IEI, International Enamellers Institute

BONDING OF VITREOUS ENAMELS ONTO ALUMINIUM ALLOYS: THE HORIZON EXTENDS
Marc Leveaux - Pemco Brugge, Belgium
Nancy Crevits - Pemco Brugge, Belgium
Koen Lips - Pemco Brugge, Belgium
Bruno Schepers - Corus Aluminium, Belgium
Ann Aerts - Corus Aluminium, Belgium
Els Verboom - Corus Aluminium, Belgium

Abstract
This study relates the development of a bonding frit suitable for aluminium alloys within families 5000 and 6000, containing up to 2.5% Magnesium, and which not requires previous surface treatment (chromatation). The mechanicalproperties of these alloys are higher than the ones of conventional alloys 3003or 4006 today in use. The whole physical and/or chemical reactions, which occurred between enamel and substrate during firing, are described.
The possible changes enhanced by this new concept for the enamelling process are discussed.

Aluminium alloys and enamels: history
The first attempts of enamelling aluminium occurred at the beginning of the 20th century in the USA, to be more precise, after the First World War [1].
For all practical purposes, this was only the subject of laboratory experiments until about 1940, for example by Du Pont de Nemours [2], which patented frits,mill additives etc. As the aluminium metallurgy was still in its infancy, thedevelopment process was slow.
The aluminium alloys, when suitable for enamelling, were not consistentenough or didn't have the requested mechanical properties. The techniquewas not truly used on an industrial scale before the beginning of the 1950s. Enamels for aluminium were lead or lead-cadmium frits, which were not highresistant and were difficult to use [3].
From the alloy side, the most used were Magnesium containing, which obliged to use a specific surface treatment before enamelling: chromatation [4].
From that time enamel suppliers developed a complete range of frits,rinding additives and pigments designed for the aluminium-enamellingindustry. The frits, or Alumails®, were still lead frits but we were continuouslyimproving their quality. Cadmium slowly disappeared and lead level was constantly being reduced.
For the aluminium alloys, the metallurgy, every day more sophisticated butalso under control, brought the development of two new families, the Aluminium-Silicium-Iron and Aluminium-Manganese-Iron, respectively [6].
A few works concerning bonding mechanism between enamel and aluminium alloys showed that the main inhibitor for the reaction was Magnesium, which limitswere step by step, established [5-7-8].
The bonding reaction was, in that time, assumed to be first of all mechanical,proceeding by dissolution-corrosion of the alloy surface by the melted frit during firing [8-9].
If this is certainly true, we have to say that from that time, the sometimes-met lack of bonding, has to have other explanation, the wetting of the alloy by the frit not being the single parameter.
Towards the end of the 60s, our Group distinguished itself by perfecting and patenting the first lead-free, low-temperature systems, using Vanadium(pentoxide) as the flux replacing lead [10].
Still today, the whole enamelling sector continues to use the most common alloys 3003 or 4006 as well as the so-called "vanadium bearing frits", which chemistry naturally improved with the time.

Purpose
It's precisely among the main characteristics of the association enamel-alloy that we find the motivations of this study.
First for the metallurgist himself, whom the limits fixed for well-known elements like Magnesium (100 ppm), or lead (50 ppm) are a real handicap.
Also, small amounts of "impurities" included into the alloys through the process (ex. use of waste alloy) or because it could modify any mechanical or physical behaviour, can be assumed as "dangerous". At least, a lot of alloys today in use, with mechanical properties much higher than the ones of the conventional alloys, could bring new developments in the field.
Second for the enameller, going forward with a cheaper, more safe, more flexible enamelling process, having then the possibility to improve his production.
So, by increasing the general knowledge and by better understanding the bonding mechanism between vitreous enamel and aluminium alloy, we could:

Definitely avoid the use of previous surface treatment, although working with Mg containing alloys.

Make the coating thickness' range larger, without any spontaneous chipping off.

Develop new enamels, with improved quantities of Vanadium or Titanium oxides, which are expensive compounds.

Bring on the market, more resistant enamels, new effects (white, matt…).

Enamel and bonding: all is not aware
General case
Bonding of enamels onto aluminium alloys is a mechanism today not fully understood, in spite of the fact that it was studied for many years [7-8-9].
This phenomenon is still now not easy to investigate, even with sophisticated methods of analises. Mainly because the frits, containing very high amounts of alkaline oxides, make the investigations near to the interface enamel-alloy unprecise and difficult.
Several methods have been proposed, in order to improve or to get the adherence of vitreous enamel layers onto aluminium alloy surfaces, specifically in the case of Mg alloys susceptible for spalling:

Chromatation of the aluminium alloy surface, the most used technique [11],

Avoiding migration of Magnesium to the surface, by tying up the whole amount in the alloy as particular phases likes Mg2Si. This is only possible

when the amount of Silicium passes over a certain limit [15-16],

The use of bonding enamel, loaded with iron oxide, and applied as a first thin layer onto the alloy surface, then recovered by the covercoat [17].

Through all these solutions as well as through the own experience in the field, this supposes that this bonding mechanism proceeds from successive and superposed steps, which chronology looks like for vitreous enamels applied onto steel surface.
These steps being in their characters, slightly different. The whole mechanism has been described on figure 1.
So the first step consists in an alkaline superficial dissolution of the substrate, achieved in two successive stages. First, during surface pre-treatment, which consists in degreasing and pickling. The medium has to be an alkaline silicates free solution, to avoid passivation of the surface. The second stage occurs when applying the enamel slip, which pH is varying between 11.5 and 12, onto the surface. There's so the possibility for the enamel to find mechanical anchorage on a rough surface. During applying and before drying, some hydroxides will form on the interface enamel-alloy.

The second step, between room temperature and softening point of the frit, consists in decomposition of the different hydroxides, overlapped to a superficial oxidation of the aluminium alloy surface.
Limited here to a few compounds, we have:

The third step, between softening point Te and firing temperature Tf, today generally between 530 °C and 560 °C consists in a juxtaposition of oxidation of the hydroxides and the alloying elements, and corrosive dissolution of the metallic alloy by the melted glass.
So in fact there's a fourth step, the vitreous enamel, basic solvent, being able to oxidise the alloying elements, being reductive, previously dissolved, to build a bonding interface between vitreous enamel and alloy.
Is there, like for vitreous enamel for steel, a reductive mechanism for a few oxides and built in of separate phases onto the surface? A self-diffusion mechanism from the enamel into the aluminium alloy? It may be.

Magnesium containing alloys
For these particular alloys, a strong diffusion of the not precipitated magnesium towards the interface takes place during firing [17]. This element will oxidise stronger and faster than the others alloying elements, according to the reaction:

Mg+1/2 O₂ -> MgO

This self-diffusive process being continuous, this reaction then stops progressively the dissolution of the alloy surface by the melted enamel.
The oxidation of this element enhances a big change also in the lattice
parameter, and involves stresses in the interface enamel-alloy, which bring progressive chipping off. We can observe first local chipping, which looks like fish scales into vitreous enamels for steel. Depending on the amount of magnesium, the phenomenon extends up to complete chipping off.
These alloys stay suitable for enamelling only thanks to a previous surface treatment of chromatation. The thin Chromate layer oxidises the diffused magnesium and forms a barrier, locking further interaction of the same with the enamel.

Experiences: practical details
Aluminium alloys
Common conventional alloys have been chosen, which magnesium content is varying between 0 and 2.7%. All references have been produced on an industrial scale. Details concerning compositions are shown in Table 1.
The pre-treatment conditions have been established, prior to enamelling, for each alloy. We have worked by immersion, the time being adjusted in order to get a weight loss on pickling varying between 7 and 9 g/m².

Degreasing 60 °C - Almeco 18 (Henkel)40g/l
Rinsing hot 40 °C
Pickling 60 °C- Almeco 57 30 g/l
Rinsing hot 40 °C
Rinsing cold
Neutralisation HNO₃ 10%
Rinsing cold (demineralised water)

Average roughness (Ra) was measured on each alloy, both on lamination direction and transversal. Measured values obtained: between 0.3 and 0.7 μm.

Enamelling
The enamel slip was sprayed onto the surface, in one or two layers. The total applied thickness (measured after firing) being 60-70 μm, within 10 μm (average) for the first layer, in the case of two coats-one fire systems. We used, as cover coat, two different systems: white and transparent enamels.

Milling fineness 0.8 - 1 Bayer cone 25000, density 1.7.
For each trial, the pH of the slip was adjusted around 11.5.

Firing
Realised in a box kiln, muffled, equipped with forced air. The temperature
was varying between 540 °C and 580 °C. Firing time was fixed at 6 minutes.

Tests
To check the bond, we used two different procedures, today in use in the field of aluminium enamelling [18-19]:

Impact test using the apparatus described in the Standart EN10209, with 4 different impacts: 10, 20, 30, 40 cm height. We so get an overview of the brittleness of the interface. We keep in mind three different impact morphologies (figure 2):

Without any reaction or with scarce reaction, (quite) complete chipping off.

From bad to medium bond, with circular impact in form of concentrical circles. The larger is the impact, the worst is the bond. The closer are the circles one to the other, the better is the bond.

Good bond, when the impact looks like a star, developed from the centre, quite without chipping. In this case, we give an indice to the bond, being the maximal height for which we have this result (example 2 for 20 cm height). For 10 cm more, the impact is having a circular design.

Test by immersion in an antimony chloride solution according to ISO Standard 13805 but performed during 72 hours. We so get an evaluation of the intensity of the reaction between enamel and alloy (figure 3).

Naturally this supposes that the enamel in use or the different phases which could form during firing are acid resistant. Our own experience showed us that it was interesting to perform this test more than the recommended 24 hours.
After a one-day period, only the non-bonded systems are clearly identified. For the systems exhibiting a medium reactivity onto the alloy, a two days period can be necessary. After three days, both the systems 100% safe and the onesshowing possible slight problems are clearly identified.
In this study we considered as "good" all the enamels for which no change in the indentation aspect was observed over a 72 hours period. The "medium ones" giving chipping off maximum 1 millimetre large, or just on corners, and the "bad ones" all the samples with chipping off over 2 - 3 mm or more.
Systematic analyses were performed on the enamels and interfaces enamelalloy after firing, by X-ray diffraction and micro-analyses, to determine diffusion profiles and phases formed during firing.

Concept of the study: used enamels
We believe we could find, like for vitreous enamels applied onto steel surface, near to the interface enamel alloy, both, reactions of oxydo-reduction and eventually formation of phases, recristalised or not. As a matter of fact, all the solutions today in use [16], consist in increasing the oxidation power of the enamel toward the alloy. If we are able to develop a new system able to get first full oxidation of the diffused magnesium, and afterwards dissolution of at least a part of the oxide formed near to the interface, we will increase or get bonding. That for magnesium contents even higher than the ones accepted today. In order to work on that particular idea, we introduced several oxides in specific frits during smelting (Table 2 and 3). The addition during milling pure frit was also considered, when possible. Depending on the oxide tested, the amount was varying from 0 to even 10%.

Results
Aluminium alloys with low Mg content
Of the low Mg alloys, which are used today, we can say, out of our test results and our PEMCO experience on this matter, the following (Table 4):

It is sometimes possible to obtain a good result with the Antimony chloride (SbCl₃) test and only a moderate result with the impact test. The other way can also happen in the case that the enamel frit is not acid resistant.

The results of the impact test have to be taken with care, especially the indices. It is not possible to compare rigorously one alloy with another because there is a direct influence of the sheet thickness and the mechanical properties of the alloy on the impact test result.

For a given frit, the adherence will diminish if one changes from alloy 1050 to 3003, to 4006 and at last to 4917. This can only be observed with very specific enamel frits, developed in order to produce limited adherence on pure aluminium. But, so we can make the following statement:

This statement is in direct correlation with the viscosity and the surface tension of the frit. The frits with a lack of fusibility or with a very high surface tension have a lower bond on the alloys 4006 or 4917 (case of frit F1 compared with F3).

Confirming in a certain way the tendencies above, the test results prove that vanadium (pentoxide) is without any doubt a very important element in the bonding phenomenon, in a way that the higher its proportion in an enamel system, the better is the result.
This element is known to reinforce the wetting of the enamel onto the substrate (coefficient of Dietzel: -6.1) and to decrease the viscosity of the frit.
Other tested promoters of the adherence:

Iron oxide Fe₂O₃. We confirm the work described in the Tefal patent [16].

Copper oxide. These two oxides being added at the melting and the milling.

X-ray diffraction measurements on the interface enamel metal after elimination of the enamel layer showed the presence of vanadium monoxide (VO). This proves that there are oxydo-reduction reactions going on at the interface.
We have not been able, by microanalyses on sections, to put in evidence, the specific diffusion of components of the enamel towards the interface.
Only an enrichment of Aluminium and Silicium in the enamel near the interface has been observed.

Aluminium alloys with Mg content: PEMCO method
We have observed, by the application of a single layer of Alumail® on magnesium containing alloys:

Chipping off at the border, partial or complete, starting immediately on cooling down,

Microanalyses on sections have proved a migration of magnesium towards the interface enamel metal. The concentration of magnesium at the interface becomes very high and also diffusion, together with aluminium, into the enamel layer takes place (Table 5).

By the application of a 2 coat / 1 fire system, with a 10 μm thick ground coat being an enamel frit loaded with oxides in the melting, or added to the mill (Table 6), we have been able to obtain adherence on alloys containing till 2.45% of Magnesium. The tests confirm the work done before with iron oxide [16]; the use of copper oxide permits to go further by obtaining adherence on alloy 5555.
In any case, having both positive results, the addition in the melt enhances a better surface aspect compared to the addition at the milling, which creates a lot of aesthetic defects.
Concerning the cover coat, the tests have been performed with both white and transparent cover coats; the results being almost identical. The addition of titanium oxide in the cover coat seems to reinforce the adherence of the white configuration.
X-ray diffraction measurements of the interface after removing the enamel layer of the pieces, also revealed the presence of vanadium monoxide, and this for all the frit compositions with iron or copper oxide.

The impact test marks reveal a grey reddish interface colour for the copper containing systems. Microanalyses (EDX) on cross-sections shows (figure 4):

a diffusion of magnesium towards the interface where it enriches, and a further diffusion into the enamel layer.

A diffusion of copper towards the interface where it is found in big quantities.

Interpretation
There are indeed during firing several reactions of diffusion and oxydoreduction for a few species, followed by cristallisation of Vanadium monoxide onto the interface vitreous enamel-alloy. This last fact occurs both for low containing as well as for high magnesium containing alloys.
For the last ones, the iron oxide or the copper oxide, both associated to the vanadium pentoxide, act as bonding elements, means oxidise chemical elements onto the aluminium alloy surface or the same previously dissolved by the melted glass.
If we detected, using X-ray diffraction, cristalisation of vanadium monoxide, any kind of phase containing iron, neither copper, was analysed. Scanning just showed self-diffusion of the element copper from the enamel near to the interface. For vanadium oxide, only the slight deviation of the Ray spectrum, compared to stoechiometry, suggests that we could have slight reduction of the same phase, and either contamination by copper itself.
If we look at the variation of the free Gibbs energy with the temperature for all oxides present or able to form during firing, through oxidation or oxidoreduction reactions (figure 5) it seems that:

metallic Magnesium is in fact a strong reductive element for the enamel,

For vitreous enamel, it becomes then clear that the higher is its oxidising potential, the more intensive could be the global oxidation of the alloying elements. So, any vitreous enamel, containing sufficient amounts of oxides able to be reduced during firing, could be bonded onto high magnesium containing alloys. It's in fact what we saw for all systems containing vanadium, iron or copper oxides. For iron and vanadium oxides, the process is broken when the oxidising potential is limited, and there's no diffusion of the same during firing. On the other hand, copper, continuously diffusing toward the interface enamel alloy can ensure this oxidation reaction.
This was observed on all selected alloys, except on the 5754, for which it seems not to be strong enough.
Except if further investigations show active participation of the element(s) copper and/or iron to the bonding mechanism, all seems to be that these two elements act as oxygen generators. On vanadium, itself further reduced, either on self-alloying elements. We could suppose, as first step that:

Which could explain the reddish-grey colour observed onto the interface
(Cu2O is red). The free oxygen being able to oxidise later on, vanadium,
magnesium or aluminium:
And then

The oxide VO then crystallising on the interface and all the free oxygen being
combined according to the reactions:

Characteristics of the new alloys
The essential interest of the new alloys enamelled with this method is their mechanical properties, which are superior to those of the conventional alloys.
Comparing the yield strength (Rp 0.2) and the tensile strength (Rm) of alloy 3003 with the medium strength alloys 5005 or 3004 (soft condition), there is a substantial rise of 25 -110% in Rp 0.2 res. 10-60% in Rm.
This is an important increase, especially if one knows that the characteristics do almost not change during the enamel firing cycle, which has the same influence as a soft annealing.
Concerning the heat treatable alloys, the alloy 6082 has almost the same mechanical properties as the alloy 4006 before enamelling.
So, there is no difference during the deformation. But, after enamelling, the mechanical properties (both yield strength and tensile strength) of the alloy 6082 are almost 25% higher.

Conclusion: perspectives
The investigation has proved that the adherence process between enamel and aluminium alloys containing magnesium is very analogous to the one observed on steel. It is based on diffusion phenomena followed by oxydoreduction reactions between alloy and enamel.
The intensity of these phenomena is before all conditioned by the wetting of the metal by the enamel.
The study has made it possible to create an adherence enamel (patent EP 099 123 342.0), in association with a conventional cover coat, with which alloys with a magnesium content till 2.5% can be enamelled.
Thanks to this new adherence frit of PEMCO, use in mono layer for dark colours or in a 2 coat / 1 fire configuration for white and pastel colours, it is now possible to use enamelling magnesium containing alloys, like 6061, 6082, 3004, 5005, 5555, without chromatation pre-treatment Reproducibility tests have started on specimens taken from industrial production of these alloys covering different months. They confirm the results obtained in the lab, but need to be continued.
This new method opens new technical perspectives, due to the fact that the mechanical properties of the magnesium containing aluminium alloys are superior to the ones of the alloys that are used nowadays: the shapes of the aluminium ware can be changed, the application of enamelled aluminium can be enlarged.
The PEMCO system also permits economical benefits for the actual products, by reducing the thickness of the substrate without decreasing the mechanical resistance of the enamelled articles as well as the long life at high temperature.
It also enables the aluminium foundry to enlarge the tolerances of magnesium in the melt.

References
1. H. Piera, galvano-organo-traitements de surface mars 1992,624.
2. US patent 2,467,114.
3. Dietzel, Die Emaillierung.
4. PEI Bulletin n ° 802.
5. H. KYRI, handbook for Bayer enamels.
6. M. Deleuze, D Marchive Revue de l'aluminium, juin 1981.
7. R. Schulze, H ;W ; Hennicke , vdfa Mitteilungen, Band 22/1974.
8. A.L. Guleger, Proceeding of the P.E.I, technical forum, vol31,1969.
9. Patent DE2,119,777.
10. L. Hiller, P. Hellmold et all, VDEfa Mitteilungen, 1994.
11. D.W.A. Verboom,VDEfa mitteilungen, 1995.
12. Brevet USA 3,222,266.
13. Brevet EP0,611,834,A1.
14. Brevet EP0,648,863 B1.
15. Brevet GB-A-840,469.
16. Brevet USA 2,932,584.
17. Brevet EP 0,686,609 A1.
18. CISP working group n° 3, 10-1997.
19. Smalto Porcellanato 2/2000.