氮硅对硅片背面的钝化-M.Hofmann-Fraunhofer.PDF

SILICON NITRIDE – SILICON OXIDE STACKS FOR SOLAR CELL REAR SIDE PASSIVATION M. Hofmann, E. Schneiderlchner, W. Wolke, R. Preu Fraunhofer Institute for Solar Energy Systems ISE Heidenhofstr. 2, 79110 Freiburg, Germany phone 49 761-4588-5360; fax 49 761-4588-9250 email marc.hofmannise.fraunhofer.de ABSTRACT Plasma-enhanced chemical vapour deposited PECVD amorphous silicon nitride films SiN x are well known in the photovoltaic community for their good optical and electrical properties when being deposited as anti- reflection film on the solar cell’s front and their low-temperature deposition process. The development of high- efficiency solar cells brings up the need for passivating films for the solar cells’ rear side when being contacted only pointwise e.g. the passivated emitter and rear cell concept PERC. Compared to a thermally oxidised solar cell’s rear side, silicon nitride already showed good but lower results in passivation and optical reflection quality in the past but it offers benefits in deposition costs, process time and heat load for the solar cells. In this work, a new surface passivating stack system consisting of a PECVD-silicon nitride and a PECVD-silicon oxide layer is presented that offers optical and electrical properties very close to those of thermal silicon dioxide without the drawback of heat load and with the benefit of using a quick and cheap process at a low temperature 350C. By sintering lifetime evaluation samples with the new stack system, the dependence of the thermal stability of the passivation on the thickness of the silicon oxide is shown. Furthermore, solar cells featuring this new stack system on their rear side are fabricated that show the expected optical and electrical benefits with a maximum solar cell efficiency of 19.3 . Keywords passivation, silicon nitride, PECVD. 1 INTRODUCTION The major goal of research in crystalline silicon solar cell technology is to reduce the price per wattpeak. Therefore, new cell designs and new production processes have been developed in the past. Widely used in laboratories around the world is the high-efficiency PERC passivated emitter and rear cell concept [1]. Due to its relatively high production costs, this cell concept has not made the step into solar cell mass production. By the use of the Laser-Fired Contacts LFC approach [2], an economically feasible technology for mass production of the rear contacts of PERC cells has been introduced. Therefore, rear side surface passivation schemes need to be adapted to the LFC process. Furthermore, another possibility to reduce the cost per wattpeak is to lower the wafer thickness. This leads to an increased surface-to-bulk-ratio and to an even stronger interest in passivating both solar cell surfaces. In other words, the solar cell’s rear side becomes more and more important. As thermally grown silicon dioxide SiO 2 layers are manufactured by a long-time energy-intensive high temperature process 1000 C [3], [4], they are not the first choice for mass production, although they provide a very good and thermally stable passivation. Plasma-enhanced chemical vapour deposited PECVD hydrogenated amorphous silicon nitride a-SiN x H; referred to as SiN x in the text films are fabricated by a low temperature processing step 300-400 C. They also provide a very good passivation quality [5], [6] but are not as thermally stable as the SiO 2 layers [7]. As the nitrogen in SiN x layers is suspected to be important for the formation of fixed positive charges in the SiN x film [8], [9] which leads to an inversion layer at the silicon surface [10], we used very silicon-rich SiN x films to reduce the risk of a current loss while contacting the inversion layer and the p-Si bulk with LFCs. To our knowledge, no PECVD-silicon oxide SiO x films have been developed so far that achieve good surface passivating qualities. In this work, SiN x films and SiN x -SiO x stacks, all PECVD deposited, are investigated as a rear side passivation layer of LFC contacted solar cells. 2 SAMPLE PREPARATION 2.1 Carrier lifetime evaluation samples In this work, symmetrical carrier lifetime evaluation samples, consisting of boron-doped float-zone FZ wafers with a resistivity of 1 Ω cm and a thickness of 250 m have been used to investigate a novel surface passivation stack system. Starting with a surface cleaning procedure using a dry plasma etch process, a PECVD-a- SiN x H layer was deposited followed by a PECVD-a- SiO x deposition. The SiO x deposition took place immediately after the SiN x deposition by only changing the process gases and electrical parameters. Next, the second side of the wafer was treated the same way. The plasma etch step was performed using a microwave SF 6 bias-free plasma in a reactor of Sentech SI. The SiN x and SiO x layers were deposited in a direct PECVD reactor of Surface Technology Systems STS. To create the SiN x layer, a gas mixture consisting of silane SiH 4 , nitrogen N 2 and hydrogen H 2 , for the formation of the SiO x film a gas mixture of nitrous oxide N 2 O and hydrogen H 2 was used. The silicon nitride used was very Si-rich n 2,9, which is quite close to amorphous silicon but still silicon nitride. The plasma etch step and the SiN x deposition always have been the same process. The SiO x deposition process was varied only in the deposition time in the range of 15s to 1min. In this way, the oxide thickness was changed. As the silicon oxide deposition took only very little time and its deposition can take place in the same 19th European Photovoltaic Solar Energy Conference, 7-11 June 2004, Paris, France 1037 reactor as the SiN x deposition, the cost impact can be assumed to be small. To characterise the as-deposited surface passivation quality, quasi steady state carrier lifetime was measured by use of a WCT-100 lifetime tester [11]. Sintering trials in a quartz tube furnace at 425 C in forming gas ambience for 15 min each including the ramping up time which leads to assumed 5 min at 425 C were performed with a measurement of the carrier lifetime after each sintering step. See Table I for a quick overview of the rundown of the processing and Fig. 1 for the scheme of the carrier lifetime evaluation samples. Table I Process sequence lifetime samples. Fig. 1 Carrier lifetime evaluation sample structure. 2.2 Solar cells To show the potential of the new surface passivation scheme for photovoltaic energy conversion, solar cells of 3.85 cm 2 were produced using boron-doped 1Ω cm float- zone FZ wafers with a thickness of 250 m. The solar cells’ front features a high efficiency RP-PERC structure with thermal SiO 2 -passivated emitter, photolithographically defined front contacts and random- pyramid front surface. The solar cells’ rear side has been cleaned by the same plasma etch step and the same SiN x - SiO x stack system was deposited as on the lifetime samples. Cells without SiO x or with a SiO x thickness of 100 nm were produced and a slight variation of the SiN x deposition parameters was carried out. Front contact Random pyramids Th. SiO 2 Emitter p-Si Local Al-BSF SiN x SiO x Fig. 2 Random pyramid passivated emitter and rear cell RP-PERC with rear side stack passivation system p- Si/SiN x /SiO x /Al and laser-fired rear contacts LFC. As reference samples, solar cells with rear side passivation layers of thermal silicon dioxide SiO 2 were manufactured. Next, the SiO x respectively the SiO 2 layer was covered with a 2 m aluminium film by electron beam evaporation. In the following step, laser-fired contacts processing LFC was applied [2] to create rear side point and line contacts. Finally, the front contacts were electroplated. After measuring current-voltage I-V and internal quantum efficiency IQE including reflection, annealing of the solar cells was performed at 425C in forming gas for 25 min. Afterwards a second measurement was done. Table II Process sequence solar cells. Cutting of the wafers using a laser system SiN x -SiO x stack deposition on first side Plasma etch surface cleaning of first side Plasma etch surface cleaning of second side SiN x -SiO x stack deposition on second side Lifetime measurements Sintering Lifetime measurements, sintering, lifetime meas., . Solar cell’s front side creation SiN x -SiO x stack deposition on rear side Plasma etch surface cleaning of rear side Al deposition on rear side Laser-fired contact formation rear side Electroplating front contacts I-V/ IQE/ reflection measurement Annealing I-V/ IQE/ reflection measurement Laser-fired contact LFC Al 0-200 nm SiO x 70 nm SiN x 250 m p-Si 70 nm SiN x 0-200 nm SiO x 19th European Photovoltaic Solar Energy Conference, 7-11 June 2004, Paris, France 1038 0 5 10 15 20 25 0 200 400 600 800 1000 1200 1400 1600 200nm SiO x 100nm SiO x 50nm SiO x 0nm SiO x L i f e tim e [ s] Sintering time [min] Fig. 3 Carrier lifetime evaluation sample results. 3 RESULTS AND DISCUSSION 3.1 Carrier lifetime evaluation samples The as-deposited carrier lifetimes of all lifetime evaluation samples showed a very high range of about 1100 s to 1300 s, independently of the thickness of the silicon oxide layer; see Fig. 3. For the performed incremental sintering the results can be split into two phases and two groups of samples. Phase no. 1 includes the first 5min of the sintering procedure; phase no. 2 is made up of the following sintering steps. In phase no. 1, the sintering shows a slightly positive effect on the lifetime of the samples with an oxide thickness of 100 nm and 200 nm; the samples with lower thickness almost did not vary. In phase no. 2 after the next sintering step accumulated 10 min the samples with silicon oxides of a thickness of 50 nm and without any oxide showed a strong decrease in carrier lifetime. Only about half of the lifetime of the previously measured value could be reached 1250 s 550 s; 1100 s 400 s. Furthermore, these samples showed a continued decrease in lifetime with the sintering time to around 100 s. Phase no. 2 of the samples with 100 nm oxide starting at a sintering time of 5 min is characterised by a steady decrease in lifetime that does not show the heavy drop between 5 min and 10 min like the samples with a lower oxide thickness. This steady decrease led to around 800 s after 25 min of sintering. Phase no. 1 of the samples with 200 nm of plasma silicon oxide is set to 0 min to 10 min sintering time – in contrast to the other samples – because of the measured rising of the carrier lifetime. After the lifetime peak, the lifetime is as well decreasing but on a much lower level than the other samples. This yields a lifetime after 25 min of sintering of 1200 s. The measurement results lead to the conclusion that a plasma silicon oxide layer of sufficient thickness on top of a surface passivating plasma silicon nitride film is very beneficial for the thermal stability of the carrier lifetime. A SiO x thickness of 100 nm or 200 nm can be called sufficient. This depends on the subsequent production steps and their heat load on the wafer. As a reason for the decrease in carrier lifetime that all samples showed can be assumed that significant parts of the – for the surface passivation very beneficial – hydrogen of the hydrogenated silicon nitride layer is lost as the sintering time is rising. Also Lauinger et al. [12] have found a strong negative influence of an annealing step on the passivating quality of single layer SiN x films. 3.2 Solar cells Solar cells 250 m, p-type, FZ, 1 Ω cm with a high- efficiency front RP-PERC were fabricated with different rear side passivation systems to compare the new passivation stack system to the single layer passivation films of PECVD-SiN x and thermal SiO 2 . In order to give rise to the energy conversion efficiency, the rear side non- Si absorption should be low therefore, the reflection should be high. In wavelength dependent reflection measurements the optical rear side behaviour can be seen in the range starting from 1000 nm. Fig. 4 shows that the rear side reflection of the thermal SiO 2 reference and the SiN x -SiO x stack system are almost identical and on a high level whereas the single layer SiN x passivation film shows a significantly lower reflection. Looking at the internal quantum efficiency IQE plots in Fig. 4, thermal SiO 2 shows the best behaviour with the SiN x -SiO x stack system being very close. This leads to the conclusion that not only the reflective but also the passivating properties of both passivation systems are of a quite comparable quality. In contrast, the SiN x layer cell shows a much lower high- wavelength IQE. This fact can be explained with a lower passivation quality of the SiN x layer. 300 400 500 600 700 800 900 1000 1100 1200 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0 After annealing iqe 4.7 SiN x iqe 6.4 SiN x -SiO x iqe 2.1 th. SiO 2 reflection 4.7 SiN x reflection 6.4 SiN x -SiO x reflection 2.1 th. SiO 2 I n t e r n a l qua nt um ef f i c i en c y , r e f l ect i on Wavelength nm Fig. 4 Wavelength dependency of the internal quantum efficiency IQE and reflection R measurement values of solar cells with different rear side passivation schemes. The optical influence of the rear side can be seen at wavelengths above 1000nm with a strongly increasing reflection. An increased rear side reflection is beneficial for the energy conversion efficiency because the potential light path through the silicon is doubled and therefore the likelihood of photon absorption is increased. The IQE at higher wavelengths starting from 800 nm is an indicator for the passivation quality of the rear surface. Taking a look at the solar cell parameter results see Table III shows that the surface passivation has a significant influence on the open circuit voltage V OC [4]. Consequently, the thermal SiO 2 solar cells show the best rear side passivation. This is consistent with the conclusion of Fig. 4. In consistency with Fig. 3 are the ranges of V OC of the single layer SiN x and the stack 19th European Photovoltaic Solar Energy Conference, 7-11 June 2004, Paris, France 1039 system SiN x -SiO x . The sintering step 425 C did not show a uniform result when looking at the SiN x solar cells Sample 7.3 very Si-rich SiN x shows a decrease in V OC but an increase in fill factor FF. This leads to the interpretation that the aluminium on top of the SiN x destroyed parts of the passivation effect by improving the Al contact to the silicon. The short circuit current J SC did not change significantly. On the other hand, Sample 10.3 Si-rich SiN x showed almost the opposite behaviour increase in V OC , decrease in J SC and FF Comparison of the two stack systems with SiN x and SiO x samples 6.4 and 10.4 shows the same behaviour for both samples increase in V OC , J SC and decrease in FF