2017 Volume 57 Issue 8 Pages 1420-1424
In the production of steel sheet, the surface of the steel is inevitably oxidized. Removal of oxide layers with acid pickling has been necessary to maintain quality. Monitoring acid concentration in the pickling solution is of primary importance in controlling the pickling process. The acid concentration of the pickling solution fluctuates dramatically in short periods because of the rapid pickling process. This makes it difficult to follow changes in the concentrations of the acid and other constituents in the pickling process with conventional chemical analyses. Currently, near-infrared spectroscopy (NIR) has been attracting attention as a potential continuous analytical technique to determine the concentration of chemical species. Therefore, the authors examined the performance of NIR for monitoring the concentration species in acid pickling solutions of the pickling process. NIR spectra in the region of 9500 cm–1 to 5000 cm–1, which covers the first stretching over-tone of OH bonds (6800 cm–1), were investigated. The peak shape of the OH bond in aqueous solutions is altered by the interaction of water molecules with dissolved species. The concentrations of total acid and iron in the solutions were estimated by multivariate analysis using the NIR spectra of the solutions. The parameters of the multivariate analysis were optimized to improve the correlation between the analytical values obtained by the multivariate analysis of the NIR and those obtained by chemical analyses. The correlation coefficient reached 0.98, which is enough to apply the multivariate analysis to practical operations. The NIR analyzer enabled us to simultaneously measure more than two species in 1 minute. The applicability of the system to an actual pickling process was tested, and good results were obtained.
Oxide layers called “scale” form on the surface of steel sheet during some heat treatment processes. This scale is usually removed by pickling to obtain improved appearance and properties of steel surfaces. Over-pickling degrades the surface layer of the steel sheet and shortens the life of the pickling solution. Thus, control of the acid concentration in the pickling solution is extremely important to secure stable product quality. As processing technology has improved, fluctuations of the acid concentration in the pickling solution have tended to increase because of the higher line speeds required for rapid production. Therefore, precise control of the pickling solution is desired.1,2)
The main parameters to control pickling solutions are the concentrations of acids and metal ions. Neutralization titration methods have been widely used to measure the acid concentration. Atomic absorption spectrophotometry (AAS) and inductively coupled plasma atomic emission spectroscopy (ICP-AES) have been commonly adapted to determine the concentration of metal ions in pickling solutions.3) However, the titration method is rather laborious and requires long measurement times. This method also requires not only preparation of standardized solutions as titrants and other working solutions of reagents, but also regular treatment of the waste solutions. Although AAS and ICP-AES provide quick determination, installation of these methods at a practical pickling line has been restricted due to their size and delicacy for the required environment. In addition, we have to use these techniques independently in order to determine both concentrations of acid and metal ions.
In recent years, near-infrared spectroscopy (NIR) has been applied to rinsing solutions and etchants in semiconductor and electronic device production processes as an analytical technique to determine acid concentration.4,5,6) NIR has attracted attention as a potential technique for rapid on-line or on-site measurement, because it offers short measurement time and easy maintenance. Moreover, this technique requires neither sample preparation nor reagents. In spite of these advantages, there have been only a few studies on the application of NIR to the steel industry. In this paper, we investigated a continuous measurement technique for acid and iron ion concentrations in pickling solutions using NIR. Multivariate analysis conditions were also optimized by comparison with results obtained by conventional methods. We also demonstrated long term NIR measurement tests carried out on a practical pickling line. The robustness of the method was also discussed.
The schematic diagram of the apparatus used in this research is shown in Fig. 1. The measurement conditions are shown in Table 1. A near-infrared spectrometer, NR800, manufactured by Yokogawa Electric Corporation, was used for continuous monitoring of NIR spectra. Incident light is introduced into the measurement cell with an optical fiber, and the transmitted light is detected with a photodetector. The absorption spectra are obtained by Fourier Transform from measured interferograms.
Schematic diagram of apparatus.
| Measurement probe | Flow through cell |
|---|---|
| Optical path length | 1 mm |
| Optical path diameter | 3 mmφ |
| Measurement wavenumber | 12500–4000 cm−1 |
| Resolution | 16 cm−1 |
| Cumulative number | 256 times |
| Measurement time | 40 s |
| Number of measurements | 5 |
Table 2 summarizes the concentration ranges of acid and iron, and the temperature of the model solutions. In this research, 10 model solutions and actual pickling process solutions were used. The model solutions were prepared by dissolving iron in nitric or sulfuric acid solutions with different acid concentrations. The concentrations of the acid and iron were prepared so as to form random distributions within the ranges, as shown in Table 2. The acid concentration and iron concentration in the solutions were determined by neutralization titration and ICP-AES as conventional methods, respectively. In the neutralization titration, sodium oxalate was used as a masking reagent to suppress the unfavorable effects of coexistent iron ions.
| Type | Concentration Range, wt% | Temperature [°C] | |
|---|---|---|---|
| Acid | Fe | ||
| Sulfuric acid | 10–40 | 0–20 | 60–80 |
| Nitric acid | 5–20 | 0–8 | 40–60 |
The absorption spectra of the first stretching over-tone of the OH bond of water molecules (around 6800 cm−1) change according to the interaction of the water molecules with the acid and metals species in the aqueous solution. Therefore, the concentrations of iron and acid in practical samples were estimated by using the multivariate analysis of the NIR. Partial least squares regression (PLS) based on Eq. (1) was used to develop a calibration model. Namely, to measure the concentrations by NIR, the coefficient ai of the Eq. (1) for each wavenumber was determined by multivariate analysis. Multiplicative scatter correction (MSC) was applied to the spectra as a pre-treatment for the multivariate analysis. The range of the wavenumbers used in the multivariate analysis was optimized so as to obtain the best correlation between the analytical values from NIR and those from the conventional methods. In addition, optimization of the wavenumber range was performed by using spectra obtained at the temperature of the upper limit in each acid solution (nitric and sulfuric acid) and the actual pickling solutions (refer to Table 2).
The analytical precision was evaluated by the correlation coefficient and the analytical accuracy (σd; error of mean square, shown by Eq.(2)), based on the correlation diagram of the relationship between the analytical values from the conventional method and those from the NIR.
| (1) |
C: concentration (mass%)
ai: coefficient
xi: absorbance (a.u.)
| (2) |
Ni: concentration obtained by NIR (%)
Ci: concentration obtained by conventional method (%)
2.4. Influence of TemperatureBecause the temperature of the pickling solution is usually controlled within a range of 20°C in the actual pickling line, it is very important to realize the influence of variation of the solution temperature on the analytical values obtained from the NIR. The conditions of multivariate analysis in both acid solutions were optimized at the upper limit temperature, as already stated in section 2.3. The NIR spectra were acquired under three different temperatures: the upper limit; 10°C below the limit; 20°C below it.
2.5. Actual Pickling Line TestsThe NIR apparatus was installed in an actual pickling line. The tests were performed with a polypropylene filter having a pore size of 30 μm in front of the measurement cell because it was not possible to obtain an adequate amount of transmitted light due to the effect of suspended solids (sludge, etc.). The tests were carried out under the measurement conditions previously determined in the laboratory.
To evaluate short-term stability, we carried out continuous measurement tests for 5 hours at intervals of approximately 1 minute. To evaluate long-term stability, we carried out continuous measurement tests over a 3-day period at intervals of several hours. In both tests, analytical values obtained by the conventional method were obtained at intervals and compared with those obtained by NIR.
Figure 2 shows the NIR spectra obtained from simulated solutions of sulfuric acid with various concentrations. The NIR spectra change according to the change in the concentration of sulfuric acid. It is important to apply the appropriate range of wavenumber range so as to obtain accurate analytical values from the multivariate analysis. Therefore, the wavenumber ranges for the multivariate analysis were investigated.
NIR spectra of aqueous solutions with different sulfuric acid concentrations.
The ranges of wavenumber for estimating each analytical species were 7300–6600 cm−1 for sulfuric acid, 10761–5361 cm−1 for iron in the sulfuric acid solution, 7135–5361 cm−1 for nitric acid, and 9450–5600 cm−1 for iron in the nitric acid solution, respectively. Figure 3 shows the correlation between the analytical values from the conventional methods and those from the NIR obtained under optimum wavenumber ranges. Satisfactory results were obtained for all analytical species, as the correlation coefficients were 0.98 or higher.
Comparison of the analytical values of the NIR method and the conventional method.
The influence of the solution temperature on the correlation between the analytical values from NIR and those from the conventional methods was investigated. Figure 4 shows the analytical results of acid and iron concentrations in a sulfuric acid solution at various temperatures. Deviation of temperature from the upper limit caused a shift in the relationship between the analytical values from NIR and those from the conventional methods. A descent of 10°C of sample solutions decreased the NIR analytical values 2.0 wt% on average, compared with those from the conventional methods. A descent of 20°C caused a decrease of 3.3 wt% on average. On the other hand, in the case of the iron concentration in sulfuric acid, a descent of temperature of sample solutions by 10 or 20°C caused an increase 1.2 or 2.3 wt% on average respectively in the analytical values from NIR, compared with those from the conventional methods.
Comparison of the analytical values of the NIR method and the conventional method (Sulfuric acid bath).
A similar tendency was observed in nitric acid solutions. A descent of temperature by 10 or 20°C caused a decrease of 2.0 or 3.3 wt% on average respectively in the analytical values from the NIR. However, in the case of the iron concentration in sulfuric acid, the drop in temperature by 10 or 20°C caused an increase of 1.2 or 2.3 wt% on average respectively in the analytical values from the NIR. The reasons why the influence of temperature changes on acid differs from that on iron are not clear at the present time. As the absorption spectra of the OH bond of the first stretching over-tone (around 6800 cm−1) reflect the information of the hydrogen bonds in water,7) it is considered that the effect of temperature on the hydrogen bond is influenced by the ions because of hydration. As mentioned earlier, since actual pickling solutions are controlled within a range of 20°C, errors caused by fluctuation of the solution temperature may occur in analytical values from NIR. Therefore, NIR spectra obtained at the temperature within a range of 20°C shown in Table 2 were analyzed by using the multivariate analysis in order to reduce the influence of fluctuation of the solution temperature.
Figure 5(a) shows the correlation between the analytical values from NIR and those from the conventional method for sulfuric acid solutions. Considering the NIR spectra obtained under three temperatures to optimize the multivariate analysis, the difference between the analytical values from the NIR and those from the conventional method was reduced from 3.3 to 0.19 wt%. On the other hand, accuracy at those temperature ranges decreased approximately 1.4 fold.
Comparison of the analytical values of the NIR method and the conventional method (Restraining influence of temperature). a) Using spectra of various temperatures for multivariate analysis b) Using spectra of various temperatures and expanding wavenumber region for multivariate analysis.
Accordingly, in addition to utilization of the NIR spectra obtained under the above conditions, the optimization of range of the wavenumber was also studied for further improvement of accuracy. Figure 5(b) shows the correlation between the analytical values from the conventional method and those from the NIR under new optimized conditions. Expansion of the wavenumber range from 7300–6600 cm−1 to 7400–5400 cm−1 suppressed the influence of temperature on analytical values from NIR to obtain good correlation resulting in improvement of accuracy to 0.85 wt%.
Figure 6 shows the NIR spectra of the sulfuric acid solution with various concentrations and temperatures, as a typical example. The spectral bands around 7000 cm−1 were altered with changes in the solution temperature and iron concentration. On the other hand, the spectral bands around 5500 cm−1 were just influenced by iron concentration. Expansion of the range of wavenumber in NIR spectra up to around 5500 cm−1, to optimize the multivariate analysis, achieves more accurate analysis by NIR without the influence of the change in the temperature of the solutions.
NIR spectra of sulfuric acid solutions with different conditions.
Figure 7 shows the correlation between the analytical values of nitric acid and iron from NIR and those from the conventional methods. In every relationship, satisfactory correlation was obtained by the same conditions as the sulfuric acid without change in the range of the wavenumber. Thus, the multivariate analysis enables us to obtain accurate analytical results of acid and iron under actual temperature conditions.
Comparison of the analytical values of the NIR method and the conventional method (after restraining influence of temperature).
Figure 8 shows the results of continuous measurements for the concentration of acid and iron in a nitric acid solution by the NIR installed in an actual pickling line. This figure also shows the results by the conventional method acquired at intervals. The analytical values from the NIR were in good agreement with those from the conventional methods. In continuous measurements at intervals of approximately 1 min, the increase in the acid concentration by periodic addition of nitric acid was precisely followed, showing that rapid changes in the concentration of the pickling solution can be accurately monitored.
Results of continuous measurements for actual pickling solution.
Figure 9 shows the results of the test of long-term stability in terms of concentration of acid and iron in sulfuric acid by the NIR and the conventional methods. The analytical values from NIR and those from the conventional method showed extremely good agreement over the 3-day period for both acid and iron concentration in sulfuric acid solution. In all cases, accuracy was 0.5 mass% or less. During continuous measurement tests on the actual line, the measurement cell and other equipment were found to be free from contaminants that potentially cause unexpected faults, resulting in practical monitoring with high stability.
Long term stability of the NIR measurement for actual pickling solutions. a) Sulfuric acid b) Fe in sulfuric acid.
NIR was studied as a practical method to determine the concentration of acid and iron in pickling solutions of steel-making processes. The following results were obtained.
(1) Measurement of the acid and iron concentrations in pickling solutions is possible by applying multivariate analysis using the absorption spectra from NIR under various conditions obtained with both simulated and actual pickling solutions.
(2) Without influence of solution temperature, analytical values were obtained by optimizing the wavenumber range and temperature conditions of the absorption spectra to be applied to the multivariate analysis. The accuracies of analytical values from NIR for sulfuric acid, iron in sulfuric acid solution, nitric acid and iron in nitric acid solution were 0.85 wt%, 0.29 wt%, 0.54 wt% and 0.37 wt% respectively.
(3) An on-line measurement test performed at an actual pickling line, indicating that satisfactory agreement between the analytical values from NIR and the conventional methods was obtained to demonstrate practical applicability of the method to continuous long-term measurements. The possibility of measurement at intervals of approximately 1 min by this method was also demonstrated.
Based on the results summarized above, rapid and highly accurate measurement of the concentration of the species in pickling solutions in steel-making processes is possible from NIR.
In the future, it is thought that application of NIR as an online analysis method for pickling solutions in steel-making processes, and development of a chemical management system making it possible to keep the concentration of pickling solutions within proper ranges at all times, will make an important contribution to the stable supply of steel products with high surface quality.
