Being bombarded

“Higher efficiency, lower cost” has been and will remain the mantra of the PV industry. Even if module prices have reached unprecedented lows over the years, manufacturers, through the value chain, have sought to further reduce their costs to vie for the prize of grid parity and that which lies beyond. The upstream developers are working on not just building more intelligent systems with better technologies but also further reducing wafer sizes and cutting steps to lower costs.
The production value chain involves a high number of complex steps to bring silicon from its raw form into a gleaming module mounted and generating power. One very crucial step is that of emitter formation. When looking at the workhorse of PV crystalline silicon, without the emitter step, silicon by itself is rather banal as it is not the best conductor. The emitter, or the p-n junction, forms the core of crystalline silicon solar cells.
By doping, in every sense of the word introducing impurities, silicon transforms into a more efficient conductive device. A large percentage of the silicon cells in the market are produced using high temperature diffusion of dopants into the crystal lattice.
P-type doping creates an abundance of holes in the lattice. They are usually acceptor dopants like boron. N-type doping involves donor dopants that create an excess of electron charge carriers, like phosphorous. These p-n junction sandwiches can be thought of as the building blocks of most crystalline solar cells.
Diffusion has been the widely used method for doping. The p-type dopant, boron, is introduced during silicon processing after purifying for PV production use.
Phosphorous diffusion in a tube furnace is a commonly used method in the industry for the creation of the n-type. This diffusion involves the wafers being loaded on quartz boats and then being diffused with a liquid phosphorous oxychloride (POCl3 ) in furnaces before being heated. The temperatures for heating can reach 900 degrees Celsius. By-products of the reaction are phosphorous silicate glass (PSG) and silicon dioxide (SiO2 ). Phosphorous then forms the n-type emitter.
Leading PV equipment manufacturers like centrotherm photovoltaics and Schmid Group have developed furnaces for such diffusion applications.
Ion implantation is another solution that has been around for a while now but only recently has gained ground in PV with Applied Materials bringing to the market their Solion technology.
Vice President and General Manager, Applied Materials Varian, Jim Mullin explains, “Ion implantation is a potentially disruptive technology that has been proven to be production-worthy in the semiconductor industry over the last 30 years and provides the significant benefits of high cell efficiency and process simplification in PV manufacturing.”

Bombarding or diffusing

Ion implantation has been used in the semiconductor sector as mentioned before. This method basically does the same job as diffusion; modify the characteristics of the surface layer of the substrate, the silicon, by introducing dopants.
In a nutshell, ion implantation refers to the bombardment of a substrate by a beam of highly energized ions. The accelerated ions penetrate the solid surface, in this case, the silicon surface. An ion source holds the ions of the desired elements that need to be produced for the bombardment. An accelerator electrostatically accelerates the ions to a high energy level. A target chamber is where the ions impinge on the target. Currently ion implantation is gaining ground, touting great potential for the efficient formation of the p-n junction for crystalline silicon cells.
Suniva Inc. and the Georgia Institute of Technology’s paper “High-throughput ion-implantation for low-cost high efficiency silicon solar cells” states that ion implantation actually provides a unique opportunity to obtain grid parity because it simplifies the fabrication of high efficiency advanced cell structures by reducing processing steps, complexity and costs. There are some reasons cited by Mullin as being the winning factors for ion implantation. Additionally, pv magazine spoke with Dr. Giso Hahn, Professor at the University of Konstanz’s physics department on this technology’s plus points and possible issues to consider.

Factors to consider

Homogeneity: With the implementation of ion implantation, a higher degree of dose uniformity to produce very uniform single side junctions can be achieved compared to diffusion, Applied Material states. The company’s Solion technology claims that precise uniform doping can be achieved. Hahn says, “You can make more homogeneous doping profiles. That is mainly because it is a different way of incorporating the doping atoms into the silicon. During ion implantation you bombard the surface with the ions, and so the highest concentration of the ions will be somewhere below the surface, because they will be stopped only inside the material and not on the physical surface.” Now the ions are bombarded onto the surface and so there will be damage. Annealing is a step required to repair such lattice damage.
Hahn explains, “The opportunity arises in the second annealing step, which you always need because you create a lot of damage via the ion bombardment and this damage has to be annealed, so you need a high temperature step. During this high temperature step, in combination with the dose, you have the freedom to create profiles which are very homogeneous and could look in principal different from the ones of conventional techniques like POCl3 diffusion.” Mullin adds, “Ion implantation precisely places the exact number of dopant atoms at the correct depth with exceptional uniformity. Ion implantation is an energetic process that uses high voltage and electrostatic optical elements to accelerate and steer the beam. Additionally, Applied/VSE ion implanters utilize a state-of-the-art control system to measure the ion beam composition in real time and adjust as necessary to ensure the emitter or field layer is created without defects and is exactly uniform. Applied/VSE ion implanters can deliver less than 3% uniformity performance and less than 2% repeatability performance. This allows our customer to dramatically improve their cell binning performance. Some of our customers are showing a 2-3 times improvement in binning using Solion.” Homogeneity is definitely an issue when there are low doping levels. And when ion implantation is used, better uniformity is achieved but higher temperatures are necessary. “You always need this high temperature annealing step. These temperatures are higher than those usually required in standard diffusion methods,” says Hahn.
Can all these PV materials withstand such high temperatures? Materials like monocrystalline, with their high silicon purity, do not seem to have this problem. Hahn says that with the use of multicrystalline material, this might still be an issue because of the higher concentration of point defects, the existence of extended defects and their interaction at high temperatures.
Mullin explains however that implantation processes in production today have anneal temperatures similar to POCl3 diffusion. He says, “In fact, ion implantation can enable shorter processing time in the furnace because the activation efficiency for implanted phosphorous is better than diffused phosphorous.”
Gettering: Gettering is employed to remove the device degrading impurities from the wafer. Impurities vitiate electrical properties. During this phosphorous diffusion step there is a gettering step to extract impurities from the bulk to the surface.“But only if there is a sink for these metal impurities,” Hahn elaborates. “That is the case if you have standard diffusion because you have a very high concentration of phosphorous dopants at the wafer surface. For example, silicon phosphite precipitates form close to the surface. This is not built-up in ion implantation.” Gettering impurities are important, especially for multicrystalline materials. Hahn asks whether better quality with ion implantation is offset by the theoretical lower ability of gettering impurities which is quintessential especially for multicrystalline materials.
Does this mean that multicrystalline materials are just not well suited to ion implementation?
Mullin answers saying that the advantage with ion implantation is that the dopant profile can be specifically tailored to achieve maximum efficiency. “A traditional diffusion process only has two ‘knobs’ (time and temperature) for controlling dopant concentration, depth and gettering efficiency. Ion implantation allows for specific control of dose rate, uniformity of dopant, and the amorphization rate of the silicon. These additional ‘knobs,’ combined with the ability to change activation temperature in the furnace, improve both emitter quality and gettering efficiency. Ion implantation has been tested with multicrystalline cells and proven [to have] equal or better efficiency across a wide range of substrate qualities,” he explains.
Edge isolation: With the advent of ion implantation, the process steps are also reduced, as Applied Materials reports (see graphic “Cell process flow” above). Edge isolation is the removal of phosphorous diffusion around the edges of the cell so that the front emitter is electrically isolated from the cell rear. Since ion implantation does not undertake the diffusion process, this step is skipped.
Hahn expands on this saying, “You have a wafer which is on a table or something similar, and it is bombarded only from the front side. So ideally there are no atoms entering the wafer from the rear side which is the case when you place the wafer in a gas phase and the gas is all around the wafer on both the front and rear sides.” Gettering again rears its head here. Gettering ability might then be reduced as there is lower concentration on the front side and there is no gettering ability at all on the rear side. With phosphorous on both sides, as in diffusion, gettering is more effective. But on the other hand, as Hahn counters, there is no need to then remove the doped rear side after diffusion, which is the case when using POCl3 diffusion, for example with novel, more complex cell processes.

Efficiency

Where does this take us in terms of efficiency? The cell efficiency distribution of Solion-implanted cells, as a result of uniformity, have a tighter efficiency distribution curve. The number of cells that deviate from the average are lower than that given for POCl3 cells. Firstly the cell efficiency on average is higher at 19.2% than POCl3 that peaks at 18.7%, as Applied Materials’ numbers show. This means that the Solion enables larger volumes of higher efficiencies. This in turn reduces the number of bins, improving yields.
Hahn says that is it important to look at the references. He says, “It could be true, as it gets more homogeneous so there is less spread. If you compare to a standard homogeneous emitter, the higher efficiency could be true. But if you compare to a state-of-the-art selective emitter process, then concerning efficiency, there might not really be a big advantage to use the ion implantation process. It really depends on the references.” With selective emitters, there is deeper emitter doping under metallization lines and shallower emitter doping elsewhere. This design in turn increases the cell’s ability to capture more light.

Cost

Hahn believes that the biggest problems are cost and throughput pertaining to ion implantation. “In PV there are huge cost pressures and for me it is hard to believe that this process on a cell level is more cost effective than standard diffusion processes at the moment. The tools seem to be expensive and the throughput is not that big at the moment. So I think there is potential but only if the machines get cheaper and the throughput can be increased.” Schmid, for example, manufactures the POCl3 diffusion furnace which also offers very high diffusion uniformity, as the company states. The company’s POCl3 process is used in the mass production of solar cells with efficiencies over 20%, as the company’s Development Process Manager, Carsten Demberger, says. He believes that the ion implanters tend to have larger footprints due to their higher energy consumption and this is a factor that potential investors need to take into account.
Applied Materials sees the cost benefits in the implementation of their technology though. Solion implanted cells offer a lower cost of ownership (CoO) per watt (W) compared to standard POCl3 technology. Cell manufacturing CoO is pegged at US$0.54/W for POCl3 and at $0.526/W for Solion. Module manufacturing CoO is $0.861/W for standard POCl3 and $0.84/W for Solion. Mullin elaborates, “In a market where every cent counts, it is critical that any new technology delivers an attractive return on investment. Solion delivers a payback of less than 2 years while enabling advanced cell architectures to be manufactured with lesser steps, higher yield, and lower $/W.” Summing it up, there are so many possibilities and processes to take into account. Single machines cannot be compared to one another as Schmid Group believes, according to Christoph Kübler from the company’s communication department. Investors need to ask themselves questions pertaining to overall costs per watt, resulting efficiency, throughput, uptime, yield, accuracy and repeatability of the overall process and flexibility of solutions, as Schmid Group highlights.

Challenge ahead

Mullin says that the technology is proven and the integration of ion implantation into solar cell processing has been proven. “For Applied/VSE the challenge is working closely with our customers to enable them to unlock the value of ion implantation in high volume manufacturing and gaining full adoption,” Mullin says.
Technologies like ion implantation as well as selective emitters and back passivation among others will continue to shape the upstream sector (see Graphic “Technologies entering c-Si production”, p. 54). However, whilst refining the design of the solar cell to achieve higher efficiencies, cost penalties should not occur and the footprint needs to be considered. Should companies successfully marry cost savings and better efficiency and a lower footprint with their new technology, then it will only be welcomed by module manufacturers.