostructures (both base diameter and height in nanoscale: moth-eye surface). The next section will detail the electrical analysis of the solar cells using the optimized micro-nanostructures.

Zoom In Zoom Out Reset image size

2.2. Electrical analysis

In the electrical domain, maximizing the collection of electron hole pairs is the foremost criteria for achieving high efficiency solar cells. One must therefore ensure that the improved optical performance does not come at the expense of the electrical properties. However, the efficiency of the solar cell is not only dependent on the front surface parameters but also on the rear surface parameters. To analyze the effects of high surface area on the surface recombination velocity at the front nanostructured surface, the effective carrier lifetime is the most crucial parameter which may be expressed as [15, 16]

$$\begin{eqnarray}&&1/{\tau }_{{\rm{eff}}}=1/{\tau }_{{\rm{bulk}}}+({S}_{p}+Sn)/d\mathrm{....},\end{eqnarray} \tag{ 3 }$$

where d is the thickness of the wafer, τ_bulk is the lifetime of the excess carrier in the bulk and S_p and S_n are the effective surface recombination velocities at the front and rear surface of the solar cell, supposing a uniform excess carrier distribution. Further

$$\begin{eqnarray}&&Sp={S}_{{\rm{f}}{\rm{loc}}}{A}^{F}/{A}_{{\rm{proj}}}\mathrm{....}\end{eqnarray} \tag{ 4 }$$

Here, S_{f loc} is the surface recombination velocity of the planar surface and A^F/A_proj is the surface area enhancement ratio.

The choice of the sheet resistivity which is inversely proportional to the dopant concentration of the homogeneous shallow emitter (200–400 nm) is a crucial factor for solar cell design (supplementary information). For low doping (∼10¹⁸ cm⁻³) the sheet resistivity will be greater than 200 Ω/sq. offering high contact resistance. For high doping (∼10²⁰ cm⁻³), the sheet resistivity is below 50 Ω/sq but the Auger recombination dominates the carrier lifetime rather than the surface area of the nanostructured geometry. Therefore, we have chosen a doping concentration in range of 3 × 10¹⁹ cm⁻³ with junction depth of 200 nm yielding a sheet resistivity of ∼100 Ω/sq. which is standard for conventional solar cells. Here S_{f localized} depends linearly on surface area. We have carried out the analysis of the increased surface area and its impact on the surface recombination velocity and carrier lifetime in two distinct domains to estimate the efficiency considering the simulated reflectance for various angles of incidence by methods outlined in [17–20].

• (i) Domain 1: S_n > S_{f loc}: This domain generally represents the commercially available industrial solar cells having a p-type silicon substrate (〈100〉, 1–3 Ω cm). The back surface recombination velocity is higher because of the full metallization of the back surface to achieve the back surface field and back contact/back mirror. The front surface recombination velocity is low due to the presence of the composite anti-reflection/passivation layer.

Figure 4(a) shows the efficiency of a solar cells (S_{f loc} = 10 cm s⁻¹ [21], S_n = 300 cm s⁻¹ [17]) having AR ranging from 1 to 4 for varying angle of incidence considering cosine effect. The bottom row depicts the response of a conventional textured nitride solar cell (TxN SC) for ready reference. In this domain, the efficiency is mainly governed by the optical parameters and even structures having AR of 4 is seen to have better opto-electrical performance as S_p remains sufficiently low as compared to S_n. The efficiency is seen to gradually increase with increase in AR ranging between 1 and 4. Therefore for nanostructured solar cells having full rear metallization there will always be an enhancement over the baseline efficiency even if we do not pay much attention towards the design of the nanostructured geometry itself [1–7]. Thus, in this domain the optical parameter (reflection) mainly governs the efficiency of the solar cells rather than the electrical parameters (carrier lifetime).

• (ii) Domain 2: S_{f loc} = S_n: This domain is rather more interesting as currently there is a tremendous thrust on the back surface passivation of solar cells. With the emergence of the PERx family of solar cells the back surface recombination velocity can now be significantly reduced as the acceleration in the PV roadmap has led to the cross-fertilization between the micro-electronics and solar cell industries leading to industrial versions of PERx solar cells. Further with the emergence of passivated contact solar cell (extremely thin dielectric tunnel oxide providing an extremely low resistance to charge carrier transport along with passivation), the day is not far when the back surface recombination velocity will actually be very less and comparable to that of the front.

Zoom In Zoom Out Reset image size

Figure 4(b) shows the efficiency of a solar cells (S_{f loc} = 50 cm s⁻¹, S_n = 50 cm s⁻¹) having AR ranging from 1 to 4 for varying angle of incidence considering cosine effect. It is seen that for higher AR, the front surface recombination velocity becomes greater than that of the rear surface as per equation (4) and therefore in this domain a thorough optimization of the nanostructure is needed as both the optical (reflection) and electrical parameters (carrier lifetime) governs the efficiency of the solar cell. Non optimized geometries for light trapping will then lead to degradation in the performance of the solar cells. It is seen that AR between 2 and 3 is optimum for such solar cells. Efficiency remains unaffected for a limited range of incidence angles (±35°) even for an optimized geometry while the simulated integrated reflectance remains unchanged for incidence angles (±55°) (figure 5) calculated as per the previous work of the authors [22]. This is because the cosine effect has been taken into consideration while calculating the efficiency. Thus before guarantying the techno-commercial success of wafer based nanostructured black silicon solar cells, it is therefore essential to first experimentally validate the angular efficiency response of a conventional silicon solar cell which is expected to remain unchanged for incidence angles (±25°) (figures 4(b) and 5).

Zoom In Zoom Out Reset image size

The angular efficiency response of conventional solar cells is measured with the help of a rotating zig which rotates the solar cell with respect to the light source. This technique inherently takes into account the cosine effect. We have provided the first (slope) and second derivatives (curvature) (figure 6) of the experimentally measured results of a state-of-the-art conventional solar cell to better understand the angular response and make it independent of the absolute value of cell efficiency. Angle of incidence greater than 60° has not been plotted as they would not be very realistic in a real working solar cell configuration. The analytical derivatives of the conventional cell (TxN SC referred in figure 4) is also provided in the same figure for ready referencing. It is important to note here that figure 6 is not a fit of negative sine function as the efficiency depends on both the cosine effect and reflection for varying angles of incidence. Further the measured change in efficiency is less as compared to the analytical calculation because of the effect of the diffused radiation during measurement.

Zoom In Zoom Out Reset image size

To further delve into the matter the second derivative has also been taken which provides the curvature and hence the angle at which the maximum change in efficiency occurs. It is seen from figure 6 that there is a maximum curvature for incidence angles (20°–25°). It is interesting to note that the measured curvature matches the expected angular transition as per figures 4(b) and 5 in spite of the effect of the diffused radiation on the solar cell surface.

Due to the strong dependence of the angular response of a conventional solar cell with respect to reflection/efficiency, wide angle light collection is crucial when solar cells have to harvest light in conditions when there is significant amount of diffused radiation (as on a cloudy day) or when the solar cells are not facing the Sun (morning, evening and winter light losses for solar panels fixed at optimum inclination angle for a corresponding latitude). But as seen in section 2 the efficiency drastically reduces mainly due to the cosine effect not only for textured conventional solar cell (beyond ±25°) but also for nanostructured solar cell (beyond ±35°). Therefore, black silicon wafer based technology indeed has the limited potential for such 180 μm silicon substrate where the efficiency remains unaltered only within a small range of ±35° (which is only ±10° enhancement over the conventional solar cell) out of the entire desired range of interest (±60°). This inference should come as no surprise as the potential of wafer based black silicon solar cells has already come under reproach [23].

Further, a severe limitation of b-Si solar cells is that as it already has a near-zero reflection from the silicon surface, module integration becomes a challenge. If existing means of encapsulation using glass/EVA (refractive index ∼1.5) are used the optical system will get disturbed which will hamper the overall efficiency of the module when compared to a module having conventional textured solar cells [17]. Moreover, the angular response of the module will be no different due to the b-Si solar cells as the relative light transmission into the module for any angle of incidence results will solely depend on the front flat glass-air interface. Therefore b-Si modules is not a technologically viable option as both the benefits of (i) reduced reflection and (ii) wide angle light collection are offset during standard cell to module technology.

So what do we do with a solar cell technology which has the potential of offering wide angle light collection, low capital cost, high values of efficiency and simple fabrication methods? The authors feel that there is a void which had existed in thin silicon solar cells technology in spite of the immense potential of silicon to achieve high efficiencies in reduced absorber thickness. In fact the limiting efficiency of wafer based c-Si solar cells was capped at 30% by Shockley and Queisser for monocrystalline single junction solar absorbers ∼50 μm thick [24, 25]. Ultrahigh efficiencies in such absorber thickness is still not a reality as the handling barrier of silicon is ∼80 μm. As emerging photovoltaic cells are the ones which can be fabricated on flexible supports supporting high speed roll-to-roll processing, it would be unjust to rule out silicon because its use as a very good mechanical material (flexible at thickness below ∼50 μm) was reported by Petersen in 1982 [26] and had opened up silicon based micro electro mechanical systems domain which includes the use of silicon diaphragm and movable cantilevers/ fixed–fixed beams at the micron-scale for a variety of applications including pressure sensors and switches. In a race to reduce the Cost/Watt and better Watt/Gram utilization, lowering of the active absorber thickness of silicon is mandatory and it is only now that industries and researchers have focused on the use of thin monocrystalline silicon substrates ≤50 μm thick for achieving the SQ limit.

At the industry front, Crystal Solar, California with their Direct Gas to Wafer™ technology approach has enabled the fabrication, handling, processing, and packaging of thin (<50 μm), single crystal silicon wafers and solar cells, which can be packaged into industry-standard modules. When this technology is transitioned into manufacturing, direct manufacturing costs for the complete solar panel approaching $0.50/W_p are achievable [25]. Astro-Watt, Texas has developed a proprietary semiconductor on metal (SOM™) technology for creating thin silicon wafers that can be processed using today's PV manufacturing tools. The company is targeting 22%–23% efficiency in just 25 μm mono-crystalline silicon on a metal substrate [27].

The research frontier being one step ahead has already initiated work on ultrathin crystalline silicon (∼3–40 μm). Realization of such ultrathin silicon from bulk wafers in the form of microscale bars and ribbons developed by Rogers's group [13] by far remains the most common technique rather than the exfoliation technique [28]. However, in such low absorber thickness, light trapping becomes crucial and it is very essential to have optimized photon–electron harvesting to achieve the SQ limit in an optimal substrate thickness. But how do we chose the substrate thickness? The authors feel that the choice is very much analogous to the desired weight of a human being. It is very essential to have a correct weight as being overweight or underweight will have its own share of problems like obesity or anorexia. Too thick substrates (conventional wafer based silicon) are undesirable from the material usage point of view and too thin substrates (conventional silicon thin films obtained by PECVD technique) will suffer from reduced absorption of photons even with optimal light trapping structures. Therefore a study has been carried out for the optimally designed nanostructure etched on different substrate thickness.

Figure 7 shows the enhancement in absorption due to the micro-nano structuring of the silicon slab for different substrate thickness by performing simulations as described in section 2 (figure 1 with PEC boundary conditions). It is seen that the enhancement factor is almost same for substrate thickness greater than 15 μm (∼35%) which clearly indicates that the absolute absorption of the substrate with and without micro-nanostructures has nearly saturated (inset of figure 7). In such substrate thickness (beyond 15 μm), the enhancement is mainly due to the anti-reflective property of the nanostructures as no advanced light management will be dominant if the substrate already has the capability to absorb most of the incident photons. So in such cases, a simple optimized anti-reflection coating will serve the purpose rather than micro nano-structuring the substrate itself. On the other hand, the enhancement factor is significantly higher (>50%) in sub 10 μm substrates as the micro-nano geometries not only act as structures providing ultra-low reflection but also play the role of a super scatterer. In such sub 10 μm substrates, although the enhancement is nearly same (55%–60%), the absolute value of absorption is more and tends towards saturation for a 10 μm substrate rather than a substrate of <10 μm (inset of figure 7). Therefore it is judicious to choose a substrate thickness in the range of 10–15 μm with optimized micro-nanostructures.

Zoom In Zoom Out Reset image size

For such substrate thickness, the advanced light management offered by the structures is because at a cross- section along any z-axis, the geometry may be considered as a 2D photonic crystal having a periodic variation of refractive index along x and y axes. In other words it can be treated as a collection of 2D photonic crystals where the width of the high dielectric region (Si) is increasing while the width of the low dielectric medium (air) is decreasing along the direction of light. As there is no periodicity in z-direction, there will be no band-gap and hence, minimum reflection will take place for light penetrating vertically. Rather as an effect of the effective graded index that increases along the z-direction, light will be vertically collimated. When the crystal symmetry is hexagonal (base of the nano-pillar array) the largest photonic band gap occurs (as in case of 2D photonic crystals). Further, there will be a gradual variation of band gap along the z-direction due the continuous variation of the widths of the high and low dielectric regions. Electric field corresponding to higher frequency will be more and more concentrated within the high dielectric material (silicon in this case) and will be maximum towards the substrate as the width of silicon along z direction is increasing effectively. Figure 8 shows the variation of electric field distribution over a unit-cell for different incident wavelength along the cut plane (in z direction). It is evident that the field intensity periodically varies within and outside the pillars as a result of the periodicity of the 2D photonic crystal. Moreover the structure is actually a lossy type photonic crystal in the visible region of electromagnetic spectra as the micro-nanostructures have sufficient absorption in this wavelength range. Hence, the structure cannot generate loss-less guided mode but this is not a problem for solar cell applications as the absorbed photon can generate electron–hole pair inside the silicon micro-nanostructures opening up the possibility of radial junctions. It is very important to point out here that record high efficiencies have already been reported on ultra-thin monocrystalline substrates [12, 13] and with an appropriate research thrust on the design of the nanostructure geometry along with the proper substrate thickness as outlined by the authors, ultra-thin monocrystalline silicon substrates will soon tend towards achieving the SQ limit.

Zoom In Zoom Out Reset image size

Therefore, the black silicon technology must evolve towards a flexible future (ability to be fitted to curved and bendable structures) where they can provide light scavenging capabilities with high values of efficiencies at very low cost. Assuming flexible ultra-thin monocrystalline silicon (∼10–15 μm in thickness) is available in the market as standard monocrystalline wafers (180 μm thick), figure 9 shows on how flexible b-Si based solar cell technology can grow in different technological environments demonstrating its technological flexibility. Further as for thin silicon absorbers the effect of surface recombination is more pronounced (as seen from equation (3)), the quality of the monocrystalline substrates in terms of minority carrier life-time would be a crucial parameter governing the absolute value of the efficiency of the solar cells. It is clearly seen the optimized nanostructure geometry can be fabricated by wet/dry etching techniques followed by the formation of a homojunction/heterojunction, realization of antireflection and passivation layers and finally obtaining the contacts depending on the availability of existing equipment. A conformal polymer encapsulation would ensure the retention of the optical properties offered by the nanostructure geometry. Thus it is clearly seen that ultra-thin b-Si solar cells not only provide a physical flexibility (allowing roll-to-roll processing) but also offers significant technological flexibility. The next section will discuss a simple yet novel fabrication methodology involving minimum infrastructure of realizing flexible black silicon solar cells having an optimized geometry for unified photon–electron harvesting.

Zoom In Zoom Out Reset image size

The fabrication methodology involves a method where flexible mono-crystalline 〈100〉 p-type substrates are obtained from a wafer having a thickness of 180 μm by a patent-pending process in which we can successfully realize flexible substrate of thickness 10 ± 2 μm. The process however can be tuned to achieve other substrate thickness. The optimized nanostructures are obtained by a MacEtch process involving nanosphere lithography where the silica nanospheres act as masks during the noble metal deposition making the process capable of realizing micro-nanostructures having different base diameter and ARs.

A mono-layer of synthesized silica nanoparticles [29] is obtained by a dip coating technique on substrates having a dimension of 3 cm × 4 cm approximately. The monolayer can also be obtained by spin-coating technique as reported in [30]. It is very important to point out here that LB trough technique can serve as an upscale process for realizing a mono-layer of silica nanoparticles on large areas. Thereafter, a thin film of gold is thermally evaporated before etching the substrate in a solution containing HF, H₂O₂ and DI-water to finally have the optimized geometry. The size of the silica nanoparticles is the most crucial parameter which determines the top diameter of the nanopillar. One may note here that the concentration of H₂O₂ and amount of gold deposited determines the lateral movement and hence governs the base diameter of the nanopillar for a chosen AR. Figure 10(a) shows a FESEM image of a mono-layer of silica nanoparticles (diameter ∼200 nm) with true hexagonal closed packing in most regions. Figure 10(b) shows the SEM image of the optimized nanopillar geometry (base diameter ∼500 nm and AR ∼3). With slight variations in the etching solution and time any desired geometry can be realized. Figure 10(c) shows the FESEM image of a tapered nanopillar having a top diameter of ∼100 nm, base diameter of ∼200 nm and a height of ∼1000 nm.

Zoom In Zoom Out Reset image size

After removing the remnant of silica nanoparticles and gold the sample is loaded in a capacitively coupled PECVD system (RF excitation frequency of 13.56 MHz) to deposit the n-type a-Si layer (1% PH₃ in SiH₄ and H₂ in ratio of 1:5). The deposition was carried out at 200 °C, 1 mbar at 12 W of RF power for 10 min. The deposition rate of the n-type a-Si layer is ∼100 Å min⁻¹. To obtain the final solar cell, contacts were realized by depositing aluminum on either surfaces (full metallization on rear and grid metallization on front side). Figure 11 shows the complete laminated flexible solar cell whose angular efficiency response is given in figure 12(a). The response of a bare silicon solar cell (∼10 μm) is also super-imposed in figure 12(a) along with the response of a micro-nanostructured solar cell with a SiN_x passivation layer.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 12(b) shows the enhancement in absorption obtained by simulation and enhancement in efficiency obtained by taking ratio of the experimental curves (micro-nanostructured solar cell over bare silicon solar cell) shown in figure 12(a). It is seen that a ∼50% enhancement in absorption is predicted by simulations for varying angle of incidence and a ∼35% enhancement over the bare substrate is obtained experimentally. The nature of both the curves shows similar response to variation in angle of incidence. The experimentally obtained enhancement in efficiency is less due to the various electrical parameters of the solar cell which are not considered in optical simulation.

A thorough analysis and optimization of the nanostructure geometry for photovoltaic applications has been presented. It is seen that micro-nanostructures having a base diameter between 400 and 600 nm and AR between 2 and 4, are able to satisfy the mutually exclusive requirements of (i) low reflection for wide angle of incidence, (ii) low surface area and (iii) low material wastage needed for optimum solar cell performance. Investigations of such nanostructures housed on 180 μm monocrystalline silicon substrate reveals that wafer based b-Si module technology may not be techno commercially viable amidst a range of competing silicon module technologies. The primary scaling factor in photovoltaics being the thickness of the absorber layer, such micro-nanostructures are found to be more potent due to their advanced light management capabilities for future ultra-thin absorbers. Further, the work clearly brings out the optimum thickness of the monocrystalline silicon absorber which is capable of achieving the well know SQ limit for solar cells. Normally c-Si bulk wafers are used to obtain the ultra-thin substrates which not only increases the processing steps but also the yield of the solar cells. Production lines involved in wafer based silicon technologies must therefore support the growth of thin mono-crystalline nanostructured absorbers irrespective of the current technology line (PERx family, n-PASHA: bifacial n-type cell with screen and stencil-printed metallization, heterojunction technology: n-type heterojunction cell with screen-printed metallization, interdigitated-back-contact solar cells) as the future belongs to thin nanostructured black silicon modules. It is therefore high time to come up with a technology to realize such thin/ultra-thin substrates as the technology road map of solar photovoltaics clearly indicates that the substrate thickness should be less than 25 μm by 2024 [31, 32]. The day is not far when all technologies should converge towards ultra-thin nanostructured silicon absorbers along with a unification of c-Si and a-Si technologies. Considering that ultra-thin silicon absorbers are not available as off-the-shelf product, the authors hope that the reported work may help in awakening of the dormant silicon ribbon technology or monocrystalline epitaxial deposition. Nanostructured solar cells of monocrystalline silicon ribbons with conformal layers of encapsulate is expected to pave the path for next generation solar cells and modules.

The authors would like to acknowledge all members of CEGESS and are grateful to Professor H Saha for providing the necessary infrastructure developed with the financial support of Department of Science and Technology (DST) and Ministry of New and Renewable Energy (MNRE), Government of India (GoI). The authors would like to thank Professor Swapan K Datta for his valuable inputs and Professor A K Barua for his constant encouragement and support. Arijit Bardhan Roy and Sonali Das would like to acknowledge DST, GoI for financial support. Sonali Das would also like to acknowledge Bhaskara Advanced Solar Energy (BASE) Program of DST, GoI and Indo-US Science and Technology Forum (IUSSTF). Mrinmoyee Choudhuri is thankful to Mr Anirban Neogi, Dr Sudhir Chandra Sur Degree Engineering College for his kind motivation.