Neutron activation analysis supports polysilicon production
10/01/2008
Ultra-trace bulk analysis of polysilicon helps meet production demands for high-purity silicon in solar-cell market
By J.D. Robertson, M.D. Glascock, and H. Newcomb, University of Missouri Research Reactor Center
The rapid growth in solar cell production has fueled demand for high-purity silicon. About 90 percent of the current solar cell market is based on solar cells using silicon, and the majority of the raw material for the process is derived from polycrystalline silicon. The demand for polysilicon from the solar industry is growing at up to 40 percent annually and it is anticipated that the use of polysilicon for solar cells will be three to four times that of the semiconductor industry in about 10 years.1 The projected global demands for polysilicon are such that the two largest polysilicon producers have announced plans to increase their production capacities to more than 58,000 metric tons in the next three years, and several other companies, with no prior polysilicon experience, are constructing facilities to enter the market.1
Although the purity requirements of silicon for solar cells (five to seven 9s) are lower than those for semiconductors (nine 9s), the power conversion efficiency of solar cells is largely dependent on impurity levels in the silicon raw materials. Measurement of element concentrations in the polysilicon raw material and the process wafers is therefore essential as new polysilicon production technologies are developed to lower solar-cell production costs and for the maintenance of quality control during the manufacturing of solar cells. The quality control of the polysilicon is especially important as new producers of this high-purity material enter the market. Since it was first applied in 1960 for the analysis of tantalum,2 instrumental neutron activation analysis (INAA) continues to be one of the most sensitive and accurate techniques for meeting industries’ needs for the trace element analysis of high-purity silicon. In keeping with the industry expansion, the demand for INAA of high-purity silicon at the University of Missouri Research Reactor Center (MURR) has more than doubled over the last three years. This article presents a brief overview of INAA of high-purity silicon.
INAA technique
The idea of using neutrons as an analytical probe for elemental analysis was first proposed and demonstrated by Von Hevesy and Levi for the analysis of trace quantities of rare earths in geological materials in 1936. Since then, the sensitivity, selectivity, and precision of INAA have made it a versatile and widely employed elemental analysis techniques. Because most materials are “transparent” to both the probe (neutrons) and the signal (gamma rays), there are few matrix effects associated with the analysis, and standardization of the measurement is simple and straightforward. Moreover, because little, if any, sample manipulation is required, INAA is a highly sensitive technique that can be applied to bulk samples and is relatively free of reagent and laboratory contamination.
In INAA, stable nuclei in the sample undergo neutron-induced nuclear reactions when the sample is exposed to a flux of neutrons. The most common neutron reaction is neutron capture by a stable nucleus (AZ) that produces a radioactive nucleus (A+1Z). The “neutron-rich” radioactive nucleus then decays, with a unique half-life, by the emission of a beta particle. In the vast majority of cases, gamma rays are also emitted in the beta decay process and a high-resolution gamma-ray spectrometer is used to detect these “delayed” gamma rays from the artificially induced radioactivity in the sample for both qualitative and quantitative analysis. A schematic illustration of the neutron capture INAA process is given in Fig. 1.
 |
The energies of the delayed gamma rays are used to determine which elements are present in the sample, and the number of gamma rays of a specific energy is used to determine the amount of an element in the sample. For example, when a sample that contains iron is irradiated, a fraction of the 58Fe atoms in the sample will capture a thermal (or low energy) neutron and become 59Fe. The 59Fe atoms are radioactive and have a half-life of 44.5 days. When the 59Fe atoms beta decay to 59Co, a 1,099-keV gamma ray is emitted 56 percent of the time. The amount of iron in the original sample can be determined by measuring the number of 1,099-keV gamma rays emitted from the sample in a given time interval after the sample has been exposed to a flux of neutrons. A description of the procedures used to quantify an analyte in INAA is beyond the scope of this article. The physical principles of the analysis are so well understood that neutron activation analysis is one of the primary techniques used by the National Institute of Standards and Technology (NIST) to certify the concentration of elements in standard reference materials.
Although there are few matrix effects in INAA, direct and indirect interferences are possible. A direct interference occurs when the radioactive species or gamma ray of interest is produced by multiple nuclear reactions. For example, measurement of 28Al that is produced by thermal neutron capture on 27Al is frequently used to quantify trace amounts of aluminum in a sample. However, in polysilicon, 28Al is also produced in significant quantities through a high-energy neutron absorption reaction followed by proton emission on 28Si. To quantify aluminum in a high silicon matrix, one must account for the alternate production of 28Al by this reaction. A direct interference can also occur when the same energy gamma ray is emitted by two different isotopes. This spectral interference can be easily accounted for by the difference in half lives between the two isotopes and/or by monitoring multiple gamma rays from each isotope. An indirect interference occurs when the activity generated by a dominant species in the sample impacts the signal-to-noise ratio of the analyte of interest by changing the background in the gamma ray spectrum. A detailed description of how direct and indirect interferences are resolved in the application of INAA to solar-grade silicon can be found in an article by Revel et al.3
 Figure 2. Silicon samples for irradiation. Photo courtesy of University of Missouri Research Reactor Center. |
Scientists at MURR have been performing trace element INAA of silicon samples for more than 30 years. A complete description of our analytical protocol can be found in an article by Herrera et al.4 The most common samples are semiconductor silicon, polished wafers, ingot chunks, polysilicon blocks, and polysilicon beads. Samples with masses typically ranging from 10 to 80 g are loaded into individual graphite containers (Fig. 2). These containers are then bundled and placed in a reactor irradiation position where the samples are exposed to a neutron flux for 54 hours. After a decay period of 48 hours, the samples are either cleaned with deionized water in an ultrasonic bath or subjected to a mild or harsh etching procedure. The mild etch is used when a light surface cleaning of the sample is required by the client, and the harsh etch procedure is employed when the client requests that the entire sample surface be removed. After the cleaning/etching procedure, the samples are dried, weighed, placed in plastic containers, and counted on low-background, high-resolution gamma-ray spectrometers. Two counts are performed on each sample. The first 30-minute count is performed immediately after the cleaning/etching and is used to measure radionuclides having half-lives in the range of 12 to 48 hours. The second six-hour count is performed after a minimum decay of 14 days and is used to measure the longer-lived radionuclides. The sensitivities able to be obtained for 40 elements in high-purity silicon using INAA at MURR are given in the table.
 Table 1: INAA limits of detection for high-purity silicon |
null
Advantages and disadvantages of INAA
The major advantage of INAA is that it provides accurate results for large, bulk samples (tens of grams) without having to dissolve or digest the sample. Moreover, by employing an appropriate surface etch procedure, it is possible to ensure that the trace elements observed in the INAA measurement are coming from the bulk material and are not a result of surface contamination at the production facility or in the analytical lab???critical information when evaluating bulk material from a new production technology or new production facility. Total reflection x-ray fluorescence analysis, secondary-ion mass spectrometry, and vapor-phase decomposition inductively coupled plasma mass spectrometry are the techniques routinely used in-house by the semiconductor industry to ensure the purity of the semiconductor and solar-grade silicon. INAA is complementary to these surface analysis techniques in that it can provide similar sensitivities on large, bulk silicon samples. An example of the use of INAA to examine impurities in silicon produced by the silane and metallurgical routes can be found in a recent paper by Holt et al.5
As with all analytical techniques, there are drawbacks to using INAA. One major disadvantage is that the technique requires access to a high-flux neutron source to obtain the sensitivities listed in the table. As a result, the technique cannot be performed “in-house” by industry. A second disadvantage of INAA is the time required for the analysis. Given the continuous production schedules on which the semiconductor industry operates, an analytical protocol that takes four to five weeks can be inconvenient. (If information is needed on only a few elements with shorter half-lives, it is possible to reduce the analysis time.) The third major disadvantage of INAA is that it cannot provide information on some of the light elements???particularly B, C, and O???that are monitored to ensure optimum performance of semiconductor devices.
INAA has had a long and successful history of application in the semiconductor silicon industry for analysis of bulk samples, and it is proving an invaluable tool to assist industry with quality control as demand for polysilicon continues to grow.
J.D. Robertson, Ph.D., is scientific director of the Analytical Chemistry Group and associate director for research and education at the University of Missouri Research Reactor Center (MURR), and a professor in the Chemistry Department at the University of Missouri. M.D. Glascock, Ph.D, is a senior research scientist at MURR and director of the MURR Archaeometry laboratory. H. Newcomb, research specialist, is the project lead for high-purity materials analysis at MURR. MURR is associated with Elemental Analysis Inc. (Lexington, KY; www.elementalanalysis.com) by contract to provide marketing and technical support for MURR’s commercial analytical services.
References
- 1. “The Sun Shines On Electronic Materials,” Chemical & Eng. News, Vol. 86, No. 28, pp. 11???16, 2008.
- 2. M. Ordogh, V. Upor, J. Veronika, Magyar Tudomanyos Akad. Kozponti Fizikai Intezetenek Kozlemenyei, Vol. 8, pp. 307???311, 1960.
- 3. G. Revel et al., J. Radioanalytical and Nuc. Chem., Vol. 85, No. 3, pp. 137???150, 1984.
- 4. R.S. Herrera, J. Radioanalytical and Nuc. Chem., Vol. 248, No. 1, pp. 39???44, 2001.
- 5. A. Holt, E. Enebakk, A.K. Soiland, Proc. the 22nd European Photovoltaic Solar Energy Conf., pp. 1155???1159, 2007.