New Post-combustion Ultraviolet Fluorescence Spectroscopy for Precise Determination of Trace Sulfur in Steel

Abstract

We have developed an analytical method which enables highly precise and rapid quantitative analysis of ultra low sulfur content in steel samples. This development has been termed the “post-combustion ultraviolet fluorescence method”, which combines a high frequency induction furnace with a continuous UV fluorescence analyzer. Despite the same ease of operation as the conventional IR method, the newly developed method showed high sensitivity and good precision. The quantitation limit of sulfur in steel was 0.5 mass ppm. In addition, it was shown that this equipment has sufficient stability for use as a process control analysis apparatus in continuous operation.

1. Introduction

It is well known that sulfur in steel affects the mechanical properties of steel products. Sulfur, for example, causes brittle fractures by segregating grain boundaries or hydrogen-induced cracking (HIC) by forming manganese sulfide. Particularly in tough, high strength line pipe for transportation of petroleum and natural gas, it is essential to reduce sulfur in the steel in order to improve resistance to HIC and stress-corrosion cracking.

For this purpose, the sulfur concentration of the molten steel during actual production has recently been reduced to less than 5 mass ppm.¹⁾

The gravimetric method,^2,3) absorption spectroscopy,⁴⁾ the infrared absorption method,⁵⁾ emission spectroscopy,⁶⁾ and the X-ray fluorescence method⁷⁾ have all been defined in Japanese Industrial Standards (JIS) as sulfur determination methods for iron and steel. Among these methods, the combustion infrared absorption method (IR method) has been widely used in steel making because it gives precise, accurate, rapid, and easy analysis.⁸⁾ Currently, ultra-low sulfur steel, with sulfur content below 5 mass ppm, is regularly produced. The IR method lacks sensitivity because the method is designed to measure higher sulfur content. Hence, a combustion IR method with higher sensitivity is commercially available today. The determination limit below 5 mass ppm is achieved by using an adsorption-concentrator column of sulfur dioxide gas (SO₂) with this analyzer, but the long analytical time required prevents it from analyzing samples of the refining process. Although many methods for the determination of trace sulfur in steel have been developed,^{9,10,11,12,13)} these methods have never been implemented in steel making because of this long analytical time and their complicated procedures.

With the combustion IR analyzer that is commonly used in the steelworks, the steel sample is combusted rapidly using a high-frequency (HF) furnace with oxygen blowing. Sulfur in the steel sample is gasified as SO₂ and is analyzed by the infrared absorption method. We will report on a new, accurate and rapid analytical method for measuring sulfur content of steel. In this technique, we combined rapid HF combustion with the fluorescence method.

2. Experimental

2.1. Samples and Reagents

Certified reference materials (CRMs) shown in Tables 1 and 2 were used in this experiment. A steel sample of approximately 1.0 g, accelerators of 0.4 g Sn and 1 g W, and ceramic crucibles preheated to 1050 degrees, in air, were used for this experiment unless otherwise specified. A moisture trap was filled with magnesium perchlorate as a dehydrating reagent.

Table 1. Certified values of certified reference materials used (mass%).

CRMs	C	Si	Mn	P	S	Ni	Cr	Mo	Cu	V	Al	N
JSS 001-6	2.4*	1*	(0.03)*	(0.5)*	1.5*	(0.2)*	(<0.6)*	(<0.2)*	0.36*	(<0.3)*	(<1)*	2.1*
JSS 242-9					0.030
JSS 244-9					0.0020
JSS 652-14	0.0358	0.624	1.177	0.0315	13.5*	10.60	16.88	2.06	0.177		0.0030	0.0191
JSS 653-14	0.0564	0.621	1.575	0.0307	9.4*	13.74	22.36	0.146	0.201		0.005	0.039
JSS 654-14	0.0421	0.680	0.880	0.0193	4.2*	19.13	24.90	0.0361	0.0430	0.112	0.0099	0.0214
SRM 131g	35.3*				4.255*

*: mass ppm, ( ): reference value.

Table 2. Certified values of high carbon steel samples used (mass%).

CRMs	C	Si	Mn	P	S	Ni	Cr	Mo	V	Co	W	N
JSS 606-8	0.76	0.28	0.31	0.016	0.0008	0.065	4.00	0.58	0.83	0.12	17.16	0.0290
JSS 608-8	0.80	0.36	0.33	0.025	0.0028	0.044	3.99	0.41	0.99	9.09	17.03	0.0320
JSS 611-8	0.86	0.37	0.30	0.025	0.0013	0.13	3.97	4.88	1.88	0.40	6.27	0.0548

2.2. Fluorescence-spectrum Measurement of the Sulfur Dioxide in Steel Combustion Gas and the Influence of Co-existing Components

A schematic diagram of the experimental apparatus is shown in Fig. 1. A HF furnace (Horiba EMIA510, Japan) and a fluorescence chamber were connected. The steel sample and accelerators in a crucible were placed inside quartz tube surrounded by the induction coil, and were combusted in a 3.0 L/min oxygen flow. The combustion gas of the steel sample was continuously introduced into the fluorescence chamber and irradiated with the excitation light in the fluorescence chamber. The fluorescence was measured with an ICCD detector after dispersal by a Czerny-Turner spectrometer. A 10W xenon flash lamp Hamamatsu E6186 was used as the excitation light source. The peak wavelength and half value width of the excitation light dispersed by the monochromator (Edmund Optics MINI-CHOROME, USA) were approximately 200 nm and 20 nm, respectively. An ICCD detector (Andor Technology DH-520-18F-05, GB) mounted on an imaging spectrometer (Horiba iHR320, Japan) was used. The flashing frequency of the excitation light source was 10 Hz. The flash was synchronized with the measurement of the ICCD detector by a digital pulse/delay generator (Stanford Research System DG535, USA). A long-pass filter mounted in the entrance of the spectrometer prevented stray light from interfering with measurement.

Fig. 1.

Schematic diagram of experimental system with ICCD detector.

Instead of the ICCD detector mounted on the spectrometer, a photo multiplier tube (PMT) Hamamatsu R8548 was used to measure the fluorescence of SO₂ by using a UV selective transmitting filter Asahi Spectra RRX340 having 340 nm CWL and 80 nm FWHM, as shown in Fig. 2. A sensor was installed at the outlet of the fluorescence chamber for continuous detection of the chamber pressure (approximately 130 kPa). The fluorescence intensity was integrated for about 30 seconds. This intensity, after being normalized with respect to the pressure of the chamber, was correlated with the sulfur content in the steel. The normalization was carried out by dividing the fluorescence intensity by the pressure measured at the same instant. In order to evaluate the influence of moisture, a moisture trap filled with magnesium perchlorate was installed between the fluorescence chamber and the combustion furnace. The CRMs were measured under the condition of existence and nonexistence of the moisture trap, respectively. For comparison, the same experiments were also performed by the IR method. In both experiments, the absorbance or fluorescence intensity without a moisture trap was measured, and the sulfur content was determined by a calibration curve obtained by using the moisture trap. The analytical values were then compared with the certified values.

Fig. 2.

Schematic diagram of experimental system with UV filter and PMT.

2.3. Development of the Analyzer for Process Control Analysis

An analyzer for operation in the steelworks was manufactured by modifying a high-frequency carbon and sulfur analyzer (Horiba EMIA-920V, Japan) and an ambient sulfur dioxide monitor (Horiba APSA-370, Japan) as shown in Fig. 3. Measurements of the PMT were synchronized with the flash of the xenon lamp. The light of xenon lamp was separated by the reflection-type wavelength selection filters to become the excitation light. The fluorescence intensity obtained by the PMT was compensated by the intensity of the excitation light at each flash as shown in the Eq. (1). The compensated fluorescence intensity was integrated for 60 seconds or until 1% of the peak value to determine the sulfur content in the steel samples. This correction is essential to prevent the intensity reduction of the excitation light which is caused by the temporal deterioration of the optical systems and the light source from affecting the analytical values of sulfur. The fluorescence intensity is corrected as follows:   

$$F=\frac{I_0}{I^{\prime }}\cdot F^{\prime },$$(1)

where F is the corrected fluorescence intensity, F′ is the measured fluorescence intensity, I′ is the measured intensity of excitation, and I₀ is the initial intensity of excitation light.

Fig. 3.

Schematic diagram of analyzer for process control analysis.

The experimental conditions of combustion and gas flow rate were standard, except for removal of the moisture trap after the HF furnace. To compare this system with the conventional combustion IR method, a HF combustion carbon sulfur analyzer (Horiba EMIA510, Japan) was also used.

2.4. Long-term Testing

The analyzer shown in Fig. 3 was placed in the steelworks for a long-term test. As shown in Table 3, the analytical conditions, i.e. the interval of standardization and the number of analytical samples were almost the same as that of the conventional combustion IR method. Not only the CRMs but various routine samples were analyzed by the new analyzer so as to check reliability and stability. The intensity of the excitation light gradually weakens with time. Hence, its influence on the repeatability of results for CRMs was investigated under weakened intensity conditions or under intentionally reduced intensity by voltage adjustments of the light source. In addition, reproducibility was measured and compared with that of the conventional IR method.

Table 3. Analytical conditions for factory testing.

Light source (Xe flash lamp)	uninterrupted operation
Moisture trap	NA
Sample	JSS654-14
Sample weight	0.5 g
Calibration interval	One month

3. Results and Discussion

3.1. Fluorescence Spectrum of SO₂ in Combustion Gas Generated from Steel Sample

Figure 4 shows a temporal change of the spectrum obtained by the excitation light having a wavelength of 220 nm in the combustion gas. The steel samples were combusted in an oxygen atmosphere by using the apparatus from Fig. 1. A broad peak that is probably due to SO₂ was observed in the wavelength of 300 to 350 nm from 10 seconds after the start of combustion of JSS242-9. Conversely, no change was observed in the combustion gas of JSS001-6.

Fig. 4.

Fluorescence spectra of combustion gas flow of steel samples.

The excitation and fluorescence scheme is represented by the following processes. The processes (I) to (IV) denote those of excitation, fluorescence, quenching by co-existing gas M, and dissociation, respectively. According to Okabe, it is estimated by using thermochemical data that the threshold energy of the nonfluorescence process is 5.65±0.01 eV.¹⁴⁾   

$$S{O}_2+h{v}_1\to S{O}_2*$$(I)

  

$$S{O}_2*\to S{O}_2+h{v}_2$$(II)

  

$$S{O}_2*+M\to S{O}_2+M$$(III)

  

$$S{O}_2*\to SO+O$$(IV)

Characteristic fluorescence spectra were observed in the combustion gas generated from steel sample containing high sulfur content. We measured the fluorescence at the wavelength of 300 to 400 nm in the combustion gases from steel samples shown in Table 1, by means of a UV selective transmitting filter and the PMT, as shown in Fig. 2. Figure 5 shows the temporal changes of the spectrum obtained from steel CRMs. The relationship between each peak area of SO₂ fluorescence intensity in Fig. 5 and the amount of sulfur in the steel samples is shown in Fig. 6. Because the pressure in the fluorescence chamber saw a large change at the beginning of the combustion of the samples, the fluorescence intensity in Fig. 5 was normalized with the pressure of the chamber. Oxygen pressure in the chamber usually was approximately 130 kPa and sometimes decreased about several percent because of oxidation of steel sample and clogging of a dust filter. The sulfur amount in steel CRMs, calculated by multiplying sample weight (approximately 1.0 g) and certified value, showed a good correlation with the fluorescence intensity. Thus we can conclude that the continuous spectrum observed was originated from the fluorescence of SO₂. Consequently, the sulfur content in steel can be determined by measuring the intensity of this spectrum.

Fig. 5.

Temporal changes of fluorescence intensity of SO₂ generated from steel CRMs.

Fig. 6.

Relationship between fluorescence intensity and sulfur weight in steel CRMs.

The technique of measuring SO₂ in the atmosphere by using the ultraviolet fluorescence method was studied by Schwarz et al. and Okabe et al.^14,15,16,17) Their conclusions were that the fluorescence intensity of SO₂ and partial pressure [M] of coexisting gas M has the following Stern-Volmer relationship.   

$$\frac{{\mathrm{F}}^0}{\mathrm{F}(\mathrm{M})}=\text{1}+{\mathrm{a}}_{\mathrm{M}}[\mathrm{M}]\mathrm{\hspace{0.17em} },$$(2)

where F(M) and F⁰ are the SO₂ fluorescence intensity with and without coexisting gas M, respectively. Okabe et al. calculated Stern-Volmer constants a_M shown in Table 4 by means of experiment using a Zn light source (216 nm).¹⁷⁾ From Eq. (2), the ratio of SO₂ fluorescence intensity in oxygen and air is described by the following formula:   

$$\frac{\mathrm{F}({\mathrm{O}}_2)}{\mathrm{F}(\mathrm{A}\mathrm{i}\mathrm{r})}=\frac{1+{\mathrm{a}}_{\mathrm{A}\mathrm{i}\mathrm{r}}[760]}{1+{\mathrm{a}}_{{\mathrm{O}}_2}[760]}\mathrm{\hspace{0.17em} }.$$(3)

Table 4. Stern-Volmer constants of various gases for SO₂ fluorescence exited by a 216 nm continuum source.¹⁷⁾

Quenching gas	Quenching constant, aM (10−3 Torr−1)
Air	1.16±0.01
O2	2.68±0.09
CO2	0.74±0.07
H2O	25

Substituting the constants shown in Table 4 into Eq. (3), the SO₂ fluorescence intensity in oxygen is calculated to be 0.62 times that in air. Though SO₂ fluorescence intensity should be significantly reduced by oxygen, we were able to detect the signal because the generation of SO₂ was sufficient for quantification.

3.2. The Influence of Co-existing Components

Besides oxygen and SO₂, other gases such as carbon monoxide and carbon dioxide from carbon, and moisture from hydrogen, in the steel exist in the combustion gas of the steel samples. The high carbon steel samples shown in Table 2 were measured for the purpose of assessing the influence of carbon monoxide and carbon dioxide on the SO₂ fluorescent measurement. The results are summarized in Table 5. Since the analytical results of the sulfur in high-carbon steel CRMs were in good agreement with the certified values as shown in Table 5, we can conclude that the influence of carbon monoxide and carbon dioxide is negligible.

Table 5. Analytical results of sulfur in high carbon steel.

Sample	C, mass%	S, mass ppm
		Certified	Analyzed
JSS606-8	0.76	8	6.7
JSS608-8	0.80	28	27.2
JSS611-8	0.86	13	12.5

Assuming that a steel sample of 1 g, having a carbon content of 1 mass%, is combusted in an oxygen stream of flow rate 3 L/min, it generates approximately 18.7 mL of CO₂ under standard temperature and pressure. If the generation of CO₂ from a steel sample is completed in about 30 seconds, as expected from the SO₂ extraction curves of Fig. 5 under Gaussian distribution, its maximum generation rate should be 18.7×60/ $5\sqrt{2\pi }$ =90 mL/min. Since the oxygen flow rate is 3 L/min, the maximum concentration of generated CO₂ is estimated to be 3.0 vol.%. Since 2.68×10⁻³ and 0.74×10⁻³ are the Stern-Volmer constants of oxygen and carbon dioxide, respectively, the fluorescence intensity of SO₂ in an oxygen atmosphere containing 3.0 vol.% CO₂ was calculated as approximately 1.01 times that in a pure oxygen atmosphere from Eq. (4).   

$$\frac{\mathrm{F}(\%\mathrm{C}{\mathrm{O}}_2)}{\mathrm{F}({\mathrm{O}}_2)}=\frac{1+{\mathrm{a}}_{{\mathrm{O}}_2}[760]}{1+{\mathrm{a}}_{\mathrm{C}{\mathrm{O}}_2}[\mathrm{C}{\mathrm{O}}_2]+{\mathrm{a}}_{{\mathrm{O}}_2}[{\mathrm{O}}_2]}$$(4)

The above result shows that the sulfur analytical value from this method should not be influenced by CO₂, even under the simulation where all the carbon in 1 mass% carbon steel becomes CO₂ when combusted.

In order to estimate the influence of moisture in the combustion gas, the analytical results for sulfur in steel CRMs measured by the combustion UV fluorescence method and the conventional combustion IR method without a moisture trap are shown in Table 6. In the conventional IR method, analysis without using a dehydrating agent causes an error of about 30 mass ppm in the sulfur analysis, owing to the overlap of the absorption band of SO₂ and water vapor. On the other hand, in the UV fluorescence method, the presence or absence of a dehydrating agent does not affect the sulfur analysis value. This is because neither the excitation light nor the fluorescence were absorbed or quenched by moisture in the UV case compared to infrared. It is difficult to decide the replacement timing of the dehydrating reagent during operation, because the appearance of magnesium perchlorate changes little even as its dehydration capability decreases. Our combustion UV fluorescence method does not require a dehydration reagent, unlike the conventional IR method. Thus the UV method has an advantage over the IR method. The UV method does not give abnormal analytical values due to the deterioration of the dehydrating reagent.

Table 6. Analytical results of sulfur in CRMs analyzed without moisture trap.

Sample	Certified, mass ppm	Analyzed, mass ppm
		UV fluorescence	IR absorption
JSS001-6	1.5	1.5	37.3
SRM131g	4.255	4.3	33.7
JSS653-14	9.4	9.5	38.2
JSS652-14	13.5	13.5	46.7

The influence of moisture is calculated similarly to the above CO₂. The Stern-Volmer constant of H₂O, α_H2O=25×10⁻³,¹⁷⁾ is larger than CO₂. However, the content of hydrogen that combusts to be moisture is several mass ppm in steel. Even if a steel sample having a hydrogen content of 10 mass ppm is combusted, it is calculated that the SO₂ fluorescence intensity is mostly not influenced by H₂O.

3.3. Development of An Analyzer for Process Control Analysis

From the results mentioned above, we designed an ultra-low sulfur analyzer for steelmaking process control. The target of the specification is that the new analyzer is as fast and easy to use as the conventional IR analyzer. The determination of five CRMs and blank testing was continuously performed 10 times respectively, and the repeatability errors were calculated. The analytical results of sulfur in CRMs from this analyzer and the conventional IR analyzer are shown in Table 7. The repeatability for each sample analyzed by this analyzer was 3 to 10 times better than that of the conventional IR analyzer. Furthermore, the limit of quantitation was estimated to be 0.5 mass ppm, because the standard deviation of the blank measurement was 0.05 mass ppm. Figure 7 shows the extraction curves of this analyzer and the conventional IR analyzer when JSS001-6 of 0.5 g was combusted in oxygen. It is evident from Fig. 7 that this analyzer shows a remarkably high S/N ratio as compared with the conventional IR analyzer.

Table 7. Repeatability error of analytical results of sulfur in CRMs by UV fluorescence method and conventional IR absorption method.

Sample	Steel type	Certified value mass ppm	Repeatability error (n=10), mass ppm
			UV fluorescence	IR absorption
Blank(*)	–	–	0.05	0.24
JSS001-6	Pure iron	1.5	0.04	0.26
SRM131g	Low alloy silicon steel	4.255	0.04	0.41
JSS653-14	SUS 309S	9.4	0.16	0.55
JSS652-14	SUS 316	13.5	0.07	0.52
JSS244-9	Carbon steel	20	0.07	0.52

(*):  accelerators only

Fig. 7.

Comparison of SO₂ extraction curves obtained by combustion of JSS001-6. (A) UV fluorescence, (B) IR absorption.

3.4. Long-term Testing

The relationship between temporal change of the excitation light intensity resulting from the degradation of the fluorescence lamp and analytical repeatability was evaluated to test the robustness of this system for daily use. Figure 8 shows the relationship between the intensity of the excitation light and repeatability. Repeatability for CRMs had an excellent value of 0.08 mass ppm or less, even when the intensity of the excitation light decreased to 45% of the initial value during the test without changing the optical system. Repeatability slightly decreased to 0.10 mass ppm when the intensity of the excitation light was set to 20% of the initial value by lowering the ramp voltage. From this result, the lower control limit of the excitation light was determined as 20% of the initial value. Figure 9 shows the temporal change of the excitation light intensity. The intensity of the excitation light fell rapidly to about 80% in the two months from the start of testing, and then decreased gradually to 45% after 12 months. The intensity of the excitation light recovered to about 75% of the initial value by replacing the lamp. The intensity was not returned to 100% because of the degradation of the lenses and filters. When the test was continued without a break, the intensity of the excitation light fell to about 20% of the initial value 15 months after replacing the lamp (27 months after the start of testing). A lamp and the lens were replaced at this time because the intensity of the excitation light reached the lower control limit. The intensity of the excitation light recovered from about 20% to about 60%. We saw a better recovery of the light intensity this time because we replaced the lamp and lens simultaneously. A slight devitrification was observed in the quartz lens removed. It is thought that some contamination on the lens has been exposed to strong UV light for a long time and caused the devitrification. Thus, it is expected that the intensity recovers completely by exchanging the optical filters as well. From the above, it is thought that this device can maintain stability and reliability as a process control analyzer in steelworks by constantly managing the intensity of the excitation light, and maintaining it as needed. Reproducibility for a 20 day period has been measured with almost the same conditions by using untreated crucibles for the comparison of this analyzer with the existing analyzer (conventional IR method). The reproducibility of this analyzer was 0.27 mass ppm, and that of the existing analyzer became 0.54 mass ppm. This shows the advantage of the newly designed analyzer.

Fig. 8.

Relationship between excitation light intensity and repeatability error of sulfur in JSS654-14 analyzed by UV fluorescence method. ●, performed during factory testing; ○, performed under low excitation intensity condition.

Fig. 9.

Decrease of incident light intensity with time.

4. Effect of Heat Treatment of Crucible

The heat treatment of crucibles has been a routine procedure to determine ultra-low sulfur in steel. Therefore, the sulfur amount contained in the crucible was determined by the new analyzer. Figure 10 shows a comparison of the extraction curves of SO₂ generated from high purity iron (JSS001-6) of 0.5 g combusted using treated and untreated crucibles. It is evident from Fig. 10 that using a treated crucible reduces the peak of the SO₂ extraction curve. The difference of the peak areas between treated and untreated crucibles was found to be equivalent to a sulfur amount of 0.4–0.5 μg. Reproducibility of sulfur in high purity iron (JSS001-6) analyzed with treated crucibles was 0.04 mass ppm and that analyzed with untreated crucibles was 0.13 mass ppm. The effect of treating crucibles was confirmed clearly and quantitatively by this method.

Fig. 10.

Extraction curves of SO₂ for crucibles with and without preheat.

5. Conclusions

We developed a new combustion-ultraviolet fluorescence spectroscopy, which can determine ultra-low sulfur content in steel. The lower limit of quantification was determined as 0.5 mass ppm, about 1/5 that of the conventional method. The measurement time and ease of use of the new analyzer are almost the same as those of the conventional IR method. This method has the advantage of not requiring a dehydrating reagent because measurement is not affected by moisture. Thus, our method is superior to the conventional IR method for daily use. In addition, from the results of extended testing, it was proved that this equipment has sufficient stability for use as a continuous process control analysis apparatus for the steelworks.

Acknowledgements

The authors would like to express their deepest gratitude to A. Hirano, and T. Inoue, and T. Iwasaki of Horiba Ltd. for their generous support and useful comments.

References

• 1)   M.  Sasabe: Bull. Iron Steel Inst. Jpn., 15 (2010), 562.
• 2)  JIS G 1215-1, Iron and Steel-Determination of Sulfur Content- Part 1: Gravimetric Method after Separation of Iron, JIS, Tokyo, (2010).
• 3)  JIS G 1215-2, Iron and Steel-Determination of Sulfur Content-Part 2: Gravimetric Method after Chromatographic Separation, JIS, Tokyo, (2010).
• 4)  JIS G 1215-3, Iron and Steel-Determination of Sulfur Content-Part 3: Methylene Blue Spectrophotometric Method after Separation of Hydrosulfide, JIS, Tokyo, (2010).
• 5)  JIS G 1215-4, Iron and Steel-Determination of Sulfur Content-Part 4: Infrared Absorption Method after Combustion in an Induction Furnace, JIS, Tokyo, (2010).
• 6)  JIS G 1253, Iron and Steel-Method for Speak Discharge Atomic Emission Spectrometric Analysis, JIS, Tokyo, (2013).
• 7)  JIS G 1253, Iron and Steel-Method for X-ray Fluorescence Spectrometric Analysis, JIS, Tokyo, (2013).
• 8)   I.  Inamoto and  T.  Otsuki: Tetsu-to-Hagané, 80 (1994), 685.
• 9)   H.  Yoshikawa,  H.  Iwata,  S.  Ariga and  T.  Takahashi: Bunseki Kagaku, 40 (1990), T1.
• 10)   K.  Chiba,  I.  Inamoto,  K.  Tsunoda and  H.  Aikawa: Analyst, 119 (1994), 709.
• 11)   H.  Naka: J. Anal. At. Spectrom., 11 (1996), 359.
• 12)   N.  Omi,  M.  Ishii,  K.  Yumi and  M.  Yamada: Tetsu-to-Hagané, 89 (2003), 979.
• 13)   H.  Koga: CAMP-ISIJ, 18 (2005), 663.
• 14)   H.  Okabe: J. Am. Chem. Soc., 93 (1971), 7095.
• 15)   F. P.  Schwarz,  H.  Okabe and  J. K.  Whittaker: Anal. Chem., 46 (1974), 1024.
• 16)   H.  Okabe,  P. L.  Splitstone and  J. J.  Ball: J. Air Pollut. Assoc., 23 (1973), 514.
• 17)   H.  Okabe: Anal. Chem., 48 (1976), 1487.