H2SO4 using a voltage of 25 V. The second approach to selective anodization was achieved by inkjet printing 50% (w/w) of H3PO4 (electronic grade, J.T Baker) ...
SELECTIVE ANODIZATION Pei Hsuan Doris Lu, Stuart Wenham and Alison Lennon The University of New South Wales, Sydney, NSW, SPREE TETB 2052, Australia Abstract — Selective anodization is a process that can enable the formation of isolated conductive regions in a dielectric layer. It’s novelty arises from the way a single layer of conductive metal can be anodized so as to create a dielectric material with isolated regions of a defined pattern that remain conductive. Electrical contact can be made to the underlying device via the conductive regions while the device remains passivated by the anodized regions. This paper reports on two different methods for achieving the selective anodization of aluminium and presents a novel interdigitated rear contact device design that could use this selective anodization process to achieve a pattern of metal and dielectric regions. Index Terms — anodization, rear contact, passivation, anodic aluminum oxide, interdigitated solar cell
I. INTRODUCTION Previous studies have shown that when nanoporous anodic aluminium oxide (AAO) layers are formed by anodizing an aluminium layer over an intervening silicon dioxide or silicon nitride dielectric layer on a silicon wafer surface, the effective minority carrier lifetime of the wafer can be increased [1-3]. This paper introduces a novel way to form metal contact regions using a single aluminum layer and selectively anodizing regions of this layer to leave conductive regions within a dielectric AAO layer. This process allows the lifetime-enhancing properties of anodization to be used whilst forming metal regions for electrical contact. There are two general approaches for achieving selective anodization. One approach requires the formation of a mask over the aluminium which can prevent the electrolyte from contacting selected regions. In this approach, the masked regions are not anodized and remain metallic. The mask can be formed by printing (e.g., using an inkjet, aerosol or screen printer) a polymer layer over regions of the aluminium surface. The second approach is to isolate the aluminium regions before anodization and only apply the anodic current to those regions that are to be anodized. This paper presents proof-of-concept results for both selective anodization processes and proposes the use of the process in the fabrication of metallic fingers for localised line contacts. An interdigitated rear contact solar cell design, which incorporates the use of selective anodization, is proposed. II. EXPERIMENTAL Commercial-grade 156 mm 1-3 Ω cm p-type, alkalinetextured, Czochralski (Cz) silicon wafers were lightly-diffused with phosphorus to an emitter sheet resistance of ~100 Ω/ .
The residual phosphorus at the rear surface was removed by etching and thin SiO2 layers were thermally-grown on both surfaces. A 65 nm thick SiNx layer was deposited onto the front surface and 200 nm SiNx was deposited onto the rear surface using remote PECVD. The wafers were cleaved into fragments of ~ 4 cm × 4 cm. An aluminium layer of thickness 600 nm was either thermally-evaporated or sputtered onto the rear SiNx surface of the wafer fragments. Selective anodization according to the first approach was achieved by aerosol-jet printing a layer of acid-resistant novolac resin (Microposit FCL; obtained from Rohm and Haas) to form a mask on the aluminum layer. The wafer fragments were then heated for 10 min at 140 °C to densify the resin and then anodized as described in [1] in 0.5 M H2SO4 using a voltage of 25 V. The second approach to selective anodization was achieved by inkjet printing 50% (w/w) of H3PO4 (electronic grade, J.T Baker) to etch isolation lines in the aluminum layer, thus isolating the layer into distinct regions. The wafer fragments were heated at 200 °C for 10 min to ensure complete etching of the aluminium in the printed areas and then anodized in 0.5 M H2SO4 using a voltage of 15 V as shown in Fig. 1. In this process the anode of the power supply was electrically contacted to the region(s) to be anodized (i.e., dielectric regions). III. RESULTS AND DISCUSSIONS Fig. 1 (a) is a digital image of acid-resistant novolac resin lines printed on a layer of sputtered aluminium and Fig. 1 (b) is a digital image of the same wafer fragment after the wafer fragment was anodized and the resin removed. The white lines in Fig. 1 (b) are aluminium regions which were covered by novolac resin during anodization (i.e., which have not been anodized and remain conductive). Fig. 1 (c) is microscope image of a section of an aluminium line showing the surrounding (brownish) AAO which differs from the aluminium line in both colour and height.
(c)
Fig. 1. (a) Digital image of a wafer fragment with novolac resin lines printed on a sputtered aluminum layer; (b) digital image of the wafer fragment after anodization with the resin lines removed; and (c) microscope image of a section of an aluminium line after anodization and removal of the resin lines.
Fig. 2 (a) shows the microscope images of the printed 5 layers of resin on a sputtered aluminium surface. The profile of the resin lines appears higher than the aluminium surface. Fig. 2 (b) shows an example resin line after baking for 10 min at 140 °C and anodizing. The height of resin after anodization seems reduced from that in Fig. 2 (a) though the masking lines are still intact on the sputtered aluminium surface. Fig. 2 (c) depicts a sample aluminium finger after resin removal by immersion in acetone.
Fig. 2. Microscope images of sputtered aluminium (a) after printing 5 layers of the novolac resin on the sputtered aluminium surface; (b) after baking for 10 min at 140 °C and anodization in 0.5 M H2SO4 at 25 V; and (c) after resin removal by immersion in acetone.
Fig. 3 (a) shows the microscope images of novolac resin lines printed on an evaporated aluminium surface, where the resin lines are higher than aluminium surface. Fig. 3 (b) displays the microscope images after the resin has been removed by acetone. There is only a small colour difference between masked and unmasked regions in Fig. 3 (b), which suggests the whole aluminium layer, including the masked regions, has been anodized. The evaporated aluminium surface was smoother than the sputtered surface and consequently the adhesion of the printed resin was poorer on that surface. During anodization the resin peeled off exposing the underlying aluminium to the electrolyte and thus it was able to be anodized.
Fig.3. Microscope images of evaporated aluminium: (a) after printing of 5 layers of novolac resin and (b) after the resin has been removed by immersion in acetone.
The second approach for isolating aluminium into regions was demonstrated on evaporated aluminium because it was preferable to use a method that could work for both sputtered and evaporated aluminium layers. Fig. 4 (a) shows a digital image of lines that were etched by inkjet printing 50% (w/w) H3PO4 on an evaporated aluminum surface. Fig. 4 (b) and (c) show digital images of a patterned evaporated aluminum layer before and after anodization at 25 V in 0.5 M H2SO4, respectively. (a)
(b)
(c)
Fig. 4. (a) Image showing etched lines in the evaporated aluminium layer; (b) the wafer fragment before anodization; and (c) after anodization.
Fig. 5 shows a wafer fragment patterned into metal and AAO regions by first etching isolation lines in the aluminium layer and then only electrically contacting regions to be anodized. The white regions in Fig. 5 remain as aluminium as they were electrically isolated during anodization, whereas the grey regions were anodized to form AAO regions.
Fig. 5. Digital image of a wafer fragment on which an evaporated aluminium layer was patterned into regions by etching isolation lines in the aluminium layer. The lines were etched by inkjet printing 10 layers 50% (w/w) of H3PO4 on the evaporated aluminium surface and anodized at 15 V in 0.5 M H2SO4.
Fig. 6 shows an elemental line scan across an aluminum finger from the wafer fragment shown in Fig. 5. The scan was acquired using scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDX). The blue line indicates the silicon signal and the high intensity of that signal at 600 µm and 1100 µm indicates the region where aluminium was etched away to form the isolating lines exposing the underlying SiNx. The red line represents the oxygen signal which is detected across the AAO regions. The oxygen signal at the aluminum finger region decreases to one third of the signal intensity detected in the AAO regions. The green line indicates the intensity of aluminum signal which is 30% higher in the aluminum finger region compared to the AAO regions. The resistivity of the aluminium fingers was measured to be ~ 1.6 × 10-5 Ω cm (which can be compared to the value of 2.71 × 10-6 Ω cm at 25 °C [4] expected for pure aluminium) by measuring the I-V response of the fingers using a probe station.
regions over the n+ lines can be isolated by inkjet printing H3PO4 as described in this paper. Anodization of the remaining surface will result in AAO formation. After anodization, p+ semiconductor fingers can be formed by laser doping through the AAO layer as described in [5]. Individual unit cells can be patterned by the isolation lines, and after selective anodization, parallel strings of unit cells can be isolated and connected in series as described in [7]. Alternatively, a 2-level busbar interconnection pattern can be used, with the aluminium busbars being either thickened by zincating and plating [8] or the bulk of the metallisation provided by aligning the patterned wafers on a metallised backsheet [9]. (a)
(b) Fig. 6. EDX line scan over an aluminium finger of the wafer fragment shown in Fig 5.
The selective anodization process can be used to form contact regions for a range of solar cell devices, including bifacial cells and interdigitated rear contact cells. In addition to improving the passivation of the cell surface in the nonmetal regions, the AAO regions have a refractive index of 1.61.7 therefore enabling light to be captured from both faces of the cell. IV. AAO INTERDIGITATED SOLAR CELL Fig. 7 depicts a proposed interdigitated cell structure fabricated on an n-type wafer that uses selective anodization to form the metal and dielectric regions on the rear surface of the cell. The cell design exploits the fact that aluminium from the AAO can also be used as a dopant source to form p+ contact regions in the underlying silicon [5]. Phosphorus laser-doping can be used to form n+ fingers according to the pattern in Fig. 7 (a). A layer of ~ 2 µm aluminium can be thermally-evaporated onto the rear surface and metal contact
Fig. 7. (a) Depicts a unit cell, patterned on the rear surface of a wafer. (b) Shows a cross section of the region between two p+ semiconductor fingers lines.
Fig. 8 shows the p-type dopant profile, determined using electrochemical capacitance voltage (ECV) measurements, for regions laser-doped through an AAO layer formed by anodizing aluminium at 25 V in an electrolyte comprising 0.5
M H2SO4 and 0.5 M H3BO3. The concentration of p-type dopants exceeded 1020 cm-3 over the first 4 µm of the region formed by doping through the AAO layer. The greater p-type dopant incorporation resulting from laser doping through the AAO is primarily due to aluminium’s diffusion coefficient in molten silicon being approximately three times larger than that of boron [10], consequently when laser-doping using the same laser speed and power more aluminium can diffuse into the silicon and diffuse further from the surface than the introduced boron. By using the ECV doping profile and PV Lighthouse’s sheet resistance calculator (assuming a 1.5 Ω cm p-type silicon wafer), the sheet resistance of the p+ semiconductor fingers was estimated to be 1.27 Ω/□.
Fig. 8. ECV profile showing electrically-active p-type dopant concentration in a 4 cm2 region which was laser-doped through an AAO layer formed by anodizing aluminium at 25 V in an electrolyte comprising 0.5 M of H2SO4 and 0.5 M of H3BO3.
Assuming that contact resistance can be minimised and that sufficient conductivity can be assumed for the busbars, the series resistance, Rs, of the cell would be limited by the length and spacing of the n+ and p+ fingers (i.e., the unit cell dimensions). The fractional power loss associated with this Rs was calculated for both finger polarities using [11]: Pfr,loss=
S f J mp ρL2f 3h f w f Vmp
(1)
where hf and wf represent the height and width of the fingers, Lf is the length of the fingers, Sf is the pitch between two fingers of the same polarity, Jmp and Vmp are the current density and voltage of the cell at the maximum power point. The measured resistivity of the aluminum regions after anodization of 1.6 × 10-5 Ω cm was used to estimate the Rs for the n-type fingers. The Rs contributed from n+ finger aluminum metal grid is relatively small compared to p+ semiconductor fingers. Consequently it was assumed that the largest contributor to the Rs would be due to current flow in the p+ semiconductor fingers.
Equation (1) was used to calculate that the required Lf of the p+ semiconductor fingers to maintain Pfr,loss to be less than 5% (see [7]) was 4.7 mm. This calculation, which assumed a sheet resistance of 1.27 Ω/□ to replace ρ/hf in Eq. [1], Sf = 1000 µm, wf = 100 µm, Jmp =32 mA/cm2 and Vmp= 0.6 V, suggests that it would be necessary to also plate the semiconductor fingers when thickening the busbar regions if significant voltage losses were not to be incurred by such closely-spaced laser-doped p+ semiconductor fingers. Fig. 9 shows the dependence of Lf on the plated-up semiconductor fingers assuming a Pfr,loss of 5%.
Fig. 9. Length of fingers in relation to plated-up Cu thickness on p+ semiconductor fingers when the Pfr,loss is limited to 5%.
Plating of the semiconductor fingers could be achieved during the same electroplating process used to thicken the busbars in a 2-level grid, with the current passing from the bus bars into the fingers as shown in Fig. 10. Although this process would result in a gradient of plated finger height due to relatively low conductivity of the semiconductor fingers, the fingers would be thickest closest to the bus bar where increased conductivity was required most [9].
[4] [5]
[6]
Fig. 10. Schematic diagram of a 2-level busbar interdigitated rear contact metallization scheme enabled using selective anodization.
V. CONCLUSION This paper introduces a selective anodization method which can be used to form a pattern of dielectric and conductive regions from a single layer of aluminum. This process can be used to form localized rear contacts for bifacial or interdigitated rear contact solar cells. An example interdigitated rear contact metallization scheme was proposed. The AAO dielectric regions can serve as a passivation layer and a p-type dopant source. Furthermore, the described aluminum patterning method can be used in the fabrication of other metallization schemes (e.g., the multi-level metallization schemes for IBC cells proposed by Verlinden et al. in [13]). ACKNOWLEDGEMENT This Program has been supported by the Australian Government through the Australian Renewable Energy Agency (ARENA). The Australian Government, through ARENA, is supporting Australian research and development in solar photovoltaic and solar thermal technologies to help solar power become cost competitive with other energy sources. The views expressed herein are not necessarily the views of the Australian Government, and the Australian Government does not accept responsibility for any information or advice contained herein.
[1]
[2]
[3]
REFERENCES P. H. Lu, K. Wang, Z. Lu, A. J. Lennon, and S. R. Wenham, "Anodic Aluminum Oxide Passivation For Silicon Solar Cells," IEEE Journal of Photovoltaics, vol. 3, pp. 143-151, 2013. Z. Lu, P. H. Lu, J. Cui, K. Wang, and A. Lennon, "Self-patterned localized metal contacts for silicon solar cells," Journal of Materials Research, vol. 28, 2013. J. Cui, X. Wang, and A. Lennon, "Light-Induced Anodisation of Aluminium For Rear Surface Passivation of Silicon Solar Cells " presented at the
[7]
[8]
[9]
[10]
[11]
[12]
[13]
28th European PV Solar Energy Conference and Exhibition, Pairs, France 2013. CRC Handbook of Chemistry and Physics, 94 ed., 2013. P. H. D. Lu, B. Xiao, Z. Lu, K. Wang, and A. L. a. S. Wenham, " Laser-Doping Through Anodic Aluminium Oxide Layers Formed On Silicon Solar Cells," presented at the 21st International Photovoltaic Science and Engineering Conference, 2011. J. Benick, B. Hoex, M. C. M. van de Sanden, W. M. M. Kessels, O. Schultz, and S. W. Glunz, "High efficiency n-type Si solar cells on Al2O3-passivated boron emitters," Applied Physics Letters, vol. 92, pp. -, 2008. P. A. Basore, E. Van Kerschaver, K. CabanasHolmen, J. Hummel, L. Yafu, C. P. Murcia, et al., "All-aluminum screen-printed IBC cells: Design concept," in Photovoltaic Specialists Conference (PVSC), 2013 IEEE 39th, 2013, pp. 0900-0903. F. Granek, "High Efficiency Back-Contact BackJunction Silicon Solar Cells," Fraunhofer Institut für Solare Energiesysteme (ISE), Germany 2009. H. Pei-Chieh and A. Lennon, "Backsheet metallization of IBC silicon solar cells," in Photovoltaic Specialists Conference (PVSC), 2013 IEEE 39th, 2013, pp. 2209-2211. H. Kodera, "Diffusion Coefficients of Impurities in Silicon Melt," Japanese Journal of Applied Physics, vol. 2, p. 212, 1963. M. Green, "Solar Cells: Operating Principles, Technology and System Applications," 2 ed. Sydney, Australia: The University of New South Wales, 1998. A. Lennon, Y. Yao, and S. Wenham, "Evolution of metal plating for silicon solar cell metallisation," Progress in Photovoltaics: Research and Applications, vol. 21, pp. 1454-1468, 2013. P. Verlinden, R. M. Swanson, R. A. Sinton, and D. E. Kane, "Multilevel metallization for large area pointcontact solar cells," in Photovoltaic Specialists Conference, 1988., Conference Record of the Twentieth IEEE, 1988, pp. 532-537 vol.1.
