# Facile and precise quantitative determination of silicon in naphtha by inductively coupled plasma-optical emission spectroscopy – Journal of Analytical Science and Technology

Aug 26, 2022

### Calibration

Since naphtha was diluted in PGMEA and all naphtha was volatilized on the hot plate, almost no naphtha was present in the sample for analysis. Therefore, there is no need to consider naphtha in ICP-OES optimization, and most of the final components are PGMEA as shown in Fig. 1. For the optimization of ICP-OES, after preparing PGMEA containing 100 ppb Si to maintain the same matrix as the sample for analysis, the RF power, coolant flow, nebulizer flow, auxiliary flow, and additional gas flow were tested to determine the condition with the largest signal size compared to the equipment background. In ICP-OES, silicon can be measured with two wavelengths of 288.158 nm and 251.611 nm. In this experiment, 251.611 nm with high relative sensitivity was selected (Amais et al. 2013; Gazulla et al. 2017), and the axial view mode was used for detection mode. The test was conducted by varying the RF power from 1150 W to 1500 W. At less than 1350 W, the background did not change, but the sensitivity was lowered, and the background was increased to 1350 W or higher. In the nebulizer, the plasma was unstable at 0.5 L/min or more, so the sensitivity of the equipment was lowered, and the RSD% (relative standard deviation) fluctuated. The ideal signal was obtained at an auxiliary gas flow of 0.5 L/min. If the additional O2 gas is insufficient, a large amount of carbon is formed from the unstable combustion, making the plasma unstable. If too much oxygen is supplied, the plasma is turned off. While checking the plasma light, the amount of O2 gas was determined so that the carbon emission in the plasma was invisible. Table 1 lists the instrumental parameters of the ICP-OES measurement for Si analysis.

The intensity trend of ICP-OES for various Si concentration solutions was identified by quantification using an external calibration method. The results showed that the Si intensity increased with increasing addition of silicon in the PGMEA, and all the samples fell onto a linear line regardless of their variations on the Si concentrations. These results suggest that small variations in the Si concentration during the preparation will not significantly affect the Si determination. Consequently, ICP-OES equipment reads the corresponding data. The linearity (R2), limit of detection (LOD), and limit of quantitation (LOQ) were 1.000, 1.0 ppb, and 3.3 ppb, respectively.

The methodological LOD and LOQ were calculated using the following equation presented by the International Union of Pure and Applied Chemistry (IUPAC) (May and Wiedmeyer 1998; Thomsen et al. 2003; For routine analysis requirements and mid-range sample throughput 2017). The LOD was calculated as an average by measuring the blank value of the blank solution used for calibration and a sample of 100 ppb concentration five times each. The LOD is defined as 10·s where s is value of the standard deviation of the measurements, and the following equation was derived by arranging in more detail by the calculation method suggested by IUPAC, as follows:

begin{aligned} & {text{LOD}} = 3{text{SD}}_{{{text{blk}}}} times frac{{{text{STD}}_{{{text{conc}}}} }}{{{text{STD}}_{X} – {text{BLK}}_{X} }} \ & {text{LOQ}} = 3.3 cdot {text{LOD}} \ end{aligned}

where SDblk is the standard deviation of the intensities of the multiple blank measurements. STDconc is the concentration of the standard. STDx is the mean signal for the standard, and BLKx is the mean signal for the blank.

Gazulla et al. diluted naphtha in isooctane to detect a LOD and LOQ of 8 ppb and 25 ppb, respectively. In this study, the Si single standard which were mineral oil based was necessary. The spray chamber temperature had been set to − 9°C because of the high volatility of naphtha and isooctane. But those are not easy to build up in normal labs. In addition, many researchers have proposed an analysis method by diluting in kerosene and xylene, but this has many problems due to the organic matrices. When xylene was introduced to ICP-OES, the plasma increased significantly due to the high volatility, resulting in unstable plasma and poor reproducibility. Toluene use was discontinued owing to its high toxicity (Kumar 1999; Gazulla et al. 2017).

In terms of the precision of the analysis result, the LOD was determined by the size of the standard deviation obtained by repeated measurements of the blank and standard solution. A low LOD means that the standard deviation of the sample is small. The LOD determined by Gazulla et al. was 8 ppb, which was eight times higher than that of 1.0 ppb in this study. In other words, this study made it possible to secure an analysis result that is approximately eight times more precise than the conventional analysis result in terms of sample precision analysis.

### Evaporation ratio

After 5 g of naphtha was placed into a washed PFA beaker, 10 g of PGMEA was added and heated to 80°C, 100°C, 110°C, 120°C, and 130°C on a hot plate. The weight change was measured at 10-minute intervals until 300 minutes. The weight change was measured up to 300 minutes at 10-minute intervals. Figure 2 shows that rapid volatilization occurred up to 100 minutes regardless of the temperature. This is because naphtha, which is more volatile than PGMEA, is first volatilized, and the residual naphtha and PGMEA are then volatilized together after 100 minutes. Evaporation at 130℃ can reduce the pretreatment time due to the high evaporation rate, but PGMEA and naphtha volatilize at the same time near the initial boiling temperature of PGMEA (145℃). Hence, the temperature to be applied to the experiment was set to 120℃.

### Naphtha analysis results

The silicon concentration of the naphtha sample was measured at intervals of 30 minutes from 90 minutes to 210 minutes on a 120°C hot plate. The data were tabulated. The measured concentration Cmeas is the ICP-OES raw data, and the calculated concentration Ccal is the value converted to the actual concentration using the following equation:

$$C_{{{text{cal}}}} = frac{{C_{{{text{meas}}}} times M_{{text{P}}}^{*} }}{{Delta M_{{text{n}}} }}$$

where (M_{{text{P}}}^{*}) is the PGMEA residual amount after volatilization and ΔMn is the amount of naphtha used for volatilization.

After measuring the weight by volatilizing naphtha at 120°C from 90 to 210 minutes at 30 minutes intervals, each concentration was converted using the following conversion equation and plotted in Figs. 3, 4, 5, 6. In ICP-OES analysis, the plasma turned off in the light naphtha where the boiling point was between 30 and 90℃. The boiling point of PGMEA was 146.64℃, which is higher than that of naphtha, and the plasma was maintained stably. When the temperature was kept constant in the naphtha and PGMEA mixture at a ratio of 1:1, naphtha was volatilized first, and PGMEA was volatilized more slowly due to the difference in boiling point. In Fig. 3 the amount of volatilization increased with time, and the measured concentration increased even though the variation pattern of concentration differs depending on the Si concentration because the type of naphtha is different, as explained in the sample preparation part of the Experimental section. The reason for the decrease in concentration sample measured at 150 min in Fig. 3a, c is the nonuniformity of hot plate depending on the sample location inside the fume hood. Because silicon is dissolved in PGMEA, it gradually concentrated, and when converted to the final concentration, all samples showed a constant concentration regardless of the volatilization time. In Fig. 3b, the Cmeas values increase continuously, whereas the Ccal decreases slightly because it is thought Si is also volatized minutely during PGMEA volatilization. From the results, all three samples with silicon concentrations of 50 ppb, 150 ppb, and 200 ppb showed the same trend as the result predicted before the experiment.

The calculated result would not show a constant concentration if a loss of silicon or secondary contamination by the environment occurred during the pretreatment process. The LOD and LOQ are the detection limits using the calibration curve and are arithmetic statistics representing the detection limits of the analysis equipment. In contrast, the method detection limit (MDL) refers to the detection limits in the sample analysis (May and Wiedmeyer 1998; Thomsen et al. 2003; For routine analysis requirements and mid-range sample throughput 2017), i.e. the lowest concentration that can be analysed with an analysis device after the pretreatment. Fig. 3 presents the average concentration range of 83% for the standard 50 ppb silicon and shows an insignificant value at concentrations below 50 ppb, so 50 ppb was selected as the MDL. ICP-OES analysis showed that the measured Si concentration increases with increasing evaporation time, regardless of the Si concentration. When converted to the actual concentration, the ICP-OES background Si intensity did not vary significantly in the evaporation time from 90 to 210 min. Hence, the evaporation temperature and evaporation time did not affect the analysis results, and there were no experimental problems.

### Accuracy

#### Reference material measurement

The reference material was prepared at a Si concentration of 200 ppb using OMCTS to verify the amount of Si in the mixture. A series of analytes with known amounts of reference material were measured and calculated under the same ICP-OES parameters. Table 2 lists the results for the five samples, where the listed Si concentration in each sample is the average of three measurements. The recovery is defined as the ratio of the amount of silicon present in the naphtha sample, which was recovered using the analytical method. It refers to an amount of silicon effectively quantified in relation to the real amount present in the naphtha sample. Table 2 reveals a good recovery rate of 86.07% with an average of 172.14 ppb compared with previous reports (Creed et al. 1994; Griffiths et al. 2000), which can be applied to the results to determine the total concentration of all analytes present in the naphtha samples.

Silicon in naphtha exists as OMCTS, one of the siloxane types, not pure SiO2. The problem is that the calibration curve for ICP-OES analysis was prepared by diluting a standard material made of water-based SiO2 in PGMEA, but the silicon dissolved in PGMEA after naphtha pretreatment is in the form of OMCTS. Therefore, the matrices are different. This can cause large errors in the analysis result. In order to confirm the possibility of this error, two types of samples were prepared. In the first sample, 200 ppb of OMCTS was spiked into naphtha, and in the second sample, water-based silicon was first diluted in PGMEA and then re-diluted in naphtha to prepare 100 ppb of each RM. Because of this test, the silicon content of OMCTS was 37.3%, which caused many errors when diluted to 200 ppb or less. Hence, the reliability of the data could not be secured, so an RM at 200 ppb and water-based samples with a median concentration of 100 ppb were prepared. The pretreatment was carried out in the same manner as the naphtha sample, and the recovery rate of the spiked sample was confirmed after analysis.

#### Analyses of standard solution spike and naphtha samples

To ensure the accuracy of the analysis results, the concentration analysis method was re-verified after spiking with a concentration of 100 ppb of Si in naphtha. As shown from the spike test results in Fig. 4, the concentration coefficient and Si concentrations increase with increasing evaporation time, similar to pure naphtha. The decrease in concentration of sample measured at 150 min is due to the nonuniformity of hot plate as mentioned earlier. In addition, good recovery rates of 90% to 110% were obtained regardless of the evaporation time.

Table 3 lists the results of the concentration change according to the volatilization time after separating naphtha samples into 10 ppb or less, 50 ppb, 150 ppb, and 200 ppb by Si concentration. In the sample of 10 ppb or less, the average concentration was 5.71 ppb. Even if the volatilization amount changed, it did not deviate significantly from the expected set concentration. The standard deviation range of 41.85 ppb, 133.19 ppb, and 202.66 ppb secured good results of 4-6%. Since the evaluation sample was not an RM sample, the determined value itself is considered the concentration present in the naphtha.

Figure 5 presents the ICP-OES spectrum according to the volatilization time. As the volatilization time evaporates, the peak size also increases in proportion, and both the left and right baselines coincide. This shows that naphtha with a low boiling point was mostly volatilized, and PGMEA solvent mostly existed. If naphtha existed, there would have been a difference in the baseline.

Figure 6 summarizes the change in recovery % according to the volatilization time of OMTCTS and water-soluble Si RM, which are organic base RM materials. As mentioned before, OMCTS was 200 ppb RM, and water-soluble Si RM was 100 ppb. Regardless of the volatilization time, the sample spiked with OMCTS showed a good recovery rate of 86% on average and 96% for the sample spiked with Si in the water base. Hence, the analysis result was not affected by the form of silicon present in the naphtha, organic silicon, or SiO2, even if the volatilization time was as short as 90 minutes. It is also shown that the recovery rate of Si RM is higher than OMCTS RM because the molecular formula of OMCTS is C8H24O4Si4, and it is possible that some OMCTS was lost due to volatilization with naphtha during the pretreatment process. On the other hand, since Si RM is dissolved in water, it is thought that volatilization did not occur.