In the beginning, there was nothing.
At least there was no dielectric passivation applied to the silicon solar cells from the early production lines of the 1980s, an era when PV production companies were still thinking in terms of MWp rather than GWp factories.
The silicon solar cells produced in these companies were based, as is surprisingly still the case for many cells 2 decades later, on Al-BSF (Back Surface Field) technology with an internal electrical field passivating the rear surface. On the front surface, a TiO2-layer was the most commonly applied dielectric deposited by either spray-on coating or Atmospheric Pressure Chemical Vapor Deposition (APCVD). This layer’s main function was to serve as Anti-Reflection Coating (ARC) thanks to its favorable refractive index and limited parasitic absorption. Passivation, either surface or bulk, was not part of its repertoire.
Surface passivation was also not required as the emitters used in these solar cells were heavily doped in order to allow a good contact to the screenprinted Ag metal grid. It was also an era when a “firing-through” process was still a distant dream. Instead, the TiO2 layer was often applied after the firing of the contacts and therefore a masking or brushing of the ARC-layer on the busbars was necessary to allow soldering for interconnection.
First dielectric industrial revolution
But in the 90s, a 1st dielectric industrial revolution occurred that marked the end for TiO2. Several companies and institutes (Mobil Solar, Kyocera, Imec, UNSW, …) developed a process based on a silicon nitride coating that incorporated many advantages.
Two distinctive versions appeared: a low temperature, plasma-based (Plasma Enhanced CVD) and a high temperature version (Low Pressure CVD). While the latter was used commercially by BP Solar for the Saturn Laser Grooved Buried Grid technology, it was the PECVD version that proved the most successful and versatile in the end.
The main reason for this revolution can be summed up in 2 words: hydrogen passivation. The hydrogen contained in the PECVD-SiNx:H layer is released during the firing of the screen-printed contacts, allowing passivation of the emitter surface and of the bulk of the material. Especially for lower quality multicrystalline silicon material, it was a match made in heaven as cell efficiencies increased by a very significant amount (1 – 1.5% absolute or 10% relative).
The fact that also the process was simplified (no brushing, no masking) by firing of the printed contacts through the SiNx:H layers and simultaneously achieving a superior passivation and an Al-BSF on the rear, with all of this in one fast firing step, was the cherry on the cake. The available surface passivation also allowed more moderately doped emitters to be used, which in turn started an evolution in the development of Ag pastes that are able to contact high ohmic emitters, an evolution which continues to this date. In the end, PECVD-SiNx:H only had one real disadvantage working against it : the fact that it was a vacuum deposition. But history has shown that the benefits far outweighed that issue and many system manufacturers have commercialized PECVD-deposition systems in different forms (batch vs inline, remote vs direct plasma, …) over the last 2 decades.
No text about dielectric passivation for solar cells can be written without mention of silicon oxide. Indeed, thermal oxidation has been a very important instrument to improve the efficiency level in the R&D labs and to push the boundaries of the cell technology. The most striking certainly example is the 25.0% PERL cell made by UNSW with thermal oxide passivation on front and rear of the cell (and no SiNx:H to be seen).
While thermal oxide has always been at the forefront of solar cell technology development, it has up to now made less headway in commercial cells.
While the PERC/PERL-cell technology already appeared in the early 90s, it is only very recently that the technology is finding widespread acceptance in pilot- and production lines.
Second dielectric industrial revolution
However, it is not thermal oxide leading this 2nd dielectric industrial revolution but another high quality dielectric: Al2O3. Its most striking property can once again be described by 2 words : negative charges. This inherent property avoids one of the pitfalls of the PERC-technology for p-Si cells. Indeed, a rear dielectric passivation with positive charges results in the creation of an inversion layer on p-Si which can lead, if not properly controlled, to local loss of generated carriers by shunting with a resulting impact on Jsc, Voc and FF of the device. This detrimental effect increases in relative importance towards lower illumination levels, potentially resulting even in a lower efficiency than a sister cell based on Al-BSF technology. But a negatively charged dielectric creates an accumulation layer instead of an inversion layer and avoids this particular problem.
While Al2O3 was already used in the 1980s for Metal-Insulator-Semiconductor (MIS) solar cells, a renewed interest arose when the potential use for p-Si PERC-type cells of this negatively charged dielectric was realized by researchers at various European institutes (Imec, Eindhoven University of technology,…).
Once again, finding the appropriate system for industrial application was thought to be tricky but over the years, system manufacturers in cooperation with several research institutes have brought Al2O3 deposition systems to the market with a variety of choices: Atomic Layer Deposition (ALD) in several versions (thermal, spatial, plasma assisted), PECVD, sputtering, APCVD and even screen printing. Spatial ALD and PECVD are currently “leading the charge” in this field.
The former due to the high material utilization efficiency, the high quality of the layers and the limited thickness needed, while PECVD has an advantage when it comes to the ease of upgrading existing production lines, many which already incorporate PECVD-SiNx systems. General expectations are that in the coming years an increasing percentage of new and old production lines for p-Si solar cells will be build or upgraded based on Al2O3-based PERC-type technology.
Another expected trend according to the International Technology Roadmap (ITRPV 2014), besides the continued high market share of p-Si multicrystalline cells, is a gradual shift from p-Si to n-Si monocrystalline cells. Also here, Al2O3 will find its place since the Boron doped emitters used for these cells will require an excellent surface passivation. In-house measurements at imec have shown a drastic decrease in surface recombination velocity for Al2O3 compared to thermal SiO2. This is true to such an extent that recombination in these Al2O3-passivated emitters is determined by Auger recombination while the surface component becomes negligible.
A prime example of this are the n-Si rear junction PERT (Passivated Emitter and Rear Totally diffused) which imec is developing together with its partners in Imec’s Industrial Affiliation Program for Photovoltaics. A cross section of such a cell, with Ni/Cu plated front contacts, is shown in the attached figure. Applying the excellent rear surface passivation that is provided by Al2O3, a top efficiency of 21.5% was reached (average 21.3%) and independently confirmed by ISE CalLab.
While PERC-type cells will take an increasing slice of the market in the coming years, a new dielectric revolution might already have announced itself and this time, it might be silicon oxide‘s turn to be at the forefront.
The reason in a 2-word summary: passivated contacts. Indeed, the recombination currents related to the silicon bulk and the physical surfaces have drastically decreased by improved silicon material quality and excellent surface passivation schemes for traditional high efficiency cells.
Now it is often the metal contact area which is dominating the recombination current and open-circuit voltage of these cells. This is not always due to the presence or insufficient shielding of the metal itself but can also be caused by the damage induced to define the metal-semiconductor interface (e.g. damage from laser ablation on the dielectric to define the contact area). This can have a large effect on the recombination of generated minority carriers.
What is needed is a selective contact scheme which forms a passivating barrier for these minority carriers without forming a resistive barrier for majority carriers. One method, besides heterojunction based cells, to form such a passivated contact is using a tunneling dielectric in combination with a thin film emitter/BSF. An ultrathin oxide layer (~1 nm) , formed by chemical means or more classically, by a moderate temperature thermal oxidation, is a possible means to achieve this.
One company turning heads with this technology is Silevo, an American start-up recently acquired by SolarCity. They are showing lots of ambition, with plans for a 1 GWp factory in New York State, and excellent results up to 23% with the Triex cell concept. And as the cornerstone of their technology, they are relying on a very thin silicon oxide layer on both sides of the cell.
If this is a glimpse of the future, then the passivating dielectric could be once again close to nothing in thickness (but not in importance).