PLASMA ENAMELLING: A NEW PROCESS
D.T. Gawne - South Bank University, Great Britain
Y. Bao - South Bank University, Great Britain
T. Zhang - South Bank University, Great Britain

Abstract
The research shows that plasma spraying can produce dense coatings of vitreous enamel on steel. The enamel feedstock powder is fused separately in the plasma while the substrate remains at a low temperature.
This enables enamelling to be carried out in a single stage operation without the need for a furnace, which offers the potential of widening the applicability of enamel coatings.
A computer simulation has been developed that provides comprehensive data for understanding the process and its control.
The results from experimental trials were found to be in agreement with those from the computational model.

Introduction
In conventional vitreous enamelling, the enamel slurry or powder is applied to a component and both are heated one or more times in a furnace to approximately 850 °C.
This temperature is necessary in order to fuse the enamel to the steel but is also a source of difficulty in both processing and product quality.
Specifically, the need for the furnace treatment limits the size of the component to the furnace dimensions and eliminates the possibility of on-site enamelling.
The repair of damaged or defective enamelled articles also cannot be carried out on-site satisfactorily and complete re-processing is normally required. As regards the product, besides the risk of softening and distortion of the steel substrate, hydrogen may be evolved from the steel causing pinholing and fishscaling in the enamel.
The above limitations of traditional enamelling derive from the need to heat both the enamel and the metal substrate to the fusion temperature.
Thermal spraying, particularly plasma spraying, has the potential advantage that the heat source is separated from the substrate.
The coating powder is melted in the plasma and the temperature of the substrate can be maintained at a temperature as low as 100 °C if necessary.
There is no furnace treatment and so, in principle, no restriction on component size.
On-site enamelling of large components and on-site repair therefore become possibilities. Furthermore, the low substrate temperature is expected to eliminate pinholing and fishscaling in the enamel.
Plasma spraying consists of injecting a feedstock powder into a plasma jet in which it is heated, accelerated and projected on to a substrate to form a coating.
A schematic of the process is given in figure 1.
The high gas temperature (10 000 to 15 000K), velocity (~300 ms-1) and heat content of the jet enables glasses and ceramics to be readily melted and deposited [1,2].
The process combines particle melting, cooling and compaction into a single process unlike conventional multi-stage enamelling.

figure 1 - Schematic of plasma spraying

The formation of the coating takes place in a number of steps. Heating and fusion of the enamel feedstock particles in the hot jet, acceleration of the particles to a high velocity by the jet, impact and flow of particle into a discshaped splat on the substrate surface, and finally, the accumulation and compaction of splats into a coating as a consequence of the scanning action of the spray-gun.
This sequence of events is shown schematically in figure 2. The residence time of a particle in the flame is less than one millisecond and so kinetics rather than thermodynamics dominate the mechanisms.
Heat transfer from the plasma gas to the particle interior, the fluid mechanics of the jet flow and the rheology of the enamel flow to form splats are important for the formation of dense, high-quality coatings.

figure 2 - Schematic showing stages of evolution of deposit during thermal spraying deposition

The properties of the enamel coating are governed by its microstructure, which depends on the process parameters and the enamel composition.
Enamels have high melt viscosities and so in order to achieve adequate flow of the splats on impact, the particles need to be accelerated to a sufficiently high velocity and heated to a high enough temperature.
The heating process is complicated by the relatively low thermal conductivity of enamels, which is expected to produce large temperature gradients in the particles during their flight from the gun to the substrate.
An understanding of the process mechanisms is therefore essential for the development of a new enamelling technique based on plasma spraying.
This paper presents a computational model for plasma-spray enamelling based on heat transfer and fluid mechanics. The model simulates the process and computes the temperature profiles of the in-flight particles, the coating and the substrate during plasma spraying.
The computed results are compared with those from experimental trials. The research is aimed at investigating the feasibility of enamelling by plasma spraying.

Experimental details
The coating material was a typical black acid-resistant frit with a softening temperature of 518 °C supplied by Escol Products Ltd (Huntingdon, UK), and milled and sieved to different sizes by Corus plc (Port Talbot, UK). The substrate material was aluminium-killed sheet steel of thickness 3mm, supplied by Corus.
The steel was degreased with acetone and surface roughened by grit blasting (Metcolite C, alumina grit, Sulzer-Metco Ltd.) with a pressure-operated machine to give a surface roughness of 6 μm (Ra). Plasma spraying was undertaken using a Sulzer-Metco plasma spray system with an MBN torch and MCN control unit. A Sulzer-Metco 4MP powder feed unit and fluidized bed hopper was used to feed the powder into the plasma jet.
The temperature of the substrate during plasma spraying was monitored using thermocouples attached to the back of the steel substrate and the data was recorded on computer.

Computational model
The temperature of in-flight particles
A theoretical analysis has been carried out to predict the temperature profile of an enamel feedstock particle inside a plasma jet under various process parameters.
To simplify the calculation the following assumptions have been made:

The feedstock particle is spherical.

The heat transfer within a spherical particle immersed in a thermal plasma can be described by a special form of the general conduction equation in a spherical polar coordinate system^[3]:

(1)

figure 3 - Schematic of an inflight particle of radius R and a surface temperature Ts in a thermal plasma of temperature Tp, where r is the radial distance fron the particle centre

An energy balance method was used for the calculation of temperature at the surface node s in which its internal energy change is equated to the heat transfer from the surrounding plasma and that from the adjacent internal node s-1 within the particle. This may be expressed as follows:

(2)

where Tp represents the plasma temperature, which is a function of the axial and radial distance from the nozzle of the plasma gun. Ts-1 is the temperature of the first internal node. h is the convective heat transfer coefficient between the plasma and the particle, which is dependent on the temperature and thermal conductivity of the plasma, and the surface temperature and size of the particle. The value of h was determined by the method given by Bourdon et a1 (4), which is now the widely accepted technique (2,5-7). An integral mean value of the thermal conductivity across the boundary layer, k, is given by the expression:

(3)

which may be used for the heat transfer calculation under plasma conditions. The heat transfer coefficient is now given by:

(4)

where k(T) is the thermal conductivity of the plasma gas at temperature T. It has been shown that equations 3 and 4 can be used provided the residence time of the particles in the plasma jet is longer than 1 μs.
This condition is readily satisfied in the current experiments and equations 3 and 4 are used to calculate the heat transfer coefficient, h, of the plasma to the particles.
Accordingly, a set of finite difference equations was written based on equations 1 and 2 and used to obtain numerical solutions.
The enthalpies of fusion and decomposition together with the change in the particle radius during decomposition were considered in the calculations.

Temperature profile of substrate and coating
A coating is formed by the repeated scanning of the substrate with a plasma torch as shown in figure 1 and 2.
The horizontal distance between the nozzle exit and the vertical plane of the substrate is fixed throughout spraying, usually in the region of 100 to 130 mm.
However, the distance between the intersection of central axis of the plasma jet with the substrate and any other position P on the substrate surface must vary with the scanning action during spraying. This is shown schematically in figure 4.

figure 4 - Schematic of plasma scanning of the substrate

The temperature and velocity of the flame or hot gas immediately in front of point P on the substrate are functions of scanning time as the torch moves across the substrate.
This gas temperature governs the transfer of heat from the flame to the substrate and thus the resulting temperature of the coated component.
The component temperature during spraying will exert a substantial effect on the product quality and so it is important that it is quantified.
In order to simulate the heating of a given point P on the substrate surface, the temperature profile and the heat transfer coefficient of the flame to the substrate must be pre-defined.
Previous research work carried out by Tollmien [8] has indicated that the temperature of a jet may be described by a Gaussian distribution.
However, this is only valid for a free jet and not one impinging on a substrate.
Consequently, the authors have used a computational fluid dynamics (CFD) package to simulate the latter situation and determine the temperature distribution of the impinging jet.
The computational method is explained in earlier papers by the authors [9, 10]. The method is applied to the current case of a plasma jet and the results are given in figure 5. This predicts how the gas temperature in the plasma jet varies with the radial distance from the central axis. Now that the temperature of the impinging jet has been calculated, the next step is to determine how much it raises the temperature of the substrate. This requires a knowledge of the heat transfer coefficient. Shimizu [11] has carried out experimental work to measure the heat transfer coefficient of a jet impinging on a plate.
The measured data provide the heat transfer coefficient as a function of the radial distance from the axis of the jet. The experimental conditions are similar to those for a plasma jet impinging on the substrate and so the data may be used in the present work.
The simulation is based on a 100x100 mm mild steel substrate with thickness of 1 mm. The flame axis is normal to the substrate and the spraying distance is 120 mm.
It is assumed that there is a pre-deposited layer of glass coating of thickness 0.1 mm on the substrate.
The plasma gas is taken to be argon with 5% hydrogen and the gas flow rate is 100 standard litres per minute.
To simplify the computation model, the following assumptions have been made:

The heat transfer in the substrate is one-dimensional (since the thickness is much smaller than the width of the substrate);

Based on these assumptions, the governing equation for the heat transfer in the coating and substrate can be expressed as:

(5)

where T is the temperature within the deposit or substrate at a distance z from the cold face of the substrate and D is the thermal diffusivity, which is a function of temperature.

figure 5 - The calculated plasma jet temperature immediately above a fixed point (P) on the substrate surface during plasma scanning

At the boundary between the substrate and the coating, a combined element is considered. Since the coating material is an amorphous glass and the temperature is always lower than the melting point of the steel substrate,
no latent heat changes are involved in the computer model.

Results and discussion
Effect of plasma flame on the heating of in-flight particles
The computational model described above was applied the current case of enamel particles in a plasma jet.
Initially, the effect of the hydrogen content to the heating of in-flight particles (diameter 60 μm) during plasma deposition was predicted using the model.
The results are given in figure 6, which shows the predicted temperatures at the surface, half radius and centre of the enamel particle as it travels from its injection near the nozzle exit towards the substrate during spraying.
figure 6a first gives the temperatures of a particle in a pure argon flame and shows a large temperature gradient from the surface to the centre of the particle.
Furthermore, the temperature of the particle only reaches 270 °C and 90 °C at the surface and centre respectively. In physical terms, the particles remain in the solid state and rebound off the substrate on impact.
As a result, deposition of an enamel coating using an argon flame is predicted to be impossible.
The addition of hydrogen to argon in the plasma gas is known to increase both the heat transfer coefficient and the gas temperature.
Hydrogen contents of 5 vol% and 10 vol% were therefore added to the plasma gas and introduced into the computational model.
The results are given in figure 6b for the 5% hydrogen-argon plasma and immediately show a dramatic increase in the particle temperature.
The surface temperature of the particle now reaches 900 °C as compared ith 270 °C in the pure argon plasma. The temperature gradient within a particle is important, since it is necessary to obtain melting throughout the thickness not only at the surface.
At short distances from the nozzle exit, the temperature gradient the between the surface and centre of the particle is large and reaches 400 °C at a distance of 30 mm.
However, the gradient gradually reduces and is only 40 °C at 120 mm, which is approximately the spraying distance. This means that the particle has a virtually uniform temperature by the time it impacts with the substrate.
The reason for this equalization of temperature is because the temperature of the plasma flame falls rapidly along the axis of the flame: from approximately 10,000 °C at near the nozzle exit down to 2,500 °C at 120 mm along the central axis. This lessens the heat transfer from the gas to the particle surface, while the rate of internal conduction from the surface to the centre of the particle remains high.
The effect of adding 10% hydrogen to argon was also computed using the model and the results given in figure 6c.
The temperature of the particle surface now reaches 1500 °C with its centre being only 40 °C lower. These temperatures are so high that they may possibly cause degradation of the enamel. Experimental trials were carried out to test the above predictions.
It was found that no coatings could be produced with a pure argon plasma whereas coatings were formed using a 5% hydrogen-argon mixture. This behaviour is in agreement with the computational model.
However, the 10% hydrogen-argon plasma produced dense enamel coatings with no evidence of thermal decomposition.
This is surprising in view of the known decomposition characteristics of enamel. It is likely that this is a consequence of the extremely rapid kinetics of plasma spray deposition rather than a limitation of the model.

figure 6a

figure 6b

figure 6c

figure 6 - Calculated temperature of a 60 μm enamel particle at its surface, half radius and centre under different plasma conditions: (a) pure Ar, (b) 5% H2 - Ar, (c) 10% H2 - Ar

The velocity of an enamel particle entrained in the high-speed plasma jet will be 100-200 m/s over a spraying distance of 100 mm, which gives a residence time of less than one millisecond.
The results indicate that decomposition of the enamel does not take place, possibly due to nucleation difficulties of decomposition products, such as gases, in the extremely short time period.
This means that the enamel can exist at much higher temperatures than under equilibrium conditions and this can bring benefits to the coating quality.
In order to form a dense coating, the impacting particle has to flow extensively into a thin splat in less than a millisecond as illustrated in figure 2.
Enamels, like most other glasses, have inherently low viscosities and the abnormally high temperatures achievable in plasma spraying facilitate largescale viscous flow, which enables the formation of dense coatings.

The effect of particle size on the heating of in-flight particles
The model was applied to calculate the effect of the particle diameter on the temperature of in-flight particles.

figure 7 - Calculated temperatures of a 90 μm enamel particle at its surface, half radius and centre at varying plasma conditions

figure 7 gives the temperature profile of a 90 mm particle (D50 = 90 μm) in a 5% hydrogen-argon plasma and a 10% hydrogen-argon plasma. The results show that in a 5% hydrogen flame, the particle only reached a maximum of 370 °C at its surface and so is predicted to remain in the solid state.
With 10% hydrogen, the results in Figure 7 show that the heating is expected to be much improved as the surface and centre of the particle are raised above 630 °C.
Trials were then undertaken with 90 μm particles but this time using a 10% hydrogen-argon plasma. figure 8a gives the microstructure of a throughthickness cross-section of the coating deposited under these conditions.
The coating is porous and of low quality owing to insufficient flow of the particles on impact.

Figure 8a	Figure 8b
figure 8 - Cross-sections of coatings deposited using feedstock particles of diameter: (a) 90μm, (b) 60μm

Referring back to Figure 7, it is seen that the temperature is barely above the enamel fusion temperature and that substantial temperature gradients remain between the surface and centre of the enamel particle.
Further spray trials were carried out with the 10% hydrogen-argon plasma but now using much finer feedstock powder: D50 of 57 μm.
The microstructure of the resulting coating is given in Figure 8b. This shows a high-quality, dense coating of enamel.
The corresponding computational data are given in figure 6b. The implication from these results is that high particle temperatures and velocities together with fine feedstock powders are required to deposit sound enamel coatings under the conditions used.

The temperature profile of coatings during deposition
The temperatures in the coating and substrate during plasma scanning were computed using the model and are presented in figure 9.
This gives the temperature at the top surface of the coating, the interface and the back of the substrate.
The temperature of the surface of the coating is predicted to increase very rapidly in the first second, since the point P (figure 4) is close to the centre of the flame and the gas temperature and heat transfer coefficient are both very high.
The temperature in the substrate increases much more slowly.
Although the thickness of the substrate is five times the thickness of the coating, the temperature gradient in the coating is still much higher than that in the substrate.
This is caused by the fact that the heat transfer at this stage is controlled by conduction and the thermal conductivity of the steel substrate is much higher than that of the enamel.

figure 9 - The temperature distribution in the coating and substrate

At the fourth second, the temperature in the front of the coating is lower than that at the back of the substrate.
This is because the gas temperature of the coating surface at P is lower than that at the substrate and furthermore, the heat transfer coefficient is higher at the coating surface than that at the back of the substrate due to much higher gas velocity.
As scanning continues, the temperature difference between coating and the plasma gas at P is smaller and the rate of temperature rise becomes slower. After four complete scans, there is no significant change between the coating and substrate temperatures and thermal equilibrium is reached.
These results from the computational model show that during each plasma scan, the plasma torch produces a high thermal shock to the front surface of the coating and this generates a high temperature gradient in the coating. The effect is particularly marked for low-conductivity coating materials such as enamels.
In order to test the computer predictions, thermocouples were welded to the back of the steel substrate. The temperatures measured during spraying were found to be similar to and follow the same trends as those in the model.
The work in this paper focuses on how the temperature profiles depend on the plasma gas composition and the feedstock particle size. However, other process parameters will also have an influence.
For instance, the model also shows how the temperature in the coating depends on the scanning method and the component size. If the interval between each scan is extended or the area of the substrate is increased, the computations indicate that the temperatures of the coating and substrate will reduce by a substantial amount.

Conclusion
Computer models have been developed to simulate the temperature profiles of feedstock particles during their flight in the plasma and those of the coating and substrate during spraying.
The computational model for the in-flight particles shows that a process window for the formation of dense enamel coatings can be determined in terms of the plasma gas composition and the particle size.
The model for the heating of the enamel coating and substrate during scanning indicates that the plasma jet produces a thermal shock at the surface of the coating and a high temperature gradient through its thickness.
Experimental spraying trials confirmed the computational models and dense coatings were successfully produced.
Further work is planned with the aim of upscaling the laboratory process to practical operation.

Acknowledgements
The research was sponsored by the Engineering and Physical Sciences Research Council and the UK Department of Trade and Industry and carried out in collaboration with Corus plc and Escol Products Ltd. under the LINK Programme in Surface Engineering.
The authors would like to thank the above for their support.

References
1. R. McPherson, Thin Solid Films, 83 (1981) 297-310.
2. M. Vardelle, A. Vardelle P. Fauchais and M.I. Boulous, AICHE i., 29 (1983) 236-243.
3. D. R. Croft and D.G. Lilley, "Heat transfer calculations using finite difference equations". Applied Science Publishers Ltd., London, 1977, pp 1-87.
4. E. Bourdon, P. Fauchais and M. Boulos, Int. J. Heat Mass Transfer, 26 (1983) 567-582.
5. C. Bonet, Pure Appl. Chem. 52 (1980) 1707-1714.
6. X. Chen and E. Pfender, Plasma Chemistry and Plasma Process, 2 (1982) 185-212.
7. S. V. Joshi and R. Sivakumar, Mat. Sci. & Technol., 8 (1992) 481-488.
8. W. Tollmien, "Berchnung turbulenter ausbreitungvorga nge", NACA TM 1085 (1945).
9. Y. Bao, T. Zhang and D. T. Gawne, "Analysis of residual stress generated
during plasma spraying of glass coatings " in Thermal Spray: Meeting the Challenges of the 21st Century, Vol. 1, Edit. C. Coddet, ASM, Ohio, USA, pp575-580 (1998).
10. T. Zhang, D.T. Gawne, B. Liu and Y. Bao, "Computer simulation of shrouding on the profile of a plasma jet", Proc. Internatl Thermal Spray Conf., Singapore, 2001, Editor: C. Berndt, ASM International, Materials Park, Ohio.
11. A. Shimizu, "Impinging jet heat transfer with gaseous-solid suspension medium, "Advances in enhanced heat transfer", Symposium volume of the 18th National Heat Transfer Conference, San Diego: pp 155-160 (1979).