In recent years, the need for evaluating trace metals has constantly increased as the purity of the components used for various products has improved. Although the silicon contained in steel and other materials above a certain level can be analyzed using the gravimetric silicon quantifying method [1], it requires considerable time and effort, involving repeated processes such as addition of acid, heating, filtration, separation, weighing, redissolution, and high-temperature heating. Although ICP-OES, ICP-MS, and molybdenum blue absorptiometry are some of the methods used for measuring silicon at low concentrations, measurements by ICP-OES and ICP-MS have the problems of silicon elution from the torch tube and molecular ion interference. While molecular ion interference can be prevented by using a reaction cell or a double focusing mass spectrometer, it is difficult to analyze trace levels of silicon in solutions containing hydrofluoric acid as silicon elutes from the torch tube [2]. It is also necessary to dilute the sample and conduct matrix matching or the standard addition calibration curve method when measuring a sample containing a large amount of the main component. Meanwhile, although molybdenum blue absorptiometry is a simple measurement method, various elements can cause interference [3].

We therefore quantified silicon by using molybdenum blue absorptiometry after vaporization separation of the silicon from other coexisting substances as silicon tetrafluoride gas and collection in boric acid solution [4-12]. However, the analysis procedure was complex and susceptible to contamination, and the separation method needed to be made more efficient in order to measure silicon at lower concentrations.

We therefore developed a silicon analysis method using continuous flow analysis (CFA), in which the entire process from reagent addition, stirring, gasification and separation, and absorption in the collecting solution, to quantification by molybdenum blue absorptiometry in a thin tube can be fully automated.

Principle of the CFA Method

The basic principle of the CFA method was invented in 1956 by L. Skeggs, a biochemist in the U.S. The method can almost fully automate the normal chemical analysis procedures done manually in equipment, including sample collection, pipette operation, dilution, mixing, filtration, heating, extraction, dialysis, distillation, reaction detection.

Since all of these reactions are completed within a thin tube, the method is hardly affected by contamination with external factors.

Developed Analysis Method

In the analysis method we developed, the equipment suctions the sample, to which hydrofluoric acid has been added, with an autosampler, mixes it with sulfuric acid, and agitates it to generate silicon tetrafluoride gas. The gas is then separated from solution and absorbed in boric acid solution to conduct quantification by molybdenum blue absorptiometry.

We used a combination of equipment manufactured by BL TEC K.K. and that by Seal Analytical Inc. as the CFA equipment. Figure 1 shows the equipment used.

Figure 1: Developed equipment.

Figure 2 shows the flow diagram for the developed equipment.

Figure 2: Flow diagram of a continuous-flow analyzer for the determination of silicon

R1: Sulfuric acid (5+1)

R2: Boric acid solution (4%)

R3: Ammonium molybdate tetrahydrate solution (10%)

R4: 3 mol L^–1 Hydrochloric acid

R5: L-Ascorbic acid solution (1%)

S: Sample or water

1: Pump

2: Air

3: Mixing coil

4: Vapor-liquid separator

5: Waste

6: Spectrophotometric detector (810 nm, 50 mm)

Sampler

We used an RAS8000 autosampler (BL TEC K.K.) as the sample suction part indicated by S in Figure 2. The sample to be measured and cleaning water are introduced into the thin tube alternately. The analysis cycle per sample is approximately 6 minutes. The standard solution and the sample to be measured that has been made into a solution are set in the sampler.

Pump

We used a PUMP IV peristaltic pump (Seal Analytical Inc.) as the pump indicated by 1 in Figure 2. It supplies reagents, sample, air, etc. to the equipment.

Reaction Cartridge

We used a reaction cartridge (Seal Analytical Inc.) as the reaction cartridge indicated by 3 in Figure 2, including the mixing coil. To prevent the elution of silicon from glass, all the mixing coil and liquid feeding tubes were made of resin.

The reaction formula for converting silicon into silicon tetrafluoride is as follows:

The sample to be measured is added to sulfuric acid and stirred so that the silicon is converted into silicon tetrafluoride gas, which is transferred to the gas phase side. Figure 3 shows the reaction inside the mixing coil.

Figure 3: Reactions inside the mixing coil.

The solution and silicon tetrafluoride gas are separated by the vapor-liquid separator indicated by 4 in Figure 2. The separated gas is collected in the boric acid solution. It is possible to conduct separation in 1.5 minutes per specimen by using this method.

Figure 4 shows a schematic diagram of the vapor-liquid separator.

Figure 4: Schematic diagram of vapor-liquid separator.

The silicon tetrafluoride that was separated in the vapor-liquid separator is absorbed in the boric acid solution, which is then fed to the color reaction part.

Molybdenum blue is formed by adding ammonium molybdate solution and L-ascorbic acid solution to the silicon that is collected in the boric acid solution.

Detector

We used an SCIC 3 (BL TEC K.K.) as the detector indicated by 6 in Figure 1. This is a double-beam absorption spectrophotometer that uses a halogen lamp as the light source. It measures the absorbance of molybdenum blue in a cell with an optical path length of 50 mm by using a wavelength of 810 nm.

Lower Limit of Quantification

We calculated the lower limit of quantification using the following method [13].

Ultrapure water to which HF (1+100) was added at the ratio of 0.5 mL in 100 mL was used as the operation blank. We calculated the average peak height (X_d) and standard deviation (S_d) by performing seven measurements on the operation blank. We then performed eight measurements on the 0.25 mg/L silicon standard solution added with hydrofluoric acid (1+100) at the ratio of 0.5 mL in 100 mL, and calculated the average peak height (X₂).

We calculated the lower limit of quantification as follows:

Lower limit of quantification for the method (MLOQ) = 10 × S_d / k₂

k₂: Slope of calibration curve [(X₂ − X_d) / C₂]

C₂: Concentration in standard solution (0.25 mg/L)

The lower limit of quantification was 0.0032 mg/L.

Comparison of Silicon Analysis in Ta₂O₅

Since Nagaoka, et al. reported a comparison of the analysis results between the manual analysis method and CFA in Ta₂O₅ [14], we introduce the case here.

Lower Limit of Quantification for Manual Silicon Analysis Method in Ta₂O₅

4 mL hydrofluoric acid (1+1) was added to 1 g of the sample to conduct heating decomposition for 5 hours at 150ºC in a heating and pressurizing sealed container (HUT-25, HUS-25, SAN-AI Kagaku Co., Ltd.). After transferring the decomposed sample into a vaporization bottle injected with 50 mL sulfuric acid (5+1) and aerating with nitrogen gas at the flow rate of 1 ml/min for 30 minutes, we let the silicon in the sample be absorbed in 50 mL of 4% boric acid solution as silicon tetrafluoride gas. Figure 5 shows a schematic diagram of the gas expulsion equipment.

Figure 5: Schematic diagram of gas expulsion equipment.

We poured the absorbed solution into a 100 mL flask, added 5 mL of 10% ammonium molybdate and 1 mL of 5% L-ascorbic acid solution, added water to make the total volume 100 mL, and measured the absorbance at the wavelength of 810 nm using a cell with the optical path length of 50 mm.

We repeated the blank operation to a total of eight times. Table 1 shows the results as silicon content rate conversion values.

Table 1: Blank test of silicon by the conventional manual method.

We conducted an analysis to separate trace levels of silicon in samples as silicon tetrafluoride gas, absorb it in boric acid solution, and perform color reaction with molybdenum blue with the CFA method. Since the reaction is conducted inside a sealed tube, contamination with external factors is suppressed, and both the lower limit of quantification and the time required for separation of silicon (1.5 minutes per specimen) were improved.

Since silicon is measured after separation, we prepared the standard solution without adding matrices other than the acid. Regarding samples, it is possible to take measurements even when they contain high levels of matrices, including 1 g of Ta₂O₅ dissolved in 4 mL, for example. It is therefore possible to take measurements by concentrating the sample to the limit.

Although components used in semiconductors, and gallium nitride and gallium compounds which are expected to be used in next-generation semiconductors, are affected by silicon and other metallic impurities, there had been no method of measuring silicon at low concentrations. The results of the present study show that our method may be used to take simple measurements on low-concentration silica in various industrial products such as semiconductors and semiconductor components.

1. JIS M 8214: Iron ores−Methods for determination of silicon content. 1995.
2. Takahashi J, Yamada N. Development of collision/reaction cell for reduction of spectral interference in ICP mass spectrometry. Bunseki Kagaku. 2004; 53: 1257.
3. Inorganic Applied Colorimetric Analysis. 1974; 5: 38.
4. Nakamura Y, Hasegawa S, Okochi H. Determination of microgram amounts of silicon in vanadium, zirconium, niobium and tantalum by molybdosilisic acid blue spectrophotometry following fluoride separation. Bunseki Kagaku. 1992; 41: 479-484.
5. Takeyama S, Hosoya M. Bunseki Kagaku. 1984; 33: 80.
6. Inamoto I, Turuhara K, Uesugi Y. Determination of microgram amounts of silicon in iron oxide for ferrite by fluoride separation-graphite furnace AAS and Molybdenum Blue spectrophotometry. Bunseki Kagaku. 1986; 35: T67.
7. Kiyokawa M, Yamaguchi H?Hasegawa S. Determination of trace silicon in high purity iron by molybdosilicic acid blue spectrophotometry following fluoride separation. Bunseki Kagaku. 1992; 42: 219-222.
8. Yamaguchi H, Kiyokawa M, Hasegawa S. Fluoride Separation / Blue Absorptiometry of Molybdosilicate Quantification of Trace Silicon in High Purity Titanium and High Purity Chromiu. Bunseki Kagaku. 1995; 44: 647-650.
9. Imakita T,Onawa K, Nakahara T. Fluoride Separation-Quantification of trace silicon in niobium and tantalum by molybdenum blue absorptiometry. Tetsu to Hagane. 1999; 85: 135.
10. Yamaguchi H, Ito S, HasegawaR, Kobayashi G. Fluoride Separation / Molybdosilicate Blue Absorption Quantification of trace silicon in high-purity aluminum by absorptiometry. Tetsu to Hagane. 2001; 87: 129-131.
11. Kikuchi, G. Kaiser, G. Tolg. Spectrophotometric determination of silicon in gallium arsenide at μg / g levels after separation of silicon tetrafluoride by distillation. Bunseki Kagaku. 1983; 32: E231-E238.
12. JIS H 1403: Methods for chemical analysis of tungsten materials. 2001.
13. JIS K 0126: General rules for flow analysis. 2019.
14. Nagaoka M, Ishihara Y, Yoshinaga F, Koyanagi A, Hatamoto S. 69th Conference of person in charge of the field in mine and smelting works. Analytical Proceedings. 2019; 45.