SOLAR CELL

Effect of Junction Depth on the FF of Monocrystalline Si Solar Cells
As discussed in Section 2, most conductor pastes contain a small amount of glass fiit. The frit serves to improve adhesion to the substrate by conforming to the surface topology.
Additionally, for Si substrates, the frit etches a small distance into the Si material. If the firing process is too aggressive, the glass fiit along with the metal particles will begin to encroach on the n+p junction. This encroachment manifests “itself as decreased ~ and increased J02. As
indicated by the modeling results in Fig. 1.2, low ~ and high J02 can destroy the device fill factor.
In this section, the importance of junction depth on the quality of SP contacts is explored. A set of phosphorus diffusions was carried out using ceriurn pentaphosphate solid sources. The diffision time was fixed at 30 minutes, and in each case the peak temperature was varied. The resulting sheet resistances were in the 40-90 Wsq range, and the junction depths are shown in  Fig. 1.7.

The data in Fig. 1.6 was used to model and compare the power loss expected for solar cells with pure Ag contacts (1.6 j.dlcm) and SP Ag contacts (3.5 pf2-cm, 700”C hotzone, 30 sec dwell time). The following device parameters were used for simulation purposes: solar cell active area of 2 cm by 2 cm, 8 grid fingers, a single tapered bus bar, and a 40 Wsq emitter sheet resistance. The width and height of each finger were fixed at 130 pm and 8 pm, respectively (tYPical values for screen-printed SOlar cells). The simulations show that the increased metal resistivity of SP Ag compared to pure Ag leads to an ~ri~, increase of 0.12 C?-cm2 and an additional power loss of 0.14 mW/cm2. In other words, SP fill factors are inherently lower than those of a pure Ag metallization by approximately 0.010 due to higher p~,ti and ~,ti~,.

Effect of SP Firing Treatment on Conductor Paste Resistivity
The conductor paste used in this work was made by Ferro Corporation (3349 Ag Conductor).
After printing, the following procedure was used to form the contacts. First, the solvents were removed by baking on a hotplate at 150°C for 2 minutes. This was followed by firing in a 3-zone IR-belt furnace in which the lengths of zones 1,2, and 3 were 7.5”, 15“, and 7.5”, respectively.
The first two zones were set to 425°C and 580”C and used to burn off organic materials in the printed paste. The hotzone (zone 3) temperature was varied to suit the particular investigation.
The overall firing time was determined by the beltspeed through the fi.umace. Beltspeeds of 15“/min and 40’’/min were implemented in this study, which correspond to hotzone dwell times of 30 seconds and 11 seconds, respectively.
First, the Ag resistivity was determined so that basic model calculations could be performed.
(It is instructive to note that the resistivity of pure Ag is 1.6 p~-cm.) As shown in Fig. 1.6, this parameter is a fi.mction of hotzone firing temperature. In fact, the resistivity changes by more than a factor of 2 (from 5.3 to 2.2 pf2-cm) for a hotzone temperature swing of 300”C.

firing cycle and the material qualities of the conductor paste. If the overall resistance becomes excessive, then the solar cell fill factor will be lowered. The junction leakage and shunting behavior depend primarily on the junction design and the contact firing cycle. If the junction is compromised during the fting cycle, the lowered shunt resistance and increased junction leakage will cause severe fill factor degradation. The impact of series resistance ~~ri~,), shunt resistance (~~u~~, and junction leakage (JOZand nz) on device fill factor can be simulated numerically using the solar cell equivalent circuit model shown in Fig. 1.2. (The JOZdiode and its corresponding ideality factor model the effect of junction leakage via depletion region recombination.) This equivalent circuit was employed together with a device simulator (PCID-#) to model the fill factor change as a fiction of ~,ri~,, ~hunv ‘d ’02- ‘e ‘esults (Fig” 1 ‘3-Fig. 1 ‘5) cm be ‘Seal ‘o ‘Omulate ‘he ‘O1lOwing guidelines ‘or
attaining high fill factor: ~~unf>lOOO !2cm2, ~ti.,<0.50 Q-cm*, and JOZ<l0-8 A/cm2. In the following sections, the experimental behavior of screen printed metallization in the context of these parameters is presented. Different characterization techniques, such as diode (dark) IV,
solar cell lighted IV, contact resistance, and conductivity analysis, are used to extract the parameters which govern fill factor response.

One of the most difiicult aspects of large scale solar cell production is forming high-quality front contacts. The metallization techniques used in laboratory settings (which involve vacuum  evaporation, lift-off photolithography, and plating) are too time consuming and impractical for large scale application. On the contrary, screen-printing (SP) offers a simple, cost-effective
contact method that is consistent with the requirements for high-volume manufacturing. The problem with SP, however, is that the throughput gains are attained at the expense of device performance. The losses associated with SP metallization fall into three categories: 1) increased minority carrier recombination in the required heavily doped n+ regions, 2) increased shading due to wide grid fingers (> 100pm), and 3) fill factor degradation due to poor contact quality. The purpose of this section is to provide a detailed study of the third issue: contact quality. This is important because contact quality determines the device fill factor, and therefore, affects the
overall cell efficiency (q= ?’Oc-Jsc-IV’). Though high fill factor petiormance has been demonstrated in the past with SP [I], most commercial solar cell processes which implement this technology result in relatively low fill factors (= 0.750) [2]. No comprehensive study has been conducted to isolate the causes for low fill factor in SP cells and relate them to specific process
conditions.

Finally, the individual rapid and potentially low-cost processes are integrated to form high efficiency devices. RTP solar cell efficiencies of 17’XOand >19Y0 are achieved on monocrystalline silicon with screen printed and photolithography contact, respectively.

Rapidly formed screen printed cells in a commercial beltline machine also resulted in 17°/0 efficient cells on monocrystalline silicon and multicrystalline string ribbon material.
4.9V0 efficient cells.

Evergreen solar. Solar cells were fabricated with photolithography contacts as well as screen-printed contacts. Solar cells fabricated with phosphorus and aluminum gettering and FGA hydrogenation showed an increase in efilciency of 1.2°/0 (absolute) over cells with the same gettenng treatments but without FGA hydrogenation. Without the gettering treatments, FGA had little effect on the bulk lifetime. Cells processed with conventional furnace processing and photolithography contacts had an average efficiency of 14.6V0 with a maximum of 15.4%. A lifetime study of the optimization and application of beltline gettering and passivation techniques indicates that lifetimes over
50 ps are achievable even though the as-grown lifetime values are only about 1 ps. The first 100 pm thick fully screen-printed cell with a bekline diffksed emitter (BLP) of 45 Q/ produced efficiencies as high as 10.9Yo. The main loss components of the screenprinted devices are in the blue response and low shunt resistance. The shunt resistance of screen-printed devices was increased from 200 f2-cm2 to over 5000 f2-cm2 by implementing a spike in the contact firing profile. An increase in the red response resulted in cells that were spike fired and may be due enhanced bulk hydrogenation from
the SiN film. Cell efficiencies as high as 14.9V0 were achieved on 250 pm substrates using beltline processing and screen-printing.
A novel simultaneous boron and phosphorus dlfision technique is presented to produce simple, high efficiency n+ pp’ silicon solar cells in one thermal cycle. This tecluique uses boron and phosphorus spin-on dopant films to fabricate limited solid doping sources out of dummy silicon wafers. This approach results in the delivery of a fixed dose of P205 or B2 03 to the diffused sample.

Screen-printing and rapid thermal annealing have been combined to achieve
an aluminum- alloyed back surface field (A1-BSF) that lowers the effective back surface recombination velocity (S,ff) to approximately 200 cm/s for solar cells formed on 2.3 Q-cm Si. Analysis and characterization of the BSF structures show that this formation process satisfies the two main requirements for achieving low S,~ 1) deep p+ regions and 2) uniform junctions. Screen-printing is ideally suited for fast deposition of thick Al flms which, upon alloying, result in deep BSF regions. Use of
a rapid alloying treatment is shown to significantly improve the BSF junction
uniformity and reduce S.fi. The A1-BSFs formed by screen-printing and rapid
alloying have been integrated into both laboratory and industrial-type fabrication sequences to achieve solar cell efficiencies in excess of 19.OVO and 17.oYo, respectively, on planar 2.3 Q-cm float zone Si. For both process sequences, these cell efficiencies are 1-2% (absolute) higher than analogous cells made with un-optimized A1-BSFs or highIy recombinative rear surfaces.

atomic hydrogen from the SiN film to the Si-Si02 interface, thereby reducing the density of intetiace traps at the silicon surface. Compatibility with this post-deposition anneaI makes the stack passivation scheme attractive for cost-effective solar cell production where a similar anneal is required to form screen-printed contacts. Model calculations are also performed to show that the RTO+SiN surface passivation scheme may lead to greater than 17%-efficient thin screen-printed cells even with a low bulk lifetime of 20 p.s.