by Helge Haverkamp, Christian Buchner, Schmid Group GmbH & Co.
A new type of crystalline solar cell with selective emitters can be easily integrated into the manufacturing process and increases efficiency by up to 0.7 percentage points.
Most crystalline silicon (c-Si) solar cells, like the human eye, are UV blind. The quantum efficiency drops at a wavelength of under 600nm, which corresponds to the color yellow-orange. With a violet blue wavelength of 400nm at the border, to the invisible UV range, the quantum efficiency lies at ~0.67. This means that the solar cell in this short wave spectral range does not convert almost a third of the photons into charge carriers. The photons still produce pairs of charge carriers and holes in the semiconducting material, but before these reach the electrodes where they can drive electric consumption, some of them recombine, producing only unwanted heat.
The cause of this UV-blindness has been known to the manufacturers of crystalline solar cells for decades. In the conventional manufacturing process, the ~200μm thick silicon wafer is first doped with boron atoms and is thus p-conductive. In the next process step, phosphorous atoms are diffused into the wafer at 850°C that convert a thin zone on the surface into an n-conductor, which acts as an emitter. The solar cell is thus nothing other than a large diode.
The n-doping of the emitter loses intensity the deeper it gets. At a depth of 300-500nm, the number of phosphorous atoms is the same as the number of boron atoms, ~1016 phosphorous or boron atoms to every 1022 of silicon atoms/cm3. However, on the surface, where the phosphorous atoms have penetrated, the n-doping is considerably higher. Here, every tenth atom is a phosphorous atom, ensuring that the transition resistance between the semiconductor and the metal contacts, which are subsequently applied by screen printing, is as low as possible. This high concentration of phosphorous atoms impairs the silicon crystal to such a degree that nearly all charge carriers on this layer recombine before they reach the contacts. The top most 50nm of a c-Si solar cell is therefore known as the "dead layer" -- i.e., it is of no use for producing electricity.
Solutions for the "dead layer" problem
Concepts for solving this problem have been available since the 1970s, i.e., so-called selective emitters, which combine a low resistance of the emitter directly under the contacts (high n-doping) with a somewhat higher resistance in the areas between the contacts (low n-doping). However, these concepts always required two separate processes for diffusing the phosphorous in which the other area always had to be covered by a mask. This double doping was not only expensive, it also proved to be susceptible to faults because heating the substrate twice considerably shortened the life span of the charge carriers in the thin crystalline wafers.
A newly developed technology called inSECT (inline selective emitter cell technology) uses the concept of selective emitters, but applies it in a simpler manner (see figure). The concept involves only one proven standard diffusion process instead of carrying it out in two stages. Only one subsequent process step is necessary to reduce the surface between the subsequent contacts down to a depth of 50nm. The high rate of doping with phosphorous atoms is purposefully reduced simply by thinning the material. However, the location on which the contacts are later printed remains intact. Viewed under the microscope, this thinning out is visible as a small lower-level step next to the contacts. For the selective etching process, which produces these steps, a mask is again required. However, for this purpose, wax needs to be applied to the surface with a special ink jet printer. These proprietary contact-free printers achieve a positioning precision of ±15μm at a printing resolution of 900dots/inch (dpi).
|Process sequence with etched-back emitter.|
For etching, the inSECT technology uses a thin solution of hydrofluoric acid, nitric acid, and water. In contrast to a two-stage diffusion, this process is considerably gentler on the wafers, and also more economical. The concept of selective emitters develops its effect only when the etching depth is precisely met ± a few nanometers over the entire area of the wafer. The trick: the acid produces only fine pores in the upper layer of the wafer. As this porous layer increases in depth, it acts as a membrane that becomes increasingly thicker and stops the inflow of fresh acid and the outflow of silicon. The result: the etching process slows as it gets deeper, i.e., it virtually controls itself.
A favorable aspect of the new process is that the porous silicon acts as a reflective layer for light and the wafer changes color as the etching process progresses. Once a depth of 50nm is achieved, the wafer shines in a gold color, which provides a simple method of visual control. Sophisticated measuring techniques indicating the correct time to end the etching process are not required. The correct dosing of acid and duration of exposure are sufficient to achieve the correct etching depth. The etching process is followed by immersion in a caustic potash solution. This removes the wax layer of the mask and also the porous silicon. The deeper space between the contacts of the selective emitter is then finished.
The inSECT process has already proven its advantages. Like many insects, solar cells with selective emitters "see" blue and UV light better and the degree of quantum efficiency lies at 0.86, which means the losses are more than halved. The electrical degree of efficiency is increased in this way from, as an example, 16.7% using homogenous emitters, to 17.4% using selective emitters, which corresponds to an increased output of 4%. This brings benefits to manufacturers, i.e., for a gain in efficiency of 0.7 percentage points, a surcharge of (currently) ~28 euro-cents can be achieved for a standard solar cell with an edge length of six inches. The additional processing stage, including investments, consumed materials and amortization costs only 8 euro-cents. It is easy to integrate this new process in an inline production line adding only 5m to its length -- it is not necessary to take out the wafers and process them separately. In our research center in Freudenstadt, final trials are currently being held with test wafers from end users; serial production is due to start later this year.
Helge Haverkamp received his PhD in physics at the U. of Konstanz and is a process engineer at Schmid Group GmbH & Co., Robert Bosch-Str., 32-34, 72250 Freudenstadt, Germany; ph.: +49 7441 538 368; e-mail [email protected].