Simultaneous Analysis of Soluble and Insoluble Oxygen Contents in Steel Specimens Using Inert Gas Fusion Infrared Absorptiometry

Abstract

In order to evaluate cleanliness of steel samples which contain oxygen in two different forms - chemically dissolved form and physically dispersed form in steel matrix, identification of each oxygen in the steel is important. A simple but promising method for simultaneous analysis of these two types oxygen in steel samples was developed in the present study using Inert Gas Fusion Infrared Absorption Method. By utilizing different carbothermic reaction temperature for each type of oxygen, the chemically dissolved oxygen (soluble oxygen) was first separated from the steel specimen at a low reaction temperature, while the physically dispersed oxygen (insoluble oxygen in the form of oxide inclusion) was separated at a higher reaction temperature. This idea was applied to a number of Al-killed ultra low carbon steel specimens, which contain alumina inclusions. It was shown that separation of the soluble oxygen and the insoluble oxygen was possible. The obtained oxygen content in this new method was independently validated by a conventional Inert Gas Fusion Infrared Absorption Method for the total oxygen content and by cross-sectional analysis of non-metallic inclusion for the insoluble oxygen. A possible supersaturation state of liquid steel after RH process was observed.

1. Introduction

Generation of non-metallic inclusion in liquid steel is inevitable in the secondary refining process after the basic oxygen furnace steelmaking. Upon tapping of the steel (or during RH (Ruhrstahl Heraeus) process for Ultra Low C (ULC) Steel), appropriate deoxidizer is charged in the ladle in order to kill O, which was dissolved form in the steel. Deoxidation reaction proceeds, and the O is deoxidized along with generating non-metallic inclusions. The most typical deoxidizer is Al (in ferro-Al alloy), and it proceeds:   

$$2\mathrm{A}\mathrm{l}(\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{r}\mathrm{o}\mathrm{\hspace{0.17em} }\mathrm{a}\mathrm{l}\mathrm{l}\mathrm{o}\mathrm{y})+3\underset{\_}{\mathrm{O}}(\mathrm{d}\mathrm{i}\mathrm{s}\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{v}\mathrm{e}\mathrm{d})=\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{i}\mathrm{n}\mathrm{c}\mathrm{l}\mathrm{u}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n})$$(1)

Other types of deoxidation by Si, Mn, Ti, Cr, etc. are also possible. In the present study, only the Al deoxidation and alumina inclusion are considered. Reaction (1) results in a transformation of the dissolved O into a non-dissolved form of O in the alumina inclusion. Hereafter, the dissolved O is called S. O (Soluble Oxygen) and the non-dissolved O in the alumina inclusion is called I. O (Insoluble Oxygen). These two types of O co-exist in the liquid steel. In addition to this, once the liquid steel is solidified, surface of the steel (cast slab, bloom, billet and small size specimen for other purpose) is oxidized. It generates an oxide film, which is thought to be iron oxide.^1,2,3,4) When a steel is subject to analyze for its O content, a small specimen is prepared from the steel: sampled from liquid state of the steel or cut from solidified state of the steel. Analysis of O content in metal may be made by wet-chemical analysis⁵⁾ or radio-chemical analysis.^6,7) For quick and easy analysis for the O content in steel specimen, combustion method either in vacuum or in inert gas atmosphere has been frequently used.^{2,3,4,5,8,9,10,11,12,13)} In these days, the combustion method under He gas as an inert gas has been extensively used. Its principle is discussed in Sec. 2.1. Since it is fast, easy, and accurate once calibrated with certified standard specimens, this method is a standard in analyzing O content in steel specimen.

However, the combustion method provides only total sum of all the oxygen contents present in the steel specimen ([T. O]):   

$$[\mathrm{T}.\mathrm{\hspace{0.17em} }\mathrm{O}]=[\mathrm{S}.\mathrm{\hspace{0.17em} }\mathrm{O}]+[\mathrm{I}.\mathrm{\hspace{0.17em} }\mathrm{O}]+\mathrm{O}\mathrm{\hspace{0.17em} }\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{\hspace{0.17em} }\mathrm{i}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{o}\mathrm{x}\mathrm{i}\mathrm{d}\mathrm{e}\mathrm{\hspace{0.17em} }\mathrm{f}\mathrm{i}\mathrm{l}\mathrm{m}$$(2)

where the oxygen content in the oxide film is known to be in the range of 1 to 2 ppm.¹⁾ In many steel plants, O content in steel specimens has been analyzed using the combustion method. For Al-killed steel, the analyzed [T. O] (total O content) is regarded as [I. O] (insoluble O content) in practice, assuming that nearly all the S.O has been killed. Therefore, the analyzed [T. O] is considered to represent content of alumina inclusion existing in the steel. Consequently, it has been regarded as an index of steel cleanliness. On the other hand, typical laboratory scale investigations on Al deoxidation have employed the combustion method to analyze O content in the steel specimen. The analyzed [T. O] is regarded as [S. O] (soluble O content), assuming that suspending alumina inclusions have been well floated up. These contradictory interpretations of analyzed [T. O] by the combustion method are due to analyzing O content using the identical method for the different types of O in steel specimen. In the present article, a new method is proposed which provides [S. O] and [I. O] simultaneously. The current combustion analysis (Inert Gas Fusion Infrared Absorptiometry) was employed, therefore, its advantage has been mostly retained. Separation of [S. O] and [I. O] during the combustion analysis was attempted using various Al-killed steel specimens. The proposed method was independently validated. Finally, a practical importance of separating the analysis for [S. O] and [I. O] is discussed.

2. Oxygen Analysis in Steel Sample

2.1. Principle of Infrared Absorptiometry

Figure 1 shows a flow diagram of Inert Gas Fusion Infrared Absorptiometry.¹⁴⁾ In the present study, LECO ON-836 (St. Joseph, MI, USA) was used. A surface-cleaned steel specimen is loaded in a graphite crucible under the inert gas (He). The crucible is placed in a vertical electric impulse furnace and heated in order to melt the specimen. C from the crucible diffuses into the molten steel specimen and reacts with O in the specimen (soluble and insoluble, both). C existing in the specimen also reacts. This reaction generates CO gas with a small portion of CO₂ gas. The generated CO and CO₂ gases are swept by the carrier He gas into the Non-Dispersive Infrared (NDIR) Cell, and pass through several units. The mixed gas passes through NDIR cells (two kinds in case of LECO ON-836). Absorption of infrared rays emitted in the NDIR cell by CO and CO₂ in the mixed gas is compared with that absorbed by a reference gas (He in case of the LECO ON-386). This is quantified to the amount of O (CO and CO₂), thereby providing O content in the loaded specimen. In general, O content in steel specimen is low (below a thousand ppm). Therefore, the portion of CO₂ is small. In this case, all the generated CO is oxidized to CO₂, and NDIR cell-2 is used to detect the extent of absorption of the infrared ray. For specimen of higher O content, CO and CO₂ are independently detected by NDIR cell-1. For the sake of simplicity, further discussions in this article mention only CO gas.

Fig. 1.

Schematic figure of the Inert Gas Fusion Infrared Absorptiometry flow diagram: (a) He gas cylinder, (b) heating chamber with steel specimen, (c) NDIR cell-1 of CO₂ (d) NDIR cell-1 of CO, (e) heated reagent, and (f) NDIR cell-2.¹⁴⁾ (Online version in color.)

Following three reactions can be considered:   

$$\underset{\_}{\mathrm{O}}+\underset{\_}{\mathrm{C}}=\mathrm{C}\mathrm{O}(\mathrm{g})$$(3)

  

$$\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})+3\underset{\_}{\mathrm{C}}=3\mathrm{C}\mathrm{O}(\mathrm{g})+2\underset{\_}{\mathrm{A}\mathrm{l}}$$(4)

  

$$\mathrm{F}\mathrm{e}{\mathrm{O}}_x(\mathrm{s})+x\underset{\_}{\mathrm{C}}=x\mathrm{C}\mathrm{O}(\mathrm{g})+\mathrm{F}\mathrm{e}$$(5)

where underscore, (g), (s) mean the dissolved form, gas, and solid, respectively. Each reaction generates CO gas from S. O, I. O, and oxide film.

The crucible and the specimen are heated by the electric impulse furnace. Heating power is controlled by current applied to the electrode (Fig. 1(b)). Temperature of the crucible was known to be proportional to the power. Temperature inside the crucible is not homogeneous, and this is intended in order to induce intense turbulence in the molten specimen by convection flow.¹⁴⁾ For typical O content analysis, temperature of the crucible is high enough to proceed the three reactions (Reaction (3) to Reaction (5)) together. Therefore, it provides [T. O].

2.2. Separation of Infrared Absorption Signal

The intensity of the CO gas evolved vs time obtained in the conventional combustion method is schematically shown in Fig. 2(a). The crucible is heated at a constant electric power. Therefore, it is nearly at a constant temperature in average (marked by a dashed line). This is referred to as “one-stage heating” in this article. The CO gas intensity increases, shows a peak, then diminishes. The area below the intensity-time curve corresponds to the O content in the specimen. This provides [T. O] including [S. O], [I. O], and O in the oxide film.

Fig. 2.

Signal of CO gas evolution from a specimen to analyze the O content (not to scale): (a) conventional heating pattern (one-stage) (b) two-stage heating pattern proposed in the present study. (Online version in color.)

Ise et al.¹⁾ considered that the oxide film is FeO. They reported that the Reaction (5) occurs relatively at lower temperature than two other reactions, because of lower melting temperature of the FeO. They controlled the heating pattern applied to the crucible in two-stage (first a slow but continuous heating to proceed the Reaction (5) up to 900°C, followed by a subsequent intense heating to keep the temperature at 2700°C for 60 to 70 seconds). Using this modification of the heating pattern, they were able to separate the O content in the oxide film from that of bulk oxygen content (sum of [S. O] and [I. O]). They applied this method to analyze O content in ultra clean steel (containing single digit [T. O]) and a certified standard specimen (JIS-GS6a). They found that the O content in the oxide film was 1 to 2 ppm, as mentioned before. In case of the ultra clean steel, such a low O content from the oxide film is considerable to the [T. O] of single ppm level. For other steel grade, this extremely low O content from the surface oxide may be neglected.

In typical Al-kill steel products, [T. O] would be higher than single ppm. Therefore, the O content arising from the oxide film is not a critical issue. More important point is how much O content comes from alumina inclusion or from dissolved O, out of the analyzed [T. O]. Since O in the alumina should be extracted by the carbothermic reaction (4), the temperature in the crucible should be high. The carbothermic reaction temperature for the alumina is high, therefore it is distinguished from a fusion temperature of C-saturated iron where the Reaction (3) occurs. The [S. O] can be separately analyzed apart from the [I. O] by controlling the heating power applied to the crucible containing Al-killed steel specimen. Figure 2(b) shows the idea of the present study that the intensity of CO evolution can be separated: one for [S. O] and the other for [I. O]. In the present study, this idea was tested for a number of Al-killed steel specimens which contains alumina inclusions. By optimizing the heating pattern, a “two-stage heating” method was developed which allows one to measure the [S. O] and the [I. O] in a single measurement.

The O content from the oxide film may be additionally separated by the method of Ise et al.,¹⁾ but it was not considered in the present study. Therefore, all of [S. O] reported in the present study contains the contribution from the oxide film, although its portion is negligible.

It has been reported that some steel specimen containing high Al content can cause the “gettering effect”.^8,12,15) The high content of Al reacts with O to form Al₂O gas compound, which leads to the reduction of the analyzed O content. Inoue and Suito reported that the gettering effect was not observed for a steel specimen containing Al as much as 0.079 pct.⁸⁾ Because the Al content in all steel specimens used in the present study was lower than 0.079 pct., this effect was not considered in the present study.

3. Experimental

In order to optimize the two-stage heating pattern, a series of experiments were carried out. Various steel specimens were used either with Al or without Al.

3.1. Sample Preparation

Electrolytic iron (EI) has no alumina inclusion. It is assumed that [T. O] of the EI is the same as the [S. O]. The EI specimens were prepared by cutting high purity electrolytic iron plate (99.9 pct. purity, < 150 ppm O, < 15 ppm N, < 30 ppm C, < 5 ppm Si). Surface of the electrolytic iron plate was ground by a diamond grinder. The plate was cut into several pieces of rectangular parallelepiped shapes (H~10 mm, a mass of 0.6 g to 1.0 g). Those were put in acetone, cleaned in an ultra sonic for 10 seconds, and dried completely.

A number of Al-killed steels (AK), which contain alumina inclusions, were used as specimens containing both S. O and I. O. The compositions of the AK specimens (AK1 to AK14) are shown in Table 1. AK1 to AK14 are the lollipop specimens provided by Steelmaking department, Gwangyang works, POSCO, during the production of ULC steel. AK1 and AK2 were sampled at the beginning of the RH process before Al deoxidation. Nevertheless, these include some alumina inclusions, which were formed due to a piece of Al loaded in the lollipop sampler. AK3 to AK14 were sampled after Al deoxidation, therefore these specimens have alumina inclusions due to the Al deoxidation (sampled by the lollipop sampler without Al). AK3 to AK5 were sampled after the RH process (RH), and AK6 to AK14 were sampled during the tundish process (TD). Size and mass of the lollipop samples (AK1 to AK14) before machining were approximately 31 mm × 39 mm ×15 mm and ~100 g, respectively. The AK specimens were further processed as was done for the EI (cut into several pieces of a mass of 0.6 g to 1.0 g, cleaned in acetone under the ultra sonic for 10 seconds, and dried completely).

Table 1. Composition of the specimens used in the present study. Those of AK specimens were provided by Steelmaking department, Gwangyang works, POSCO.

Sample No.	Mass pct.								Note
	C	O	Mn	S. Al	Ti	Cr	N	Si
EI	<0.003	<0.015					<0.0015	<0.0005
AK1	0.0157		0.081		0.0012	0.017			RH arrival
AK2	0.01		0.095			0.017		0.002	RH arrival
AK3	0.0256		0.075		0.0012	0.014			after RH
AK4	0.0157		0.081		0.0012	0.046			after RH
AK5	0.011		0.091	0.0285	0.0264				after RH
AK6	0.021		0.149	0.0344	0.0012	0.046			TD
AK7	0.022		0.549	0.0286	0.0270	0.067			TD
AK8	0.0015		0.158	0.0350	0.0010	0.020	0.0014	0.0560	TD
AK9	0.0017		0.142	0.0330	0.0010	0.020	0.0022	0.0570	TD
AK10	0.0012		0.147	0.0320	0.0010	0.020	0.0014	0.0540	TD
AK11	0.0014		0.154	0.0340	0.0010	0.020	0.0019	0.0600	TD
AK12	0.0013		0.154	0.0340	0.0010	0.020	0.0022	0.0550	TD
AK13	0.0016		0.142	0.0380	0.0010	0.020	0.0022	0.0580	TD
AK14	0.0011		0.164	0.0360	0.0010	0.020	0.0014	0.0530	TD

3.2. Calibration: Temperature and Oxygen Content

Temperature inside a graphite crucible could not be accurately measured due to geometry of the equipment. Moreover, as mentioned in Sec. 2.1, it is not homogeneous spatially. Only average temperature may be referred, which was controlled by the heating power applied to the electric impulse furnace. In the present study, the temperature controlled by the heating power was calibrated by inspecting melting point of pure Cu, pure ZnO, and pure Ni in graphite crucibles. In case of Cu and ZnO, those should melt at their melting point (1085°C and 1975°C, respectively). In case of Ni, it dissolves C. Therefore, a eutectic temperature of 1326°C was considered as the melting point.¹⁶⁾ Pellet of Cu, Ni, and pelletized ZnO were put in graphite crucibles, respectively, and melting behavior of these pellets was inspected after heating the graphite crucible at various heating power. The observed result is shown in Fig. 3. Almost linear relationship was found, which covers the melting temperature of Fe saturated by C and the carbothermic reduction temperature of alumina (approximately 1900°C).¹⁷⁾ Therefore, temperature referred in the present study for the two-stage heating mode is indeed a translated temperature of each specimen in a graphite crucible (in average) from the controlled heating power (in kW).

Fig. 3.

Relationship between the heating power of LECO ON-836 and melting point of Cu, Ni (saturated by C), and ZnO.

Oxygen content analyzed by the combustion method with LECO ON-836 was calibrated using certified standard samples, supplied by LECO. The conventional combustion method (one-stage) was used. Information of the standard samples (ID, mass, certified oxygen mass (g)) and heating pattern are listed in Table 2. “Calculated composition” was obtained as follows. When a standard sample was heated, CO gas was generated, and detected at the NDIR cell. The intensity of the detected signal was plotted over time (Fig. 2(a)). The area below the signal corresponds to the amount of O in the certified standard sample, after a necessary correction.¹⁴⁾ The same procedure was performed repeatedly for four kinds of standard samples. Finally, the calibration line was plotted by these obtained data, as shown in Fig. 4.

Table 2. Standard samples used in the present study. “Certified O mass” and “certified composition” were also provided by LECO. “Calculated composition” was measured in the present study using the conventional “one-stage” combustion analysis from the adjusted area (2050°C (4.5 kW) for 30 seconds).

No.	Name	Sample mass (g)	Certified O mass (×10−5)	Certified composition (ppm)	Calculated composition (ppm)
1	502-917 Lot No. L0375-9	1.0017	1.3022	13 ± 3	13.61
		1.0010	1.3013
2	502-643 Lot No. 0612	0.9987	3.1958	32 ± 5	31.46
		1.0013	3.2042
3	502-644 Lot No. 0483-40	1.0006	6.7040	67 ± 6	66.64
		1.0039	6.7261
4	502-199 Lot No. J0381-1	0.9904	20.5010	207 ± 10	207
		1.0008	20.7170

Fig. 4.

Calibration of the adjusted area of CO gas intensity by certified standard samples given in Table 2.

3.3. Heating Patterns

Various heating patterns were used in order to optimize the two-stage heating pattern which is suitable to separate the signals of S.O and I. O in AK specimens. It is a) one-stage heating (isothermal holding), b) multi-stage heating (step-wise isothermal holding), and c) two-stage heating. The last two-stage heating pattern was indeed the optimized two-stage heating pattern, which will be discussed in Sec. 4.2.

4. Results

All the experiments were conducted under an He atmosphere of 1.5 atm using the inert gas fusion infrared absorption method (LECO ON 836). Default procedure to measure [T. O] is as follows: heating a steel specimen in a graphite crucible loaded in the equipment for 30 seconds at 2050°C (4.5 kW), recommended by LECO. This procedure is considered to be enough to extract all S. O (chemically dissolved in steel matrix) and I. O (physically separated from the matrix in forms of oxide inclusion).

4.1. Experiment A: Isothermal Holding

The experiment was conducted using EI and AK1 specimens. O in each specimen was extracted by holding at a constant temperature (900, 1080, 1300, 1530, 1780, and 2050°C for 150 seconds, respectively). The results are shown in Fig. 5. Both specimens did not melt at 900 and 1080°C. Therefore, extracted O content was low. The AK1 specimen partially melted at 1300°C. All other cases, the specimens melted completely. In case of EI specimen, holding at 1300°C and higher resulted in noticeable O content (~50 ppm). The [T. O] was nearly constant. This means that O in the EI specimen was almost extracted when the temperature was higher than 1300°C. Since the EI specimen does not contain alumina inclusion, the analyzed [T. O] is thought to be [S. O] with negligible [I. O] ([T. O] = [S. O]). In case of the AK1 specimen, holding at 1300°C and higher also resulted in considerable O content, but holding at higher temperate resulted in higher [T. O]. Since the AK1 contains both S. O and I. O as alumina inclusion, and from the observation of the EI specimen, it is likely that the Reaction (3) occurred at lower temperature and the Reaction (4) occurred at higher temperature. Increasing the holding temperature resulted in more carbothermic reduction of alumina during 150 seconds employed in the present study. From these observations, it is concluded that S. O can be extracted at lower temperature, presumably below 1530°C. Isothermal holding at higher temperature results in the carbothermic reduction of alumina, which corresponds to I. O. According to a thermodynamic analysis by Krasovskii and Grigorovich,¹⁷⁾ start temperature of alumina reduction by C depends on Al content in steel. In our case, majority of the specimen contain Al in the range of 0.028 to 0.038 pct. For this range, the start temperature of alumina is approximately 1580°C to 1600°C.

Fig. 5.

Analyzed O content ([T. O]) in (a) EI and (b) AK1 specimens, isothermally reacted (Experiment A). Temperature and O content were calibrated as described in Sec. 3.2.

4.2. Experiment B: Stepwise Incremental Heating

The results shown in Sec. 4.1 were based on melting different specimens, even for the same type of specimen (EI, AK1). The analyzed O content at various temperature might be influenced by irregularity of each specimen. In order to eliminate this probable error, O was extracted in single specimen by taking step-wise isothermal holding. Temperature profile is shown in Fig. 6 by a dashed line. A (black) thin curve represents the intensity of O in an EI specimen, while a (blue) thick curve represents that in AK2 specimen (see the online version). Until 300 seconds (T ≤ 1080°C), no noticeable signal was observed because both specimens did not melt. Noticeable signals were detected after 300 seconds (T ≤ 1300°C) for both specimens. It is thought that the Reaction (3) resulted in the signal at 1300°C. No more signal was observed in case of the EI specimen. In case of the AK2 specimen, the signals were generated repeatedly at each step until the temperature reached 2050°C. These observations are summarized as follows. EI specimen (with no oxide inclusion) shows a reaction at a low temperature (1300°C in the present study), which reflects S. O in the specimen. AK2 specimen shows not only the reaction for the S. O (Reaction (3)), but also additional reaction at higher temperature, which reflects I. O in the alumina inclusion in the AK2 specimen (Reaction (4)).

Fig. 6.

Analyzed O content ([T. O]) in EI and AK2 specimens, step-wise isothermally reacted (Experiment B). Temperature and O content were calibrated as described in Sec. 3.2. (Online version in color.)

If the carbothermic reduction of alumina (Reaction (4)) had occurred at lower temperature, the peak of AK2 specimen at 1300°C or 1530°C would have reflected some part of I. O. However, according to a study on carbothermic reduction of alumina,¹⁹⁾ the alumina reacts with C at a slow rate at 1550°C: the reaction began after 30 minutes. Therefore, the signal generated at 1530°C (2.719 kW) within 150 seconds corresponds to the results of the Reaction (3). Contribution of I. O to the signal is likely to be low. The signals obtained at 1780°C and 2050°C are most likely generated by the carbothermic reduction of alumina, reflecting I. O in the AK2 specimen. This is because the S. O should have reacted at low temperature with C. Therefore, it is now concluded that S. O and I. O can be separately analyzed using the combustion method, if heating pattern is manipulated.

4.3. Experiment C: Separation of Soluble Oxygen and Insoluble Oxygen

From the results obtained in two previous experiments A and B (Secs. 4.1 and 4.2), a new two-stage heating pattern was developed. The first step (an isothermal holding at 1530°C for 150 seconds) and the second step (the other isothermal holding at 2050°C for 50 seconds) consist of this two-stage heating pattern.

4.3.1. A Typical Example (AK6)

AK6 specimen was used to test the proposed idea. O content in the specimen was analyzed by the conventional method (one-stage heating giving only [T. O]) and by the proposed method (two-stage heating giving [S. O] and [I. O], respectively). In Fig. 7(a), O signal in a piece of the AK6 is shown, which was obtained by the one-stage heating pattern (30 seconds at 2050°C). The analyzed [T. O] was 26.3 ± 1.3 ppm. In Fig. 7(b), O signal in the other piece of the AK6 is shown, which was obtained by the two-stage heating pattern (150 seconds at 1530°C, followed by 50 seconds at 2050°C). In the latter, two distinctive peaks were successfully observed. It is seen that the S. O and I. O in the AK6 specimen were extracted separately in a single analysis. The soluble O content ([S. O]) and the insoluble O content ([I. O]) were computed from the area under the signal from 0 second to 150 seconds and from 150 seconds to 200 seconds, respectively. These were 22.0 ± 2.6 ppm and 5.6 ± 6.5 ppm from seven tests, respectively. [T. O] can be computed from sum of the [S. O] and [I. O], which yields 27.5 ± 7.9 ppm. This is in good agreement with the independently analyzed [T. O] using the conventional method.

Fig. 7.

Oxygen signal generated from AK6 specimen: (a) conventional heating pattern (one-stage) (b) two-stage heating pattern proposed in the present study. Temperature and O content were calibrated as described in Sec. 3.2. (Online version in color.)

4.3.2. Soluble Oxygen and Insoluble Oxygen in AK Steel Specimens

Analyzed results for O content in twelve AK specimens (AK3 to AK14) are listed in Table 3. [T. O] was measured by the conventional (one-stage isothermal holding) method. [S. O] and [I. O] were independently measured by the present method (two-stage isothermal holding: 150 seconds at 1530°C, followed by 50 seconds at 2050°C). Sum of the [S. O] and [I. O] was also shown on the right-hand column. [T. O] of AK3 and AK4 specimens were high, although these were sampled after RH process. This is thought to be caused by local reoxidation when the specimens were sampled. One more specimen taken after the RH process and all other specimens, taken in TD, show that [T. O] measured by two different methods lie between 20 to 35 ppm. This looks reasonable value as the [T. O] in ULC steel before casting. The measured [S. O] and [I. O] are also shown in Fig. 8.

Table 3. Analysis results for the O content in AK specimens using the conventional (one-stage isothermal holding) method and the present method (two-stage isothermal holding). Condition of SEM analysis for counting I. O from inclusions are also listed.

Sample No.	One-stage heating		Two stage heating			SEM analysis
	[T. O]	# of analysis	[S. O]	[I. O]	[T. O]	[I. O] (ppm)			Aobs	dmin
	(ppm)		(ppm)	(ppm)	(ppm)	# of inc.	MD	AF	(mm2)	(μm)
AK3	243 ± 77	6	204.6 ± 73.5	68.4 ± 36.6	273.0 ± 106.7	509	29.2	90.0	0.541	0.12
AK4	465 ± 132	6	373.0 ± 124.3	116.6 ± 61.1	489.6 ± 116.6	1788	55.4	125.8	0.565
AK5	31.6 ± 8.1	11	23.7 ± 3.8	5.3 ± 5.1	29.0 ± 7.1	328	10.2	15.0	0.523
AK6	26.3 ± 1.3	7	22.0 ± 2.6	5.6 ± 6.5	27.5 ± 7.9	173	4.6	5.9	0.517
AK7	24.1 ± 2.8	8	22.9 ± 3.4	2.4 ± 1.2	23.9 ± 3.6	127	5.3	9.7	0.572
AK8	29.1 ± 1.6	7	19.3 ± 1.3	10.4 ± 3.3	29.6 ± 2.4	352	11.4	13.6	0.568
AK9	29.5 ± 4.2	5	18.5 ± 0.9	8.7 ± 7.1	28.2 ± 6.2	195	5.6	16.9	0.485
AK10	26.7 ± 1.3	9	21.1 ± 8.2	5.5 ± 3.7	26.6 ± 8.7	147	7.8	11.7	0.521
AK11	34.7 ± 1.6	7	25.4 ± 2.0	8.8 ± 2.9	34.2 ± 4.2	360	12.2	10.9	0.445
AK12	25.4 ± 1.6	7	20.3 ± 2.5	7.2 ± 3.9	27.5 ± 4.6	147	6.6	12.3	0.476
AK13	31.8 ± 2.9	7	24.9 ± 2.5	8.1 ± 2.5	33.0 ± 3.9	182	6.8	12.4	0.481
AK14	32.8 ± 8.6	7	21.0 ± 1.8	10.5 ± 6.5	31.5 ± 6.9	241	7.5	13.9	0.472

Fig. 8.

Analyzed [S. O] and [I. O] of AK specimens using the two-stage combustion analysis. (Online version in color.)

5. Discussions

The proposed new method in the present study shows a promising result for the O content analysis. This provides the O content in different forms: chemically dissolved O in steel and physically suspended O as oxide inclusion in the steel. Nevertheless, it is necessary to validate the present results. In the following two sections, two independent validations for the O content were carried out: one for [T. O] and the other for [I. O].

5.1. Validation for [T. O]: Comparison between Conventional Method and the Present Method

By neglecting a contribution from oxide film on each specimen, the Eq. (2) is reduced to:   

$$[\mathrm{T}.\mathrm{\hspace{0.17em} }\mathrm{O}]=[\mathrm{S}.\mathrm{\hspace{0.17em} }\mathrm{O}]+[\mathrm{I}.\mathrm{\hspace{0.17em} }\mathrm{O}]$$(6)

The obtained [S. O] and [I. O] of the AK specimens (from AK 5 to AK 14) were used to obtain [T. O]. The calculated [T. O] was compared with the independently measured [T. O] using the conventional method, and the results are shown in Fig. 9. All the data are listed in Table 3. A good agreement is seen for each specimen, within the uncertainty. This shows that the new method developed in the present study is able to analyze [T. O] in steel specimens, as the conventional method does.

Fig. 9.

Comparison between [T. O] of AK specimens (AK5 to AK14) measured by two different methods: the conventional method and the new method. [T. O] from the new method is sum of the [S. O] and [I. O]. (Online version in color.)

5.2. Validation for [I. O]: Comparison between Cross-Sectional Inclusion Analysis and the Present Method

In order to validate the analyzed [I. O] independently, it was attempted to analyze insoluble aluminum content ([I. Al]) after an electrolytic extraction of alumina inclusions in the specimens, followed by a wet-chemical analysis. However, this was not very successful, partly due to too small amount of alumina inclusion in our specimen, giving too low [I. Al] to be analyzed. This resulted in increasing uncertainty of the [I. Al] value. Therefore, it was decided not analyzing the [I. Al] by the electrolytic extraction, but analyzing the [I. Al] by SEM observation for the planar size distribution of alumina inclusion followed by conversion to spatial distribution in order to get the mass of alumina inclusions. It was confirmed that almost all the inclusions were alumina in the specimens used in the present study. An image analysis of alumina inclusions in the AK specimens was carried out by automated SEM (JEOL 7100, Tokyo, Japan) at a magnification of 3000, after polishing the cross-section of each specimen. The O of the alumina inclusions in the specimen should correspond to the [I. O] in the specimen. Total observed area was varied 0.445 to 0.568 mm² and the total number of inclusions was 127 to 1788, depending on each specimen. The minimum diameter of the inclusions (d_min) to be detected was set to 0.12 μm. Equivalent circular diameter of each inclusion was measured. This two-dimensional information was converted into three-dimensional information in order to calculate volume fraction of the inclusions in the specimen (f_V). This is then used to calculate an O content in the specimen, which corresponds to [I. O]. Two methods were employed, Mean Diameter (MD)^20,21) and Area Fraction (AF),²²⁾ to calculate [I. O] from the results of SEM analysis.²³⁾ Formulae for the volume fraction (f_V) of inclusions in a steel specimen is given in Table 4. The formulae for d_V (mean diameter) and N_V (the number of inclusions per unit volume) are the same in both methods. The f_V was computed by multiplying the cube of d_V by N_V in the MD method, and by dividing the sum of the cross-sectional area occupied by all the inclusions by A_obs, a given observed area in the cross section. The following assumptions were made to calculate the [I. O] from the calculated f_V: all the inclusions observed in the specimens were alumina that satisfies stoichiometry (Al₂O₃). Although the AK5 and the AK7 specimens contains some Ti (264 and 270 ppm, respectively), the same assumption was maintained. Previous investigations reported that Ti content in observed oxide inclusions is low and most inclusions had evolved toward alumina in Ti added Al killed steel.^24,25,26,27) Therefore, [I. O] of each specimen was computed by the equation given below, and the calculated [I. O] are given in Table 3:   

$$[\text{I}.\mathrm{\hspace{0.17em} }\text{O}](\text{ppm})=f_V\times \frac{{\rho }_{{\text{Al}}_2{\text{O}}_3}}{{\rho }_{\text{Fe}}}\times \frac{3{\text{M}}_{\text{O}}}{2{\text{M}}_{\text{Al}}+3{\text{M}}_{\text{O}}}\times {10}^6$$(7)

where ρ_Fe, ρ_Al2O3, M_O, and M_Al are the density of Fe (7860 kg m⁻³), the density of alumina (3950 kg m⁻³), the atomic weight of O (0.0159 kg mol⁻¹), and that of Al (0.0270 kg mol⁻¹), respectively.

Table 4. Mean diameter (d_V), the number of inclusions per unit volume (N_V), and volume fraction of the particles (f_V) estimation using Mean diameter(MD) method and Area Fraction(AF) method.^19,20,21,22) n, d_A(i), N_A, and A_obs are the number of inclusions, the cross-sectional diameter of an ith inclusion, the number of inclusion sections per unit area, and a given observed area in a cross section, respectively.

Method	dV	NV	fV
MD	$\left(\frac{\pi }{2}\right)\frac{n}{{{\displaystyle \sum }}_{i=1}^n\frac{1}{d_A(i)}}$	$\frac{N_A}{d_V}$	$\frac{\pi }{6}{\overline{d}}_V{}^3{N}_V$
AF	$\left(\frac{\pi }{2}\right)\frac{n}{{{\displaystyle \sum }}_{i=1}^n\frac{1}{d_A(i)}}$	$\frac{N_A}{d_V}$	$\frac{{{\displaystyle \sum }}_{i=1}^n\pi {\left(\frac{d_A}{2}\right)}^2}{A_{\text{obs}}}$

The calculated [I. O] of ten specimens (AK 5 to AK 14) are shown in Fig. 10. Those are compared with the [I. O] measured using the two-stage combustion method. Although some extent of scatters cannot be avoided, the calculated [I. O] by MD method is in favorable agreement with the measured [I. O] using the two-stage combustion method. The calculated [I. O] by AF method is seen to be somewhat higher than the measured [I. O]. According to Karasev and Suito,²³⁾ the f_V obtained from the cross-sectional data using the MD and AF methods can be greater than those obtained from chemical analysis, approximately two to three times for alumina inclusions. In this regard, the [I. O] measured by the two-stage combustion method is thought to be reliable.

Fig. 10.

Comparison between [I. O] of AK specimens (AK5 to AK14) measured by two different methods: an estimation from volume fraction of inclusions and the new method. [T. O] from the new method is sum of the [S. O] and [I. O].

5.3. Soluble Oxygen in Liquid Steel and Its Impact on Steel Cleanliness

Inspecting the measured [S. O] and [I. O] of AK specimens (AK5 to AK14) shown in Fig. 8 shows that variation of the uncertainty of the [S. O] is less than that of the [I. O]. Since the [I. O] stems from physically separated oxide particles in the specimen, the uncertainty of [I. O] would depend how the particles were homogeneously distributed. It is reasonable that the uncertainty of [I. O] is more significant than that of [S. O].

The AK specimens were sampled from liquid ULC steel deoxidized by Al, during RH and TD processes. In this regard, the measured [S. O] in the present study looks high. The [S. O] in Table 3 and the [S. Al] given in Table 1 were compared with the Al deoxidation equilibria as shown in Fig. 11. The curve was the most recently reported Al–O deoxidation equilibria at 1600°C,^28,29) using the Modified Quasichemical Model.³⁰⁾ It is seen that the [S. O] and [S. Al] (shown by open circles) were not in thermodynamic equilibrium. For reference, [T. O] were also shown by half-filled symbols at each given [S. Al]. Since the [S. O] accounts for only chemically dissolve O atoms in liquid steel, this discrepancy cannot be attributed to overestimation of O content by suspending alumina particles during the O analysis. This implies that the liquid steel might not be in thermodynamic equilibrium in terms of the Al deoxidation. Then, the supersaturation of O may be considered. Suito and co-workers have reported many possibilities of the supersaturation phenomena in Al deoxidized steel.^{31,32,33,34,35)} In order to confirm this hypothesis, two of the AK specimens (AK9 and AK11) were randomly chosen and were remelted in an alumina crucible heated by a halogen lamp in a gold image furnace. The specimens were fully melted at 1600°C in the alumina crucible under a purified Ar atmosphere. Those were then cooled to room temperature by a rate of 300°C min⁻¹. By the two-stage combustion analysis, [S. O] and [I. O] were analyzed. The results are shown in Fig. 12. Due to limited amount of both specimens (0.40 g, 0.46 g), only O content could be analyzed. Both specimens showed that [S. O] decreased and [I. O] increased after the remelting. Although Al content after the remelting was not measured, [S. Al] after the remelting was estimated by the stoichiometry of Al₂O₃ along with the decrease of [S. O]. The [S. Al] and [S. O] after the remelting was also shown in the Fig. 11 by filled symbols. It can be seen that the [S. Al] and [S. O] came close to the Al deoxidation equilibria after the remelting. [T. O] before and after the remelting did not vary significantly. This shows that the AK specimens (at least AK9 and AK11) were indeed supersaturated, and the remelting of small size specimen in the alumina crucible relieved the supersaturation. This is likely due to the alumina crucible, promoting heterogeneous nucleation.

Fig. 11.

Al and O contents of the AK specimens (AK5 to AK11): filled symbols ([S. Al] vs [T. O] as received), open symbols ([S. Al] vs [S. O] as received), and half-filled symbols ([S. Al] vs [S. O], after the remelting). Curves are the calculated equilibrium relation between [S. Al] and [S. O].²⁸⁾ (Online version in color.)

Fig. 12.

Analyzed [S. O] and [I. O] of AK9, AK11 specimens, as sampled vs remelted during the present study. (Online version in color.)

The above analysis suggests that liquid steel in a large scale vessel (such as TD in the case of AK9 and AK11) may be in the supersaturation state. In particular, reoxidation of ULC steel by open-eye formation,³⁶⁾ TD flux,³⁷⁾ or TD refractory³⁸⁾ causes increase of O content in the steel. If a critical supersaturation in terms of the free energy for homogeneous nucleation is not overcome, the O may stay as dissolved O in the steel. In the conventional combustion analysis for O in steel specimen, only [T. O] could be analyzed, and it was not possible to distinguish whether the O is S. O or I. O as is in alumina inclusion. Depending on the type of O in the steel, a way of improving the cleanliness of the liquid steel may be different. This raises a necessity of further investigations to confirm the existence of the supersaturation in the liquid steel in a large-scale vessel, and its consequences on the steel cleanliness.

6. Conclusion

A new method for analyzing O content in steel specimen was developed. The developed “two-stage” combustion analysis is indeed a modification of the conventional inert gas fusion infrared absorption method by heating the specimen at two different temperatures. Therefore, it is as simple as the conventional “one-stage” combustion analysis, which only gives [T. O] in the specimen. In addition to this, it was shown that the “two-stage” combustion analysis can simultaneously analyze [S. O], chemically dissolved O content, and [I. O], physically separated O content in oxide inclusion. The “two-stage” combustion analysis was applied to analyze the [S. O] and the [I. O] in a number of Al-killed ULC steel specimens, taken from RH and TD of a steelmaking plant. The [S. O] and the [I. O] were measured simultaneously. The measured [I. O] and [T. O] (= [S. O] + [I. O]) were independently validated, and were shown to be reliable. By the [S. O] and the [I. O] of the Al-killed ULC steel specimens, a possibility of the supersaturation was observed, thereby being away from the deoxidation equilibria. Remelting the supersaturated steel specimens resulted in approaching to the deoxidation equilibria.

Acknowledgement

The authors appreciate their gratitude to Mr. Jea-Bok Choi and Mr. Gwang-Chun Kim, Steelmaking Department, Gwangyang Works, POSCO for supplying the AK specimens, Dr. Woo-Yeol Cha, Technical Research Laboratories, POSCO and Prof. Sung-Mo Jung, POSTECH for valuable comments.

References

• 1)   T.  Ise,  Y.  Nuri,  Y.  Kato,  T.  Ohishi and  H.  Matsunaga: ISIJ Int., 38 (1998), 1362. https://doi.org/10.2355/isijinternational.38.1362
• 2)   K.  Ito and  M.  Koike: Tetsu-to-Hagané, 87 (2001), 756 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.87.12_756
• 3)   H.  Yasuhara,  M.  Shimura,  K.  Yoshioka,  K.  Abiko,  H.  Iwai and  T.  Niida: Phys. Status Solidi A, 189 (2002), 133. https://doi.org/10.1002/1521-396X(200201)189:1%3C133::AID-PSSA133%3E3.0.CO;2-5
• 4)   H.  Uchihara,  A.  Bando,  M.  Ikeda and  T.  Nakahara: Tetsu-to-Hagané, 89 (2003), 988 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.89.9_988
• 5)   S.  Yanagisawa and  T.  Sirai: Bunseki Kagaku, 15 (1966), 1161.
• 6)   T.  Shigematsu: Oyo Butsuri, 55 (1986), 1138.
• 7)   Y.  Miyashita: Tetsu-to-Hagané, 52 (1966), 1049 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.52.7_1049
• 8)   R.  Inoue and  H.  Suito: Mater. Trans. JIM, 32 (1991), 1164. https://doi.org/10.2320/matertrans1989.32.1164
• 9)   F. M.  Evans: Ph.D. thesis, Iowa State University, (1962), https://doi.org/10.31274/rtd-180813-4275, (accessed 2017-12-14).
• 10)   B. D.  Holt and  H. T.  Goodspeed: Anal. Chem., 35 (1963), 1510. https://doi.org/10.1021/ac60203a023
• 11)   T.  Koizumi,  F.  Tsugane and  M.  Kamakura: Tetsu-to-Hagané, 50 (1964), 787 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.50.5_787
• 12)   T.  Koizumi,  F.  Tsugane and  M.  Kamakura: Tetsu-to-Hagané, 50 (1964), 925 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.50.6_925
• 13)   T.  Yoshioka,  H.  Yamaguchi and  H.  Okochi: J. Jpn. Inst. Met., 49 (1985), 650. https://doi.org/10.2320/jinstmet1952.49.8_650
• 14)  ON836 Oxygen/Nitrogen Analyzer Instruction Manual, Ver. 1.9.x ed., LECO Corporation, St. Joseph, MI, (2013).
• 15)   K.  Furuya and  H.  Kamata: Bunseki Kagaku, 14 (1965), 708.
• 16)   M.  Singleton and  P.  Nash: Bull. Alloy Phase Diagr., 10 (1989), 121. https://doi.org/10.1007/BF02881419
• 17)   P. V.  Krasovskii and  K. V.  Grigorovich: Russ. Metall., 2002 (2002), 108.
• 18)   T.  Zienert,  S.  Dudczig,  O.  Fabrichnaya and  C. G.  Aneziris: Ceram. Int., 41 (2015), 2089. https://doi.org/10.1016/j.ceramint.2014.10.004
• 19)   R.  Khanna,  M.  Ikram-Ul Haq,  Y.  Wang,  S.  Seetharaman and  V.  Sahajwalla: Metall. Mater. Trans. B, 42 (2011), 677. https://doi.org/10.1007/s11663-011-9520-7
• 20)   R. L.  Fullman: JOM, 5 (1953), 447. https://doi.org/10.1007/BF03398971
• 21)   R.  Dehoff and  F.  Rhines: Trans. Metall. Soc. AIME, 221 (1961), 975.
• 22)   J.  Hillard: Quantitative Microscopy, McGraw-Hill, New York, (1968), 45.
• 23)   A.  Karasev and  H.  Suito: Metall. Mater. Trans. B, 30 (1999), 259. https://doi.org/10.1007/s11663-999-0055-0
• 24)   H.  Matsuura,  C.  Wang,  G.  Wen and  S.  Sridhar: ISIJ Int., 47 (2007), 1265. https://doi.org/10.2355/isijinternational.47.1265
• 25)   M.-A.  Van Ende,  M.  Guo,  R.  Dekkers,  M.  Burty,  J.  Van Dyck,  P. T.  Jones,  B.  Blanpain and  P.  Wollants: ISIJ Int., 49 (2009), 1133. https://doi.org/10.2355/isijinternational.49.1133
• 26)   M.  Wang,  Y.-P.  Bao,  H.  Cui,  H.-J.  Wu and  W.-S.  Wu: ISIJ Int., 50 (2010), 1606. https://doi.org/10.2355/isijinternational.50.1606
• 27)   E.  Zinngrebe,  C.  Van Hoek,  H.  Visser,  A.  Westendorp and  I.-H.  Jung: ISIJ Int., 52 (2012), 52. https://doi.org/10.2355/isijinternational.52.52
• 28)   M.-K.  Paek,  J.-M.  Jang,  Y.-B.  Kang and  J.-J.  Pak: Metall. Mater. Trans. B, 46 (2015), 1826. https://doi.org/10.1007/s11663-015-0368-0
• 29)   M.-K.  Paek,  J.-J.  Pak and  Y.-B.  Kang: Metall. Mater. Trans. B, 46 (2015), 2224. https://doi.org/10.1007/s11663-015-0369-z
• 30)   A. D.  Pelton,  S. A.  Degterov,  G.  Eriksson,  C.  Robelin and  Y.  Dessureault: Metall. Mater. Trans. B, 31 (2000), 651. https://doi.org/10.1007/s11663-000-0103-2
• 31)   K. R.  Lee and  H.  Suito: ISIJ Int., 35 (1995), 480. https://doi.org/10.2355/isijinternational.35.480
• 32)   K. R.  Lee and  H.  Suito: Metall. Mater. Trans. B, 27 (1996), 423. https://doi.org/10.1007/BF02914907
• 33)   H.  Suito,  H.  Inoue and  R.  Inoue: ISIJ Int., 31 (1991), 1381. https://doi.org/10.2355/isijinternational.31.1381
• 34)   H.  Suito: Proc. Ethem T. Turkdogan Symp., Iron and Steel Society, Warrendale, PA, (1994), 141.
• 35)   J.  Tanabe and  H.  Suito: Metall. Mater. Trans. B, 26 (1995), 95. https://doi.org/10.1007/BF02648982
• 36)   S.  Chatterjee,  D.  Li and  K.  Chattopadhyay: ISIJ Int., 56 (2016), 1889. https://doi.org/10.2355/isijinternational.ISIJINT-2016-239
• 37)   K.  Sasai and  Y.  Mizukami: ISIJ Int., 40 (2000), 40. https://doi.org/10.2355/isijinternational.40.40
• 38)   P.  Yan,  S.  Arnout,  M.-A.  Van Ende,  E.  Zinngrebe,  T.  Jones,  B.  Blanpain and  M.  Guo: Metall. Mater. Trans. B, 46 (2015), 1242. https://doi.org/10.1007/s11663-014-0287-5