High efficiency n-type silicon solar cells featuring passivated contact to laser doped regions

Minimizing carrier recombination at cell contacts becomes increasingly important for reaching high efficiency. In this work, the passivated contact concept is implemented into n-type silicon solar cells with laser-processed local back surface fields. The passivation and contact characteristics of the SiO2/amorphous silicon (a-Si:H) stack on localized laser doped n+ regions are investigated. We find that the SiO2/a-Si:H stack provides not only good passivation to laser doped n+ regions but also allows a low contact resistivity after thermal annealing. With the implementation of the SiO2/a-Si:H passivated contact, an absolute efficiency gain of up to 1.5% is achieved for n-type solar cells.

The main challenge in today's photovoltaics industry is to increase the conversion efficiency and to lower the production costs of cells.A route towards higher efficiency is via the implementation of selective emitter (SE) and local back surface field (LBSF) structures into silicon solar cells.A SE can reduce contact resistance by allowing heavy doping under the metal contacts, while ensuring low recombination elsewhere via lighter doping in the optically active regions, which results in a higher fill factor (FF) and a better blue response (higher short circuit density J sc ).A heavily doped LBSF can also increase the fill factor due to a decreased series resistance and rear-side recombination.The ultra-high efficiency passivated-emitter rear locally diffused (PERL) silicon solar cells have been fabricated with locally doped regions at front and rear sides. 1 However, the approach used to realize local doping requires cost-intensive and time-consuming processes, such as photolithographybased wet chemical processing and high temperature diffusions, which limit the commercial success of the PERL cell concept.
In recent years, laser processing has attracted considerable attention from industry as a fast and cost-effective technique for forming locally doped regions in silicon solar cells.3][4][5] An absolute efficiency gain up to 1% has been achieved with the implementation of laser-processed SE or LBSF.Recently, n-type PERL solar cells with an efficiency of up to 23.2% have been achieved using a laser-based PassDop technology. 6Compared to the solar cells with local doped regions formed by photolithography and diffusion, however, laser processed solar cells usually exhibit a lower efficiency, which can be attributed to the increased bulk recombination at the laser doped regions caused by the laser-induced defects. 7,8Our recent simulation results have indicated that the recombination current density of laser doped regions without passivation may be higher than 5000 fA/cm 2 (even reaching a few tens of thousand fA/cm 2 with non-optimized laser parameters), 9 which is much higher than that of conventional thermally diffused regions.As such, the high recombination at laser doped regions becomes the dominant loss that limits the efficiency of the laser-processed solar cells.
A possible approach to improve the laser-processed solar cells efficiency is to passivate the laser doped regions with an ultra-thin dielectric, which have to be sufficiently thin to allow current flow via quantum mechanical tunnelling whilst being thick enough to provide reasonable surface passivation.At the same time, the laser doping fraction could be enlarged to reduce the effect of the contact resistivity.1][12][13][14] The results indicated that SiO 2 , Al 2 O 3 , and amorphous silicon (a-Si:H) ultrathin layers with optimized thickness (typically <2 nm) could provide appreciable passivation to diffused p þ and n þ regions whilst maintaining a relatively low contact resistivity.However, the passivation stability of these ultrathin dielectrics during the final metal contact annealing is still a problem to be resolved before being applied on solar cell fabrication.In this work, the implementation of the passivated contact concept into n-type silicon solar cells with laser-processed LBSF is presented.The recent developed SiO 2 /a-Si:H stacks, which have shown excellent passivation to n þ regions and good thermal stability, 15,16 are applied at the laser processed rear side to reduce carrier recombination.
The line-structured n þ LBSF was achieved by laser chemical processing (LCP) with diluted H 3 PO 4 solution (50%), 17 which could open the dielectric layers and form the LBSF in a single step.The sheet resistance (R s ) and contact resistivity (q c ) of LCP-LBSF with or without SiO 2 /a-Si:H stacks were determined by the transfer length method (TLM).The optimal LCP parameters were selected at a scanning speed of 100 mm/s and a laser energy of 23 lJ, by which an R s and q c value of 34 X/ٗ and 1.8 Â 10 À5 XÁcm 2 can be achieved, respectively. 18To fabricate the TLM structures, p-type wafers (>100 X cm) coated with plasma enhanced chemical vapour deposition (PECVD) SiN x (70 nm) on both sides were patterned with parallel LCP doped lines on one side of the samples.The samples were then separated into two groups.One group was coated with the SiO 2 /a-Si:H stack on the laser-processed side, which was formed by a short thermal oxidation in a clean quartz tube furnace at 800 C for 20 s ($1.5 nm) and then capped with a PECVD a-Si:H film ($30 nm).TLM patterns were defined for both groups using photolithography after evaporation of aluminium ($1 lm) on the laser-processed side.Currentvoltage measurements were performed on a Keithley 2425 Source Meter after annealing in forming gas atmosphere (FGA) at different temperatures for different times.The contact resistivity (q c ) was obtained from an extrapolation of resistance versus pad spacing.
N-type PERL cells with and without rear passivated contact to laser-LBSF were fabricated on 2.5 XÁcm FZ-Si wafers with a thickness of $200 lm.The cell size is 2 Â 2 cm 2 .Figure 1 shows the sketch of n-type PERL solar cells with and without passivated contact to laser-LBSF.The front side was textured with a random pyramid structure and has a boron diffused p þ emitter with a sheet resistance of $120 X/ٗ.The p þ emitter was passivated by an Al 2 O 3 /SiN x stack ($70 nm), which is formed by atmospheric pressure chemical vapour deposition (APCVD, Al 2 O 3 $15 nm) 19 and PECVD (SiN x , $55 nm).The front fingers were photolithographically defined and consisted of an evaporated stack of Cr/Pd/Ag further thickened by Ag electro-plating.The rear-side was passivated with PECVD SiN x ($70 nm).Linestructured LBSFs with a pitch of 1.0 and 1.5 mm were applied at the rear-side.A line width of $45 lm was obtained by LCP.Half of the cells were then subjected to a short thermal oxidation and capped with a-Si:H on the rearside.Finally, the rear-side of all the cells was metallized with an evaporated aluminum layer (2 lm), and an annealing step at 300 C for 30min in FGA was performed for contact formation.Light I-V tests were performed under standard AM1.5 illumination using an in-house system.The light source intensity of this system was calibrated using a certified reference cell from Franhaufer ISE CalLab.
A crucial parameter in the success of a low q c value for the SiO 2 /a-Si:H passivated contact is the annealing temperature, which will activate the alloying between a-Si:H and overlying aluminum film to create a high conductivity mixed-phase layer.The lowest temperature at which the a-Si:H/Al alloying, and hence the contact formation will occur, has been shown to be !250C. 16 A proper annealing temperature must be chosen so that the aluminium interacts with the a-Si:H to achieve a low q c value but not with the underlying thin SiO 2 to maintain a good passivation.
Figure 2 shows the q c dependence on anneal conditions for laser doped n þ regions with and without SiO 2 /a-Si:H stack.After annealing at 250 C for 30 min, an acceptable q c value of 5.5 Â 10 À3 and 1.8 Â 10 À3 XÁcm 2 could be achieved on the samples with and without the SiO 2 /a-Si:H stack, respectively.A higher q c value is observed on the sample with the SiO 2 /a-Si:H stack with the same annealing recipe.As the annealing temperature increases, the q c value decreases continuously in both cases.The low q c values are a result of the high surface concentration ($4.5 Â 10 19 cm À3 ) of the laser doped regions. 18We should note that a low q c value of 7.6 Â 10 À4 XÁcm 2 can be achieved after annealing at 300 C for 30 min, which is the exact annealing recipe we used for contact forming annealing.
With the decreasing q c values after annealing at higher temperatures, it is important to maintain the passivation performance simultaneously.Here, photoluminescence (PL) imaging with a short pass filter (1025 nm) was used to qualitatively monitor the passivation function of annealing conditions after each step, and the passivation quality is determined in terms of the implied open circuit voltage (i-V oc ) measured by quasi-steady-state photoconductance (QSSPC) technique.A short pass filter can minimize the effect of rear-side reflection after metallization. Figure 3 shows the PL images of a SiN x passivated sample after different processing at the rear-side.All the images were acquired with the same PL parameters from the front-side.The processing begins with a quarter silicon wafer  A sharp decrease of i-V oc (À73 mV) as well as I PL is observed after this step, which can be attributed to the high passivation layer opening fraction ($6%) at the rear as well as the laserinduced defects at the laser doped areas.After the implementation of the SiO 2 /a-Si:H stack at the rear-side, i-V oc value is improved to 670 mV accompanied by an obvious increase in the I PL value [Fig.3(c)].The large i-V oc gap (38 mV) between the initial sample in Fig. 3(a) and the sample in this step can be mainly attributed to the laser-induced damage, which cannot be recovered by re-passivation.After metallization, i-V oc value cannot be measured by QSSPC anymore.The I PL value change can be used to qualitatively monitor the passivation change after metallization and annealing.After rear metallization by aluminium [Fig.3(d)], only a I PL value decrease is observed.The I PL value increases to 16 000 after annealing at 250 C for 30min [Fig.3(e)], which indicates a slight rear passivation improvement.It can also be seen that the alloying between a-Si:H and aluminium at this step does not have any effect on the passivation quality of the SiO 2 /a-Si:H stack.Further annealing at 300 C for 30 min leads to a slight I PL value decrease, which indicates that the passivation of the SiO 2 /a-Si:H stack is stable at this temperature.Therefore, a simultaneous minimization of carrier recombination and contact resistivity to the laser doped n þ regions could be achieved with the SiO 2 /a-Si:H stack after annealing at 300 C for 30 min in FGA.
Although the SiO 2 /a-Si:H stack allows good contact to laser doped n þ regions after annealing, the recombination parameters (J o ) at the contact regions still might be higher than 100 fA/cm 2 due to the laser-induced defects.According to the simulations, a small contact fraction (3%-5%) should be applied to achieve a high efficiency with the implementation of passivated contacts if J o is relatively high and q c is small. 15,16Here, we applied the SiO 2 /a-Si:H passivated contact to n-type solar cells with laser-LBSF at a pitch of 1.5 and 1.0 mm, which corresponds to a contact fraction of 3% and 4.5%, respectively.Table I shows the results of n-type solar cells with and without the passivated contact to laser-LBSF.Without the passivated contact, the cell efficiency is mainly limited by a low V oc , which is caused by the high recombination at the laser doped regions.A relatively low efficiency of 19.7% and 19.4% is achieved with rear contact fraction of 3% and 4.5%, respectively.With the implementation of a passivated contact, the cell V oc , and hence the efficiency are improved significantly.A relatively high efficiency of 20.9% has been achieved with a contact fraction of 4.5%, which means that an absolute efficiency gain of 1.5% has been achieved with the implementation of passivated contact.However, a smaller absolute efficiency gain of 0.6% has been achieved with a smaller contact fraction of 3%.These cell efficiencies are mainly improved due to a V oc increase (up to 44 mV), which is caused by the reduced surface recombination at the laser doped regions by the implementation of the SiO 2 /a-Si:H passivated contact.Slight J sc improvement is also achieved at the same time.A small loss in FF has been observed for cells with the passivated contacts, which might be attributed to the higher contact resistivity of the SiO 2 /a-Si:H passivated contact.
In summary, we have implemented the SiO 2 /a-Si:H passivated contact to n-type solar cells with laser-processed LBSF.The cells efficiencies have been improved with the implementation of the passivated contact concept, which is mainly attributed to the reduced surface recombination at the laser doped regions.An absolute efficiency gain of up to 1.5% has been achieved with the implementation of the SiO 2 /a-Si:H passivated contact.This study demonstrates that the efficiency of solar cells with partial rear contact can be improved with the implementation of the passivated contact concept.

FIG. 1 .FIG. 2 .
FIG. 1.The structure of the n-type PERL solar cell (a) without and (b) with a passivated contact to the laser-LBSF.

( 2 . 5
XÁcm) passivated by SiN x with an i-V oc value of 708 mV [Fig.3(a)].PL intensity (I PL ) and/or i-V oc values after each step are shown in the figure.All the I PL values were collected from the area marked by black dashed lines.Parallel LCP doped lines with a pitch of 0.75 mm are applied to form a 30 Â 30 mm 2 area at the rear side [Fig.3(b)].

FIG. 3 .
FIG. 3. PL images of SiN x passivated sample after different processing at the rear-side.All the images were acquired with the same PL parameters.I PL and/ or implied open circuit voltage (i-V oc ) values after each step are as follows: (a) SiN x passivated sample i-V oc : 708 mV and I PL : 51 000; (b) After LCP, 0.75 mm pitch i-V oc : 635 mV and I PL : 5000; (c) After SiO 2 /a-Si:H deposition i-V oc : 670 mV and I PL : 15 000; (d) After Al metallization I PL : 14 700; (e) After FGA at 250 C 30 min I PL : 16 000; and (f) After FGA at 300 C 30 min I PL : 15 600.

TABLE I .
Results of n-type silicon solar cells with and without the SiO 2 /a-Si:H passivated contact to laser-LBSFs.
a PC: Passivated contact.