2014 Volume 54 Issue 9 Pages 2109-2114
A new technique to study the kinetics of the carburization and decarburization processes of liquid iron is proposed. A liquid iron droplet was electromagnetically levitated in a CO–CO2 gas mixture during carburization and decarburization, and its surface oscillation was continuously recorded using a high-speed camera. The surface tension varied depending on each elementary step in the carburization and decarburization processes. This behavior was caused by transient adsorption and desorption of oxygen on the surface of the liquid iron accompanied by CO and CO2 gases. The kinetics of the carburization process were discussed and the conclusion was drawn that the desorption rate of oxygen adatoms in the form of CO2 was predominantly controlled by the gas-phase mass transfer.
Carburization and decarburization reactions of liquid iron are the most important reactions in the production of iron and steel. Many researchers1,2,3,4,5,6,7,8) have measured the reaction rates of carburization and decarburization, and investigated the rate-determining step under various conditions. Ito et al.2) measured the reaction rate of carburization using an electromagnetic levitation technique with CO–CO2 gas mixtures. They reported that the concentrations of carbon and oxygen in liquid iron rapidly increased when the CO partial pressure in the gas phase was increased during carburization. Subsequently, the carbon and oxygen concentrations gradually reached constant values. Thus, the carburization and decarburization behavior was studied based on concentration changes in the liquid iron. In the studies cited above, the samples were quenched after a certain time and chemical analysis was conducted for the quenched samples. Consequently, using this approach, it is not possible to study the rapid adsorption and desorption processes at the liquid iron surface.
Previously, we used the oscillating droplet method combined with a gas–liquid equilibrium method to demonstrate that oxygen acts as a strong surfactant in liquid iron, whereas carbon has a negligible effect on the surface tension.9,10) The equilibrium constant of the oxygen adsorption reaction at the liquid iron surface and surface excess of oxygen have been determined in previous studies.9,10) In the present study, based on surface adsorption data, we propose a continuous technique for studying the carburization and decarburization processes by measuring the surface tension using the oscillating droplet method.
Oxygen and carbon activities were controlled using Ar–He–CO–CO2 gas mixtures based on the following equilibrium reactions:
(1) |
(2) |
The experimental apparatus and technique used were similar to those described in detail elsewhere.9,10) High-purity (99.9972 mass%) iron prepared using an ion-exchange method12) was used as the sample. The iron sample of about 0.9 g was placed on a quartz sample holder and positioned in a levitation coil. The quartz tube (inner diameter: 18 mm) was evacuated to a pressure of the order of 10−2 Pa using a diaphragm pump and a turbo molecular pump, and then filled with high-purity Ar gas (99.9999 vol.%). The iron sample was initially levitated in an Ar gas atmosphere using the electromagnetic levitation (EML) facility, and an Ar–He–CO–CO2 gas mixture was introduced into the quartz tube to control the oxygen and carbon activities. Ar and He gases were purified using an Mg-deoxidizer kept at 873 K before introduction into the quartz tube. The liquid iron was carburized and decarburized by changing the CO and CO2 activities in the gas phase. The total gas flow rate was 2 L·min−1 at 273 K and 101.325 kPa, and it was estimated that the gas reached the sample after approximately 3 s based on the volume of gas tube.
The temperature of the sample was measured using a single-color pyrometer (wavelength: 0.9 μm; temperature resolution: 1 K; sampling rate: 2 Hz). The pyrometer was calibrated at the melting temperature of iron (1808 K). The temperature of the sample was controlled by changing the flow rates of Ar and He gases in the gas mixture and kept at 1873 ± 15 K. Images of the oscillating liquid iron droplet were recorded by a high-speed camera with a resolution of 256 × 256 pixels at a frame rate of 250 fps from the top of the sample.
Figure 1 shows the experimental conditions presented in the form of a log aO2 vs log aC plot at 1873 K. The initial condition (i) was controlled at log aO2 = −12 and log aC = −3 using a CO/CO2 gas mixture with aCO = 5 × 10−2 and aCO2 = 1 × 10−4. The sample was kept under this condition for 30 min to achieve equilibrium between the liquid iron and the gas phase. The surface tension was measured and the initial activities of CO and CO2 gases were then changed to aCO = 2 × 10−1 and aCO2 = 4 × 10−4 to obtain condition (ii) as indicated in the figure. Thus, the change from condition (i) to (ii) corresponds to a carburization process with constant aO2. The reverse process from condition (ii) to (i) is a decarburization process.
Experimental conditions for carburization from condition (i) to (ii) and the decarburization from condition (ii) to (i) at 1873 K presented as a log aO2 vs log aC plot.
In addition to the continuous experiments, some samples were quenched during carburization and decarburization and quantitative chemical analysis of oxygen and carbon was carried out to determine the time dependence of the composition, and also to assure the gas–liquid equilibrium. The carbon and oxygen contents were analyzed using an infrared-absorption method with a LECO CS-444 LS Carbon/ Sulfur Determinator and a LECO TC-436 Oxygen-Nitrogen Analyzer, respectively.
2.3. Determination of Surface TensionThe surface tension of liquid iron was calculated from the modified Rayleigh equation proposed by Cummings and Blackburn.13,14) The surface oscillation frequency and translational frequency of the center of gravity were determined through fast Fourier transform using sequential images of the oscillating droplet. The details of the frequency analysis were explained with consideration of sample rotation in our previous work.9) The sample radius was calculated using the density of liquid iron reported by Nishizuka et al.,15) which is given as:
(3) |
Figure 2 shows the time dependence of the surface tension of liquid iron during carburization, corresponding to a change from condition (i) to (ii) as indicated in Fig. 1. The oxygen activity in the gas phase was kept at log aO2 = −12. Here, the gas composition was changed at 0 s. The temperature variation caused by changing the gas composition was within 15 K. Although the oxygen activity in the gas phase was kept constant during carburization, the surface tension decreased significantly to 1675 mN·m−1 within 10 s of changing the gas composition [designated by (A) in Fig. 2]. Following this, the surface tension gradually increased and became constant within 300 s [designated by (B)].
Surface tension behavior of liquid iron during carburization from condition (i) to (ii) at 1873 K (solid circles). The surface tension was also estimated from the oxygen contents obtained from the quenching experiments (triangles).
After the carburization process described in the preceding section, decarburization from condition (ii) to (i) was conducted. The result is shown in Fig. 3. The temperature variation caused by changing the gas composition at 0 s was within 15 K. Although the oxygen activity in the gas phase was held constant at log aO2 = −12 during the decarburization process, the surface tension rapidly increased to 1870 mN·m−1 within 10 s of changing the gas composition [designated by (C) in Fig. 3]. Subsequently, the surface tension gradually decreased and became constant after 150 s [designated by (D)].
Surface tension behavior of liquid iron during decarburization from condition (ii) to (i) at 1873 K (solid circles). The surface tension was also estimated from the oxygen contents obtained from the quenching experiments (triangles).
Table 1 shows the oxygen and carbon contents in mass (wO and wC) obtained from the chemical analysis of samples quenched during carburization and decarburization. The values of aO2 and aC were calculated from wO and wC using the following thermodynamic relationships and data from the literature:11,16)
(4) |
(5) |
(6) |
(7) |
(8) |
(9) |
Process | Time | Results of chemical analysis | Calculated aO2 and aC using wO and wC | Estimated σ from aO2 | ||
---|---|---|---|---|---|---|
t/s | wO/mass% | wC/mass% | log aO2 | log aC | σ/mN·m–1 | |
Carburization | 0 | 0.0029 | 0.028 | –12.0 | –3.2 | 1813 |
10 | 0.0120 | 0.024 | –10.8 | –3.2 | 1653 | |
60 | 0.0054 | 0.044 | –11.5 | –3.0 | 1762 | |
200 | 0.0043 | 0.065 | –11.7 | –2.8 | 1785 | |
350 | 0.0036 | 0.074 | –11.9 | –2.7 | 1805 | |
Decarburization | 0 | 0.0030 | 0.101 | –12.0 | –2.6 | 1813 |
10 | 0.0008 | 0.095 | –13.2 | –2.6 | 1873 | |
60 | 0.0015 | 0.110 | –12.7 | –2.6 | 1856 | |
120 | 0.0022 | 0.097 | –12.3 | –2.6 | 1835 | |
400 | 0.0033 | 0.082 | –12.0 | –2.7 | 1813 |
Figure 4 displays the time dependence of aO2 and aC during the carburization and decarburization processes. Panels (A)–(D) correspond to the surface tension behavior presented in Figs. 2 and 3. The values of aO2 and aC calculated from wO and wC agree well with the values calculated from the CO/CO2 gas compositions used in the experiments at 0 s, which means that aO2 and aC were well-controlled using the gas–liquid equilibrium method. During carburization, aO2 in the liquid iron rapidly increased for 10 s after changing the gas composition, which is designated as process A. Subsequently, aO2 in the liquid iron gradually decreased with time and approached the equilibrium value (process B). During carburization, aO2 in the liquid iron rapidly decreased for 10 s after changing the gas composition (process C). Finally, aO2 in the liquid iron gradually increased with time and approached its initial value (process D).
Time dependence of activities of carbon and oxygen during carburization (solid circle) and decarburization (open squares) at 1873 K presented as a plot of log aO2 vs log aC.
The surface tension behavior during carburization and decarburization presented in Figs. 2 and 3 is discussed in this section. The overall carburization reaction is expressed by:
(10) |
(11) |
(12) |
(13) |
(14) |
(15) |
The adsorbed carbon atoms then dissolve into the iron melt, which is expressed by reaction (13). Conversely, the accumulated excess oxygen adatoms on the iron surface react with CO gas, and desorb as CO2 gas, which is expressed by reactions (14) and (15). During the desorption process, the surface tension gradually increases as shown in Fig. 2 (process B). This behavior can also be explained by the slower decrease in aO2 in the liquid iron from 10 to 350 s along the line presenting aCO = 2 × 10−1 in the gas phase (process B in Fig. 4). Thus, during process B, the liquid iron is not in equilibrium with the gas phase, but is in a transition state from the initial condition (i) to (ii). The kinetics in the process B is discussed later in the section 4. 3.
The surface tension behavior during the decarburization process shown in Fig. 3 can be also explained by the change in the number of adsorbed oxygen atoms. When the CO activity was decreased from 2 × 10−1 to 5 × 10−2 by changing the gas composition, the number of adsorbed oxygen atoms decreased, which was caused by the reverse reactions of (11) and (12). This decrease in the number of adsorbed oxygen atoms results in the rapid increase in the surface tension (process C in Fig. 3). This is supported by the experimental observation of a decrease in aO2 in the liquid iron for 10 s after changing the gas composition (process C in Fig. 4).
Subsequently, the transient deficiency of adsorbed oxygen atoms is recovered by supplying oxygen atoms from CO2 gas, as expressed by the reverse reactions of (14) and (15). As a result, the surface tension gradually decreases because of increasing oxygen adatoms (process D in Fig. 3). This behavior is also consistent with the slower increase in aO2 in the liquid iron from 10 to 400 s (process D in Fig. 4).
4.2. Changes in Fractional Surface Coverage and Oxygen Surface Excess during Carburization and DecarburizationThe equilibrium constant of the oxygen adsorption reaction on the liquid iron surface, Kad, is expressed in the form of a Langmuir isotherm:
(16) |
where θO is the fractional surface coverage. The value of Kad was determined as a function of temperature in our previous study as:9)
(17) |
Relationship between the fractional surface coverage of oxygen adatoms on the liquid iron for carburization (solid circles) and decarburization (open squares), and oxygen activity (solid line) at 1873 K.
The surface excess of oxygen, ΓO, can be obtained from the fractional coverage using the following relationship:
(18) |
Relationship between the surface excess of oxygen on the liquid iron for carburization (solid circles) and decarburization (open squares), and oxygen activity (solid line) at 1873 K.
The rate-determining step during carburization is discussed in this section. There are three steps for the carburization process:
1. Mass transfer in the gas phase
2. Chemical reactions on the liquid iron surface
3. Mass transfer in the liquid iron
The carburization process is divided into two processes: processes A and B, as indicated in Figs. 2, 4, 5 and 6. Process A was rapid and was complete within 10 s. This means that both gas-phase mass transfer and the chemical reactions taking place on the surface were fast. This was followed by process B, which took place slowly, over the course of 350 s. Considering the mass transfer in the liquid iron during process B, the surface tension values were estimated using the oxygen contents in the liquid iron obtained from the quenching experiments. The oxygen activities were calculated from the oxygen contents as presented in Table 1. The surface tension was previously determined as a function of oxygen activity as:9,10)
(19) |
The dissolved oxygen atoms transferred from the iron melt to the gas phase in the form of CO2 during the process B, which is expressed as:
(20) |
Here, nO is the amount of O atoms in the melt in mole, A [m2] is the surface area of the melt and t [s] is the time. The term on the left-hand side corresponds to the change in the oxygen atoms in units of mole per unit area and time. This is calculated from Table 1 during process B from 10 to 60 s. kg [m·s−1] is the gas film mass transfer coefficient and R is the gas constant (8.314 J·mol·K). PCO2s and PCO2b are the partial pressures of CO2 in Pa on the surface and in the bulk gas phase. PCO2s is calculated from the equilibrium relationship between CO2, O2 and C using aO2 and aC given in Table 1. PCO2 in the bulk gas phase is 4.0 × 10 Pa (aCO2 = 4.0 × 10−4). T is the absolute temperature in the boundary film in the gas phase, which is given by the average of the temperatures on the surface and in the bulk gas phase [T = 0.5 × (1873 + 300) = 1086.5 K]. Thus, kg was estimated to be 7.1 × 10−2 m·s−1 for process B from the experimental data.
Ito et al. also conducted the carburization experiments for levitated liquid iron droplets using CO–CO2 gas mixtures [2]. They concluded that the rate was initially controlled by the mass transfer in the melt, and after that, the rate was controlled by the counter diffusion of CO and CO2 in the gas phase. This well explains the behaviors found in the processes A and B obtained in the present study.
Finally, the following modeling using the Ranz–Marshall correlation is attempted for further discussion. The Sherwood number (Sh) is expressed in terms of the Reynolds number (Re) and the Schmidt number (Sc) according to the Ranz–Marshall correlation for mass transfer for a sphere under forced convection:17,18)
(21) |
The dimensionless numbers are defined by the following functions as:
(22) |
(23) |
(24) |
Here, d is the diameter of the droplet [m], which was estimated to be 6.3 × 10−3 m from the sample mass and density given in Eq. (3), DCO2 is the diffusion coefficient of CO2 in He gas [m2·s−1], ρg is the density of the gas [kg·m−3], vg is the velocity of the gas [m·s−1] and μg is the viscosity of the gas [Pa·s].
The following assumptions were made to evaluate kg using the Ranz–Marshall correlation.
1. Oxygen adatoms desorbed in the form of CO2 gas, which diffused from the liquid iron surface to the bulk gas phase through the boundary film.
2. The temperature of the boundary film in the gas phase is given by the average of the temperatures on the surface and in the bulk gas phase. DCO2 ρg vg μg are estimated at the average temperature (1086.5 K).
3. Helium is the dominant gas species during process B (PHe = 0.8 × 105 Pa).
The viscosity and diffusion coefficients for the He–CO2 system were estimated using the Chapman–Enskog equation.19) The values of Re and Sc were estimated to be 3.2 and 2.1, respectively. Using these values, Sh and kg were determined to be 3.4 and 2.7 × 10−1 m·s−1, respectively for process B. Thus, the estimated value is four times as large as the experimental one. However, it greatly depends on the temperature of the boundary film in the gas phase. Considering the uncertainty in the gas film temperature, they roughly agree with each other.
The surface tension of liquid iron was measured continuously during carburization and decarburization. The surface tension rapidly decreased in the initial stages of carburization and then gradually returned to its initial value. This behavior was caused by the transient adsorption and desorption of oxygen at the liquid iron surface accompanied by CO and CO2 gases. From the results obtained, the kinetics of the carburization process were studied, leading to the conclusion that the desorption rate of oxygen adatoms was predominantly controlled by the gas-phase mass transfer. Thus, we have proposed a new technique to study the kinetics of carburization and decarburization processes of liquid iron in a noncontact manner.
The authors thank Prof. T. Tsukada (Tohoku University), Prof. T. Hibiya (Keio University) and Assoc. Prof. S. Ozawa (Chiba Institute of Technology) for their helpful comments and discussion. This work was financially supported by SENTAN, Japan Science and Technology Agency (JST) and ISIJ Research Promotion Grant. One of the authors (KM) acknowledges support of the Grant-in-Aid for JSPS Fellows.