Steve Lowry is a senior scientist with Thermo Electron Corporation, Madison, Wisconsin.

Steve Lowry

Recent evidence that the phthalate plasticizers may cause health problems, particularly in children, has resulted in many countries prohibiting the use of phthalates in toys and clothing. The United States Child Safety Law actually prohibits the presence of several phthalate compounds at levels exceeding 0.1 wt % even though the commercial levels are often greater than 10%. While ATR spectroscopy works well for identifying phthalates at commercial levels, a more sensitive technique is required to detect trace levels.

We have created an approach where we produce a 500-µm thick film by hot pressing a sample and then acquire a transmission spectrum by measuring the infrared energy getting through the sample. This technique increases the sensitivity by almost a factor of 50, enabling the detection of phthalates below the 0.1 wt % level, if other additives in the plastic formulation do not cause spectral interference. However, we have observed that the success of the analysis at low levels is very dependent on the specific components present in the plastic.  

In this application, a spectrum is acquired from the printed area on a T-shirt in less than a 10 s. We verify that the sample is a PVC plastisol by comparing the peaks in a specific region of the spectrum to those in a reference spectrum of PVC. A sophisticated match algorithm is applied to the peaks in a region specific to phthalates. When the presence of phthalates is confirmed, we produce the report shown in Figure 1, which clearly states that this area of the shirt contains phthalates 

Figure 1: A warning report is created (red X) if the match value for phthalate is high.			

When the match value for phthalate is low, the report states that the result is "inconclusive." The ATR measurement does not have the sensitivity to confirm that phthalates are not present above the 0.1 wt % limit and recommends that the operator try the hot pressed film technique or the GC–MS method suggested by U.S. Consumer Product Safety Commission.  

Figure 2: Thermo Scientific Universal Film Maker and the spectrum from a sample press into a 500-Âµm thick film accessory.			

For thermoplastics such as polyethylene or PVC, heating a sample to the melting point and pressing a film is quite easy. Figure 2 shows the Thermo Scientific Universal Film Maker kit overlaid on a spectrum acquired from a 500-µm thick film. The scale expanded region is used to check for the presence of phthalates and reveals that the instrument sensitivity of the transmission spectrum is almost 50 times better than that obtained with the ATR technique. This provides the sensitivity to detect phthalates at the 0.1 wt % level. However, the technique also increases the intensity of any peaks resulting from the presence of other additives in the plastic and in some cases these will interfere with the peaks used to confirm the presence of phthalates which increases the standard error and the uncertainty of the results. 

By producing samples with a known thickness (500 µm), creating a classical least squares (CLS) quantitative method to determine phthalate concentration is straightforward. One advantage of using a multivariate statistical analysis technique such as CLS is the availability of a standard error term. This term is based on the goodness of the curve fit between the sample spectrum and the spectra included in the CLS model. We recommend that the reported concentration be at least three times the standard error for a reliable result with high confidence. One of the major concerns related to the presence of phthalates is when small amounts of contamination are present in a plastic where the main plasticizer is an allowed substance such as DINCH (diisononyl cyclohexane-1,2 dicarboxylate) which is an approved plasticizer for use with polyvinyl chloride polymers produced by the company BASF. Figure 3 shows the results of applying our Thermo Scientific TQ Analyst software method to a spectrum that is the result of subtracting the spectrum from a hot pressed film of PVC containing DINCH as the plasticizer from a spectrum of the same material spiked with 0.24 wt % total phthalates. 

Figure 3: The results of applying a classical least squares quantitative analysis method to the spectrum of a sample containing 0.24 wt % total phthalates after subtracting a reference spectrum.			

The SNR is excellent in the difference spectrum and the CLS method calculates a concentration of 0.17 wt % with a standard error of 0.011. The sample would fail the screen because the phthalate concentration is above the wt % allowed by law. 

Figure 4: The yellow question mark indicates that the measure concentration is leass than three times the reported standard error.			

If we do not subtract out the spectral features from the plastic reference, we still get a concentration of 0.18 wt % but the standard error term has increased to 0.16 wt %. Based on the higher standard error we would report that this result to be "inconclusive" even though the calculated concentration is the same. This report is shown in Figure 4. The only case where a sample will pass the screen is when the calculated concentration is below 0.1 wt % and the standard error is below 0.033 wt %. When this occurs, the report shown in Figure 5 is produced. 

Figure 5: The green check mark indicates that the measured concentration is less than 0.1 wt % as had a standard error below 0.033 wt%.			

Based on our present research we have chosen to report that we are confident no phthalate is present above 0.1 wt % and to not report either the calculated concentration or the standard error. 

Articles

Enhanced Sensitivity to Detect Phthalates by FT-IR Analysis

Infrared spectroscopy is a powerful analysis technique used in the semiconductor industry to ensure the quality of silicon and silicon wafers. An extensive set of ASTM test methods has been developed defining methods for silicon used in the electronics industry (now administered by SEMI, a global industry association for silicon device producers). The recent emphasis on non-hydrocarbon-based energy production has increased the interest in solar silicon applications, which in turn brought about a need for a rapid, sensitive, and well-characterized analysis tool. Although the requirements for silicon used in photovoltaic solar cells are slightly different from those used in electronics, many of the material specifications are similar.

One of the most important questions for silicon producers involves the determination of the amount of oxygen and carbon in the silicon. During the crystal growth process, oxygen and carbon are introduced in varying degrees into the molten silicon from the quartz crucible and the graphite heaters. As the ingot is pulled from the melt and the silicon solidifies, these impurities become incorporated into the crystal lattice. Because the carbon atoms occupy sites normally taken by silicon atoms in the lattice structure, this type of impurity is classified as substitutional. In addition, the oxygen atoms can find positions within the lattice structure between the silicon atoms, forming interstitial impurities. Different levels of interstitial oxygen (Oi) in the silicon have been correlated with varying effects, both physical and electrical. The requirement to control these effects leads to the need for monitoring the concentration of Oi in the wafers used to fabricate devices. Although no direct effects of the substitutional carbon (Cs) have been proven, many experiments suggest that Cs affects the way the Oi behaves when silicon is heated, such as in typical silicon-processing temperature cycles. Carbon can lead to defect formation and the formation of swirls as well as assist the precipitation of oxygen. 

The increased production of solar silicon has led to a requirement for a technique to quickly analyze the material from the ingot before final processing. To improve productivity, the analysis must analyze the C and O concentrations from multiple samples with minimal user interaction. The desired instrument should allow an operator to determine easily if a specific sample is "out-of-spec." These characteristics, coupled with the existing knowledge of silicon analysis from the electronics industry, make infrared a leading candidate. Here, it is shown how an accessory initially designed for analyzing biotech samples using the standard 96-well-plate format can be used to measure carbon and oxygen concentrations in silicon to the level of precision required by the solar silicon industry. This accessory permits users to analyze samples arranged in a precise 

 array and can be customized for any array of samples that fit within the constraints of the stage movement and beam size of the spectrometer. 

Production tools used in fabrication centers have rigid requirements, which lead to high costs. The needs of the burgeoning solar silicon industry are better met with an inexpensive, laboratory-based tool, used in a quality control laboratory rather than in a fabrication center. This is the focus of the system discussed here. 

A Fourier-transform infrared (FT-IR) system (Nicolet 6700 FT-IR spectrometer, Thermo Fisher Scientific) was fitted with a plate reader directly into the sample compartment (Microwell plate reader, Thermo Fisher Scientific). The plate reader was configured to analyze 40 samples of silicon that are 10 mm in diameter and 2 mm thick. 

The accessory consists of a high-precision 

 stage controlled via array automation software (OMNIC Array Automation, Thermo Fisher Scientific). The optical module contained an integrated camera for recording a visual record. A special cover with a small door allows the sample plate to be accessed without opening the cover (no need to break purge). The software also permits the integration of the metrology procedure. The results can be color coded using the metrology results. Thresholds can be assigned so that any samples that exceed the upper limit are colored red and any samples below the lower limit are colored purple, providing the instant feedback required for pass–fail determination.  

Spectra were acquired from various silicon samples at a resolution of 4 cm

. The analytical methods used to measure the carbon and oxygen concentrations are based upon the test methods originally developed by the ASTM and now supported by SEMI (MF1188-1107 and MF1391-1107). Figure 1 shows the spectrum of the 2-mm-thick float zone reference wafer and the spectra of the carbon and oxygen references after the silicon peaks have been removed. This spectral subtraction and intensity measurement forms the basis of the ASTM/SEMI methods. The intensity of the oxygen band at 1107 cm

  is scaled to determine the oxygen concentration, and the band at 607 cm

  is used to determine the substitutional carbon. The ASTM/SEMI methods require 2-mm-thick samples that have been double side polished. A further software package was used to perform the accurate analysis of samples of difference thicknesses (ECO/RS, Thermo Fisher Scientific). 

Figure 1: Carbon and oxygen float zone (ultrapure) standards and reference spectrum. 

To verify the performance of the accessory for silicon analysis, spectra were acquired from a set of NIST Standard Reference Material 2551 oxygen in silicon samples. The set consists of four 25-mm-square, 2-mm-thick silicon samples, mirror polished on both sides. A high resistivity float zone material with negligible oxygen content serves as the reference. The remaining three samples are cut from n-type crystals grown by a modified Czochralski process. The room temperature resistivity is >3 ohm-cm and the nominal oxygen concentrations are 17, 23, and 26 parts per million atomic (ppma). The samples were positioned in the Microwell plate reader in a 2 × 2 array and the locations were entered into the software. Figure 2 shows the array automation analysis of the four samples. The wells in the schematic in the upper left are colored based upon the concentration of oxygen in the sample calculated using one of the ECO/RS proven methods. The calculated values used to determine the color of each well are as follows: A1: 0.0 ppma, B1: 17.45 ppma, A2: 23.50 ppma, and B2: 26.90 ppma. The lower portion of the display contains the overlay of the spectra from the four samples showing the differences in the oxygen peak at 1107 cm

Figure 2: Reference samples run using the array automation software. The color-coded results permit instant visual confirmation of the passâfail status. 

The precision and stability of the system were evaluated using the standard with the highest oxygen concentration (26.303 ppma) as reported by NIST. A total of 50 repeat spectra were acquired using 4 cm

  resolution and 128 scans. The carbon and oxygen concentrations were determined and the results stored automatically. The average value for the 50 measurements was 26.629 ppma oxygen with a standard deviation of 0.083 ppma. 

Setting Up and Measuring 10 mm × 2 mm Disks

Seven samples were positioned onto a specially designed, 40-position sample holder with 10-mm slots. The array automation software contains a template tool permitting data collection from any custom plate design. FT-IR data were collected and an analysis based upon a predefined ASTM/SEMI method was performed. Figure 3 shows the results of the analysis. The images for each sample are captured during the data collection. This allows users to peruse all aspects of the collection — spectra, images, and results — in one interface. Note that one sample — shown in orange — had a much higher concentration of carbon (6.3 ppma) than the others. 

Figure 3: Analysis of silicon samples, showing visual image, spectra, and numerical results. 

Measuring Multiple Points on a Larger Sample

The plate reader was designed to measure discrete samples in an array. However, it can also be used to measure multiple points on a single sample or on multiple samples that fit within the 125 mm × 85 mm stage mount. For example, Figure 4 shows the measurement of an 8 × 12 array of points on a silicon test wafer. The results are overlaid on a display of the schematic showing the relative concentrations of oxygen across the sample. The standard deviations for both the carbon and the oxygen are less than 0.1 ppma for the 96 points. 

Figure 4: An array of measurements made on a single wafer using the plate reader. 

FT-IR has been used in silicon manufacturing for many years. The tools presented here are not fabrication plant tools, but are designed for the laboratory or near-line environment. Removing the fabrication facility requirements permits the sampling system to represent an inexpensive, rapid, and reliable system for users needing such a tool. Combining long-term experience in silicon analysis, ASTM/SEMI methodology, and the plate array accessory provides an excellent nondestructive, noncontact measurement for the solar silicon industry. With application-specific software tailored for customer operation, these determinations can be performed with little knowledge of the theory of the instrument involved. 

 is a Senior Applications Specialist, and 

 is FT-IR Product Manager with Thermo Fisher Scientific, Madison, Wisconsin.  

(2) R. Boyle, FT-IR Measurements of Interstitial Oxygen and Substitutional Carbon in Silicon Wafers, Thermo Scientific Application Note: 50640. 

(3) SEMI MF1188-1107: Test Method for Interstitial Oxygen Content By Infrared Absorption with Short Baseline. 

(4) SEMI MF1391-1107: Test Method for Substitutional Atomic Carbon Content of Silicon by Infrared Absorption. 

(5) B.G. Rennex, "Standard Reference Materials: Certification of a Standard Reference Material for the Determination of Interstitial Oxygen Concentration in Semiconductor Silicon by Infrared Spectrometry." NIST Spec. Publ. 260-121 (Aug. 1994). 

Infrared spectroscopy is a powerful analysis technique used in the semiconductor industry to ensure the quality of silicon and silicon wafers. The authors discuss the use of an inexpensive, lab-based system to measure carbon and oxygen concentrations in silicon to the level of precision required by the solar silicon industry.

Steve Lowry

Articles

Enhanced Sensitivity to Detect Phthalates by FT-IR Analysis

Analysis of Solar Silicon Using High-Throughput Spectroscopy