TITANIUM OXIDE HYDROGEN TRAPS IN TI-IF STEELS FOR ENAMELLING
Isabelle Tolleneer - Laboratory for Iron and Steelmaking, Ghent University, Belgium
Chris Rasschaert - Sidmar N.V., Arbed Group, Belgium
Franz Hörzenberger - Ocas N.V., Research Centre of the Sidmar Group, Belgium
Bruno C. De Cooman - Laboratory for Iron and Steelmaking, Ghent University, Belgium
Jan Penning - Laboratory for Iron and Steelmaking, Ghent University, Belgium

Abstract
Ti interstitial free (IF) steels designed for enamelling applications contain higher levels of carbon, nitrogen, sulphur and titanium than conventional IF steels in order to ensure that a sufficient amount of precipitates, i.e. hydrogen traps, is present. In this study the possibility of replacing part of the TiN, TiS and TiC with Ti-oxides through deoxidisa-tion of the liquid steel melt with Ti was investigated. Scanning electron microscopy at all stages of the laboratory processing allowed for an evaluation of the composition, the morphology and the size distribution of the inclusions and precipitates.
The influence of the increased oxygen content on the microstructure, the mechanical and enamelling properties was also examined in detail. Hydrogen permeation tests, pickling tests and laboratory scale enamelling were included with regard to the enamelling properties.

Introduction
Ti-stabilised interstitial free (IF) steels designed for enamelling applications contain high levels of C, N, S and Ti compared to conventional deep drawing IF steels, e.g. for auto-motive applications. This ensures that a sufficient amount of precipitates is present in the steel. Their presence increases the hydrogen uptake capacity of the steel substrate during enamel firing and lowers the risk of the occurrence of fish scale defects. The resistance against fish scale defect formation is related to the low hydrogen solubility in α-Fe, the fast diffusion of hydrogen at ambient temperatures and the enamel adherence. The hydrogen diffusion is very often evaluated by means of the hydrogen permeation index:

H₂I should be at least 6.67 min/mm² to ensure a low risk of fish scale formation. Previous investigations[1, 2] have shown that there is a direct relation between the number of precipitates in Ti IF steels for enamelling and the hydrogen permeation index. figure 1 shows the increase in H2I with increasing precipitate volume fraction. The occurrence of fish scale defects on laboratory enamelled coupons is also indicated on the figure: the risk of fish scale defect formation increases with decreasing hydrogen permeation index.
A large amount of precipitates is however known to be detrimental for the deep drawing performance of an IF steel. In addition, the higher Ti-content causes casting difficulties because of nozzle clogging. Therefore, the possibility of replacing part of the TiN, Ti₄C₂S₂, TiS and TiC precipitates with Ti-oxide inclusions was investigated. Through laboratory vacuum melting and deoxidisation with Ti an ingot containing a substantial amount of Ti-oxides was produced.

The influence of the oxygen content on both the mechanical and enamelling properties was investigated in detail. Microstructure analysis was combined with scanning electron microscopy at all stages of the processing route in order to evaluate the composition, the morphology and size distribution of the inclusions and precipitates.

figure 1 - Influence of the volume fraction of precipitates on H2I and the appearance of fish scales, for Ti-stabilised IF steels processed with different slab reheating tem-peratures (SRT) and coiling temperatures (CT)

Deoxidisation with Ti
Oxides are formed in two stages: primary oxides form during cooling of the steel melt from teeming to liquidus temperature and secondary oxides form during solidification from liquidus to solidus temperature[3]. The first are due to thedecreasing solubility of oxygen in the liquid phase, whilst the second result from the lower solubility of oxygen in the solidified steel.
After vacuum degassing and decarburisation the IF steel melt is deoxidised with Al. Primary Al2O3-oxides form, float and can be separated from the molten steel. Subse-quently the remaining content of dissolved oxygen is very low and little or no secondary oxides will be formed during solidification[4,5,6].
When the steel melt is however deoxidised with Ti, which has a weaker deoxidisation potential than Al, an increased amount of oxygen remains dissolved in the molten steel.
Thus during cooling and solidification fine oxides can precipitate. As they are fine enough they are prevented to float at the surface in the liquid steel by the viscous force. According to Stokes' law the floating velocity is given by equation (2) and the floating distance by equation (3):

The floating distance decreases with decreasing oxide size as is shown in figure 2. A particle of 10 μm diameter rises only 11 cm after 60 minutes in a static melt.
A typical oxide particle size distribution for a Ti-deoxidised steel is also given in figure 2: the average diameter is approximately 4 μm.

figure 2 - Floating distance of oxides in molten steel

Studies of Ti-deoxidised steels by Goto et al.[4,5,6] showed that almost all the oxygen in the steel is present as fine complex Ti-Al-Mn oxides. The oxides that precipitate during cooling and solidification represent about 70% of the total oxide amount in the steel and are generally smaller than 10 μm.
When the cooling rate is increased the diameter of the oxides decreases even more and the number of oxides increases.
Honma et al.[10] reported that in Ti-deoxidised steel these complex oxides are composed of Ti₂O₃ contain-ing additionally Al₂O₃ and MnO.

Materials
In this study a Ti-deoxidised steel was produced by laboratory vacuum melting and cast-ing. The chemical composition of steel T is given in Table 1.
An Al-deoxidised IF steel (steel R) with a comparable amount of N, S and C was used as a reference.

In order to obtain a Ti-stabilised IF steel the excess amount Ti has to be larger than zero. In the case of steel T also Ti-oxides are formed, which means an extra term has to be added to the formula (equation (4)) for calculating Ti_exc, i.e. Ti_tot minus the amount of Ti needed to stabilise all O, N, S and C:

Since steel T contains a small amount of Al it is assumed that also Al₂O₃ is formed by the Al that is not in solid solution, i.e. the total Al-content (Al_tot) minus the Alcontent in solid solution (Al_sol).
Part of the oxygen will not be available to form Ti₂O₃ and has to be excluded from the O-content in equation (4).
This is shown for steel T in Table 2. For steel R the Tiexc-content is 105 ppm.

Blocks of 25 mm thickness were reheated for 1 hour at 1250 °C and hot rolled to 4.5 mm in four stages with a finishing temperature above A_r3. After hot rolling the sheets were air cooled to a coiling temperature of 700 °C. Following the coil cooling simulation the sheets were acid pickled and cold rolled to a thickness of 1.5 mm (66% CR) or 0.8 mm (83%CR).
Subsequently all sheets were batch annealed (BA) under a HN_x gas atmosphere and temper rolled 0.4%.

Analysis and testing
Steels T and R were examined with both light optical microscopy (LOM) and a Zeiss DSM 962 scanning electron microscope (SEM) after casting, hot rolling and coiling and after annealing.
In addition automatic SEM particle analysis was carried out after hot rolling and coiling to determine the particle size distribution (figure 2) and quantitatively analyse the composition of the precipitates.
The mechanical properties were determined by tensile tests, whilst material drawability was evaluated by measuring the mean normal anisotropy, rmean. The hydrogen permeation time t₀ was measured by means of a Ströhlein apparatus as described by the EN10209 standard [11].
A sheet sample is fixed in a holder with at the upper side a container with a hydrogen-producing electrolyte. At the bottom side of the sample there is a funnel-shaped part entirely filled with water and connected with a capillary. As H diffuses through the steel sheet small gas bubbles will be formed and cause the capillary water level to rise.
A plotter and a PC register the water level. From this curve t₀ can be determined as the intersection point of the time axis with a tangent through the point were the water level starts rising sharply.
Using equation (1), the hydrogen permeation index was calculated as a measure for the susceptibility for fish scale formation.
This was combined with the 2 side enamelling of small flat panels (coupons of 80 x 250 mm) with a 2 coat/1 fire wet enamel application technique, currently used for sanitary ware and household appliances, in order to also visually evaluate the enamel layer.
The adherence of the enamel was tested by a conventional impact test [11]. Pickling tests were done [11] to evaluate the surface reactivity of the steels.
This is important in view of the direct white enamel (DWE) application technique where the steel sheets have to be pickled before applying a Ni-flash and enamel layer.

Microstructure analysis
Light optical microscopy (LOM) investigation of steel T showed that large precipitates, present in the as-cast condition, appeared broken and aligned parallel to the rolling direction after annealing (figure 3).
It can be assumed that the process of breaking the Ti-oxides during cold rolling increases the resistance against fish scaling due to the creation of micro-voids around these areas of broken particles.
A mechanism that is comparable with the principle of broken cementite particles in regular low carbon steel grades for enamelling applications [12,13]. In steel R no large inclusions or precipitates were visible with LOM

figure 3 - LOM of steel T in as-cast condition (left) and 66% cold rolled + batch annealed (right)

In order to identify the composition of the precipitates in steel T EDS investigations were carried out. In the as-cast condition both oxide inclusions, TiN and TiS precipitates were observed. The spectrum shown in figure 4 is characteristic for the complex Ti-oxides found in steel T: clear Ka-peaks for Ti and
O together with a small but distinct Al Ka-peak.
Most oxides were identified as this type of oxide containing both Ti and Al, referred to as Ti[Al]-oxide in this paper. Except for a few cases the spectrum of the precipitates always contained the Al Ka-peak.
Particles with an average diameter of 5-10 μm were all identified as oxides.
Whilst those with a diameter of 1-3 μm were identified as TiN or TiS but also some as oxides.
The TiN precipitates had a characteristic ideomorphic rectangular shape. The TiS precipitates were coprecipitated with TiN or Ti[Al]-oxide.

figure 4 - EDS-spectrum of Ti[Al]-oxide

In the hot rolled samples of steel R only TiN and TiS precipitates were found,
in some cases with a nucleus of Al₂O₃. A large number of particles smaller than 1 μm were observed in both steel T and R and could because of the limited resolution of the EDS-system of the SEM not be identified.
In the cold rolled samples of steel T and R the same types of precipitates were observed as in the hot rolled condition: TiN and TiS in both steels, Ti[Al]- oxides only in steel T.
However more important is that the large oxides have been broken by the rolling strain into smaller particles, which are aligned in the rolling direction. This suggests that during this process microvoids could be created around the deformed areas. A few of the larger TiN were also found broken. A typical micrograph is shown in figure 5.

figure 5 - Compositional SEM micrograph of steel T after a cold rolling thickness reduction of 66% and batch annealing

Composition of the Ti-oxides
In addition to the investigation of the microstructures with LOM and SEM hot rolled samples of steel T and R were also investigated by means of an automatic SEM particle analysis technique.
For each sample at least 300 particles were measured, determining their Ti, Al and O-content.
Since with SEM also a part of the bulk of the sample is measured the Fe intensities were very high. The intensity of the other elements such as Ti, Al or O increases as the particle size increases.
Compositional data obtained by EDS for particles with a specific stoichiometry but of different sizes must lie on a straight line with a slope determined by the stoichiometry of the compound formed.
For steel R no correlation was found between the atomic mass fractions in %of Al and O, although during SEM investigation of the hot rolled material some Al2O3 were observed.
The size distribution of the particles revealed that the average particle diameter was only 0.99 μm, which was very near to the lower measuring limitof the SEM.
In steel T many particles clearly showed a correlation between the Ti and O-content of the particles. Figure 6(a) shows that most particles have a Ti/O atomic ratio between 0.4 and 0.7 and could thus be identified as TiO₂, Ti₂O₃ or Ti₃O₅.
As mentioned above Goto et Al. [4,5,6] and Honma et Al. [10] reported that Tideoxidised steels contain complex Ti₂O₃-Al₂O₃ oxides.
The particles are conglomerates of separate Ti₂O₃ and Al₂O₃. The averageTi/O ratio of the particles measured in steel T was however 0.595 instead of 0.666, whilst on the other hand the average Al/O ratio was only 0.07.
Yamamoto et al.[14] refer to the fact that Ti₂O₃ is a cation-deficient oxide so in steel T these cation vacancies could partially be filled by Al³⁺-cations.
In both cases the average (Ti+Al)/O atomic ratio should be near to 2/3, the stoichiometric ratio of the oxides.
Figure 6(b) shows the (Ti+Al)/O atomic ratio distribution with a Gaussian fitting to determine the mean value.
The average ratio was found to be 0.638. In this investigation hardly any separate Al₂O₃ oxides were observed and because of the very low Al/O ratio the most likely composition of the oxides is that of Ti₂O₃ in which part of the Ti³⁺-cation vacancies has been filled by Al³⁺-cations.
Based on the average Ti/O and Al/O ratios the stoichiometry is Ti_2-XAl_XO₃ with x approximately 0.2.
The option of the formation of complex Ti₂O₃-Al₂O₃ oxides can however not be excluded entirely without further detailed investigation of the precipitates.

a	b
figure 6 - Correlation between Ti and O- content (a) and distribution of (Ti+Al)/O atomic ratio (b) for steel T

Influence of the oxides on the enamelling properties
The hydrogen permeation indices of both the hot and cold rolled material were determined. The thickness of the hot rolled samples was reduced from 4.5 mm to 1.5 mm by grinding them on both sides. For each material two permeation tests were carried out and the results were averaged.
Figure 7 shows the results of both the automated method (right hand column) and the visual observation measurements (left hand column).
The H₂I-values of the hot rolled samples suggest that the presence of the Ti[Al]-oxides without the mechanism of breaking them up through cold rolling has only a small influence on the hydrogen permeability.
In the hot rolled samples of steel T a few broken particles were observed, which could explain the higher H₂I of steel T compared to steel R. The finished material of steel T however shows a large increase in hydrogen permeation index. Steel T has a H₂I of approximately ten times that of steel R.
The mechanism of breaking up the oxides during cold rolling clearly improves the resistance against fish scaling.
In addition the material with 83% cold reduction has a higher H₂I than the material with 66% cold reduction since the grain boundaries also act as hydrogen traps.

figure 7 - Hydrogen permeation indices for steels R and T

The 2 coat/ 1 fire laboratory enamelling tests confirmed the results of the hydrogen permeation tests. The coupons were sprayed with a ground coat of approximately 80 μm, immediately followed by spraying on a top coat of 120 μm.

The coated sheets were dried at 80 °C and than fired at 800 °C, 820 °C and 840 °C in a gradient furnace for 4 minutes (4.5 minutes for the 1.5 mm sheets). The coupons were then left for at least 24 hours before evaluating the surface. The enamelled coupons of steel T showed no fish scale defects, whilst those of steel R were covered with fish scales (figure 8).

figure 8 - Example of fish scales on steel R (2 side enamelling - 2 coat/1 fire)

The pickling tests showed that the Ti-deoxidised steel T had a lower surface reactivity than steel R. A higher O content resulted in a lower pickling rate, but the level of reac-tivity was sufficiently high for DWE application. The drop test was performed on the three firing areas of the strip. For the 0.8 mm sheets the adhesion is excellent, especially at a firing temperature of 820 °C. For the 1.5 mm sheets the adherence to the reference steel is better than the adherence to the steel containing the Ti[Al]-oxides.
The best results are at a medium (820 °C) to high (840 °C) firing temperature.
At 800 °C the adhesion is clearly less. An overview of the adhesion ratings is given in figure 9. The graph also suggests that adherence improves with increased cold reduction.
Direct comparison of the ratings is however difficult since the drop weight falls from a different height as prescribed by the standard: 50 cm for the 0.8 mm sheets and 75 cm for the 1.5 mm sheets, so a part of the improvement could be due to that difference.

figure 9 - Adhesion performance with a 2 coat / 1 fire enamel

Influence of the O-content on the mechanical properties
The mechanical properties of steel T and steel R, determined through tensile testing, were compared. It was found that an increased O-content resulted in only a small de-rease in rmean-value, especially for the sheets with 83% cold reduction. The elongation increased slightly with increasing O-content (figure 10). With regard to the strain hardening exponent n the increased amount of oxides led to a minor decrease of 0.005.
In all cases the mechanical properties of steel T complied with standard EN10209[11]: R_p0.2< 190 MPa, r_mean> 1.6 and A_mean> 38%. In general the O-content had almost no influence on the mechanical properties.

figure 10 - Amean and rmean versus O-content

Conclusions
Ti-stabilised interstitial free steels designed for enamelling applications combine good deep drawability with good enamelling properties by increasing the Ti, C, N and S-content in comparison with conventional deep drawing IF steels. In this study the possibility of adding Ti-oxides as an alternative type of precipitatewas investigated.
Investigation of the material at all stages of the processing showed that the Ti-deoxidised steel contained large oxide particles containing both Ti and Al. All particles with an average diameter of 5 -10 μm were identified as this typeof oxide. Precipitates with a diameter of 1-3 μm were found to be either TiN, TiS or Ti[Al]-oxide. More detailed automatic SEM particle analysis allowed to determinethe most likely composition of the oxides.
Their (Ti+Al)/O atomic ratio was close to 2/3, which means that they are either complex Ti₂O₃-Al₂O₃ oxides or cation-deficient Ti₂O₃ oxides in which part of the vacancies has been filled by Al³⁺-cations.
With regard to the fish scale resistance the most important observation was the breaking of the oxides by the rolling strain into smaller particles, aligned to the rolling direction.
During this breaking up process micro-voids were created, which acted as irreversible hydrogen traps.
These observations were confirmed by the hydrogen permeation measurements. The Ti-deoxidised steel had a hydrogen permeation index almost tenfold that of the reference steel.
The laboratory enamelling of coupons confirmed their superior enamelling properties: no fish scale defects were observed. The adhesion of the enamel was good-to-excellent, especially for the 0.8 mm material. In addition the presence of the oxides had no distinct negative influence on the mechanical properties. There was only a very minor decrease in rmean and n-values.
In industrial practice the use of Ti-oxides as hydrogen traps in IF steels could offer interesting perspectives. However, apart from the trend to produce “pure” inclusion-free steels, the deoxidisation with Ti might complicate the adjustment of the Tiexc level necessary to optimise the mechanical properties.
On a laboratory scale Ti-oxides have proven to increase clearly the resistance against fish scaling.
The possibility of working with a lower O-content and still guaranteeing good enamelling properties requires further investigation.