X-ray Fluorescence Analysis Advances the Development and Manufacture of Photovoltaic Materials

As interest grows in expanding the use of renewable energy, companies like DuPont (Wilmington, Delaware) are developing new photovoltaic materials for use in solar panels. Spectroscopy recently spoke to Dr. Wayne Brubaker, Senior Research Chemist at DuPont’s Corporate Center for Analytical Sciences, about the use of X-ray fluorescence (XRF) spectroscopy in the analysis of those materials, during both development and manufacturing.

Spectroscopy: What types of photovoltaic materials do you analyze with X-ray fluorescence? And do you use other X-ray techniques as well?

Brubaker: For years, my colleagues and I within DuPont’s Corporate Center for Analytical Sciences have provided analytical problem solving to support the development and manufacture of materials that are widely used in today’s photovoltaic modules. Examples of such products include fluoropolymer and polyvinyl fluoride materials for frontsheet and backsheet applications, respectively, as well as pastes used for metallization within modules.

The intensification of research and development (R&D) efforts across the photovoltaics industry has presented new opportunities to showcase the capabilities of X-ray analysis. For example, in the development of new active-layer materials, X-ray fluorescence (XRF) spectroscopy is used to determine elemental stoichiometry, while X-ray absorption near edge structure (XANES) fingerprinting provides for quantitative phase speciation in this system. X-ray reflectivity (XRR), like XRF, can measure the thickness of thin films but can also go one step further than XRF by providing direct measurements of a film’s density. The complementary information from these various X-ray techniques allows our researchers valuable insights into their new materials.

Spectroscopy: Are there challenges in using XRF for this work?

Brubaker: XRF is often selected for photovoltaic-related measurements because of its well-known attributes: It can be fast, highly quantitative, and often performed with little to no sample preparation. Also, compared to other elemental analysis techniques, such as inductively coupled plasma (ICP) spectroscopy or neutron activation analysis (NAA), XRF has a much greater potential to be used in at-line or online testing after materials advance beyond the R&D phase and into the realm of manufacturing and commercialization. It is much easier to transfer a method from the laboratory to the frontlines when the basic underlying technology remains the same.

I regularly tell colleagues that the most frequent challenge with developing quantitative XRF methods (perhaps 9 times out of 10) is acquiring acceptable standards for calibration, and this is particularly true with photovoltaic-related materials. The most accurate methods require well-characterized, matrix-matched standards because of XRF’s susceptibility to interelement and matrix effects. Samples related to our photovoltaic research tend to be unique and “first of a kind,” meaning that standards are seldom available. We often have to characterize some samples by other techniques, such as ICP and NAA, so that those samples may ultimately serve as our standards. Other times, we have to be a bit more creative in coming up with standards. In one case, for example, we developed a fusion preparation that enabled us to make our own standards.

Spectroscopy: Could you explain how you use XRF for the characterization and quality control of metallization pastes?

Brubaker: Our metallization pastes represent a wide range of compositions, each highly specialized for an intended application. Pastes for frontside metallizations, for example, consist primarily of silver, whereas aluminum tends to be the major component of backside pastes. Various glass frit materials are also introduced into compositions to provide adhesion properties when an end user fires a paste onto a substrate.

These pastes are largely inorganic in nature; therefore, XRF provides an excellent means for their elemental characterization. By XRF we can measure a paste nondestructively —“as received” on just a thin polypropylene film support — allowing the sample to be retained for additional testing and thus conserving as much of the high valued material as possible. (Remember that silver is not cheap these days.) High-concentration elements such as Ag and Al are best determined by XRF’s fundamental parameters methodology; at the same time, a minimal amount of sample also has to be analyzed by ICP to determine any trace elements and contaminants, given the lack of XRF standards for such a variety of paste compositions. We have found this complementary approach of using both XRF and ICP gives our teams the most complete and efficient “elemental picture” of the pastes.

Spectroscopy: I understand that your company has also been developing photovoltaic materials you call “CZTS” (Cu₂ZnSn[S/Se]₄) as a more sustainable alternative to materials made from Cu(InGa)Se₂, which include elements in low abundance on the earth. What aspects of the development or manufacture of these materials have you tested with XRF? Could you explain the steps you follow in that XRF analysis, including the fusion sample preparation process you developed?

Brubaker: Our teams were investigating various routes to synthesize CZTS during the early stages of the program, and obviously they needed to determine the stoichiometries of the materials they had made. The powders initially given to us for XRF analysis posed a couple of challenges. First, no prior materials were available to serve as calibration standards. Second, the amounts of material available — in the range of tens of milligrams —were less than ideal for our traditional instrumentation.

High-temperature lithium borate fusions that we normally use for the preparation of geological samples provided us solutions in this case. First, we found that a fusion with as little as 10 mg of CZTS powder would result in a homogenous glass bead that would generate plenty of signal intensity for each element of interest. The only trick was to figure out how to retain sulfur in the high-temperature melts (at 1100 °C). That was accomplished by adding excess BaO to capture the sulfur as nonvolatile BaSO₄. Second, one of the key advantages of fusion for general XRF preparation is that one can synthesize known calibration standards from laboratory reagents, which we did here using common oxides (such as CuO, ZnO, and SnO₂) ordered from a commercial catalog. This combination of fusion preparation and subsequent XRF analysis has proved to be an accurate and robust method for our CZTS powders.

More recently, our researchers started applying their CZTS formulations to substrates (such as glass) for evaluation as thin-film photovoltaic devices. Ideally, one would measure the stoichiometries of such films nondestructively so that the device could be passed along for other testing and then could be archived as needed. Neither a fusion-XRF nor a digestion-ICP approach would fit this criterion. Therefore, we are now using direct XRF measurements for CZTS thin films. The challenge again was the acquisition of appropriate standards for the desired level of quantitation. Commercially available films containing the individual elements of interest as well as comparative in-house analyses with ICP have been used for our current calibrations.

Spectroscopy: Are there other areas in your company’s development of photovoltaic materials for which you anticipate developing XRF methods?

Brubaker: As our photovoltaic programs continue to pick up speed and more photovoltaic-related products approach commercialization, I envision more and more of our XRF activities being focused on the development of at-line and online methods to monitor our processes and end products. The capabilities of less expensive benchtop and handheld units based on energy-dispersive XRF (EDXRF) have been gaining steadily on those of more traditional wavelength-dispersive XRF (WDXRF) spectrometers, and it has long been known that well-defined XRF applications can often function as “black boxes” in manufacturing environments where experienced analysts tend to be less abundant. Photovoltaic technology is also moving beyond crystalline silicon and developing thin-film modules and materials that involve continuous processes for fabrication. It is easy to imagine online XRF instrumentation being mounted above a moving CZTS-coated sheet to noninvasively monitor its film composition or thickness in real time.