DEVELOPMENT OF A NEW TEST METHOD FOR FISHSCALE DEFECT PREDICTION IN STEELS FOR ENAMELLING
Renzo Valentini  Dipartimento di Ing. Chimica
Chimica Industriale, Università di Pisa, Italy
Walter Sabba  ILVA, Gruppo Riva
D. Mattia De Micheli, Renzo Valleggi  Scienzia Machinale s.r.l., Italy
Abstract
In this work a measurement test method for fishscale defect probability prediction in steels for enamelling is described.
This test method utilises an electrochemical permeation system and a hot wire sensor that measures the hydrogen flux which passes through the surface of an enamelling sheet. The determination of the material diffusion coefficient, together with the
utilisation of a mathematical model, allows the evaluation of the enamelling attitude of IF steels too.
Introduction
The fishscale defect is a problem, which can be shown in steel products for enamelling. Fishscaling is caused by hydrogen gas rising up pressure at the interface between enamel and steel, causing the enamel rupture.
The market introduction of new Interstitial Free steels together with the necessity to guarantee a better quality control has generated the need for a new prediction method of the fishscale defect which can overcome the limits of the current available instruments.
Currently there are two different modalities for evaluating the fishscale susceptibility in steel for enamelling: an indirect modality and a direct one.
Three different commercial instruments are available:
Ströhlein, based on an indirect measurement, evaluates the volume variation in a capillary due to the pressure of the hydrogen that cross the sample;
Hyperm, based on an indirect measurement, evaluates the current variation due to the permeated hydrogen.
Enamelling test (Smalto Test), based on a direct measurement, evaluates the behaviour of a specimen enamelled in humidity condition critical for the defect formation.
The evaluation of the results obtained by these methods pointed out a great dispersion of the data. This scattering was due to:
loss of precision for the lack of reproducibility of the operative parameters;
different sensibility of the different instrument used.
The system adopted in this research has brought to the development of a new approach to the problem with principal objectives:
no limitations in terms of chemical composition of the steel in evaluation;
no limitations in terms of thermomechanical treatments;
ranging of the class of quality of the steels, not only an "acceptable  not acceptable" response.
It is well known that hydrogen diffusion in steel is influenced by the presence of traps. The word "trap" indicates a site in the microstructure (interface matrixprecipitates or second phases, various kinds of lattice defects) which can interact with hydrogen.
Traps are normally classified as "strong" or "irreversible", when the probability for hydrogen to be released is extremely low, or "weak" or "reversible" when the probability of hydrogen release is higher.
Apart from trapped hydrogen, after enamelling a certain amount of hydrogen remains free into the ferrite matrix, and is the first responsible of fishscaling.
The Devanathan apparatus is well known to be as the more accurate, as laboratory system, to control the fishscale susceptibility. Nevertheless, one of the limits of the use of Devanathan permeation test, is the long time required to obtain the results (about 24 hours), since it is based on a double permeation and for each one it is necessary to reach the hydrogen stationary flow.
The idea which was the basis of this new method to predict fishscale susceptibility was to eliminate the time necessary to reach the stationary flow in the semicell of measure of the Devanathan apparatus, using a solid sensor which could be immediately ready to capture the hydrogen flux without secondary effects due to oxidisation of the surface.
The time of the measure will be reduced to only few minutes for thickness of 0.40.8 mm specimen.
The permeation curve contains information regarding both reversible and irreversible traps.
So from a regression of the curve it is possible to individuate the number of the reversible and irreversible traps contained in the specimen individuating the ability of the steel to avoid the fishscale defect.
This system is independent from the steel composition and it will be possible to catalogue the specimen respect to critical values.
This kind of approach is reliable and can be used at industrial scale because as sensing element for Hydrogen flux has been adopted a selective sensor based on Hot Wire catalytic technology, derived from the application in the domestic field. This kind of sensors are able to individuate small quantities of hydrogen and continue to maintain their properties during the time.
These characteristics are fundamental for industrial application, while in the recent past different sensors based on palladium membranes presented a lot of problems despite the very selective behaviour respect to Hydrogen.
Experimental
The experimental apparatus developed (figure 1) is composed by an electrolytic semicell, where hydrogen is developed through an electrochemical reaction in a diluted sulphuric acid solution, which utilises the sample as a cathod, applying an adequate cathodic current.

figure 1  View of the hydrogen permeation measurement apparatus 
Hydrogen passes through the sample and is surveyed by the flux sensor,
based on the hot wire technology.
The flux values are therefore surveyed by an electronic unit which processes the signal and transmits it to a PC which determines the hydrogen flow curve.
This curve is compared with a mathematical model that allows to find out the reversible and irreversible traps present in the steel matrix which the consequent classification of the steel (figure 2).
The system is indipendent from the steel chemical composition and will be then possible to classify the samples according to the critical values.
This kind of approach is reliable and may be utilized at industrial level since as sensible element for the hydrogen flux a selective sensor based on catalytic hot wire technology, derived from applications in other fields, was adopted.

figure 2  Permeation equipment 
This type of sensors are able to identify small hydrogen contents and keep their properties unchanged with the time. These characteristics are basic for industrial applications, while in the recent past sensors based on palladium membrane suffered various problems notwithstanding their highly selective behaviour towards hydrogen.
The measurements performed with this apparatus and with the Devanthan instrument showed a good accordance concerning the diffusion coefficient D, measured on different materials (figure 3).

figure 3  Permeation measured with a Devanathan instrument compared
with permeation measured with the new apparatus 
It is well known that each steel class has D values, as a matter of typology and number of hydrogen traps, suitable to guarantee a good enamelling attitude for which concerns the fishscaling resistance.
The D value may be automatically calcuated through the time lag method after the permeation (figure 4) but the determination of the entrapment system complex, evaluated by a physicalmathematical model, allows to obtain information scietifically and techinically more reliable in order to determinate the material behaviour during the enamelling process.

figure 4  D calculated with the time lag method 
The fishscale effect is well simulated also performing an electrochemical
permeation on a steel enamelled on one side, in order to artificially cause the
defect and then evaluate the enamelled product resistance in terms both of
hydrogen permeability and steel enamel adherence (figure 5). This test
performed with the new apparatus is very simple because the sensor is based
on the gaseous hydrogen measurement and then doesn't chemically or
electrochemically interact with the enamelled surface.
The hydrogen flux measurement reliability is guaranteed by the possibility of
calibrating the apparatus with pure gaseous hydrogen feed through
Hamilton^{®} syringes.

figure 5  Example of fishscale defect performed
by electrochemical process 
The mathematical model
The mathematical relationships which rule, the model of the general
hydrogen permeation in metals represent in fact a non linear second order
system composed by three differential equations with six parameters.
Five more equations, between outline and initial conditions, complete the
diffusion problem with the reversible ad irreversible entrapment processes in a
uniform metal sheet with parallel surfaces, at a constant hydrogen
concentration on the inlet side and zero on the outlet side.
Assuming an homogeneous entrapment locations distribution through the
sheet thickness, the well known equations of the reversible entrapment model,
extended to the irreversible entrapment case, are as follows:
Where C is the diffusible hydrogen concentration, n_{r}/n_{i} the reversible/irreversible
traps, D_{L} the diffusion coefficient in the perfect crystal lattice, k_{r}(k_{i}) the probability of
hydrogen capture by the reversible/irreversible traps, p the probability of hydrogen
release by the reversible traps (for the irreversible ones p=0 by definition, at room
temperature).
Normally, in case of monodimensional interest a form of expression with
nondimensional variables is utilised, which is easily adapted for the solution
with numeric calculation of the finite differences.
To utilise this mathematical model in the permeation results interpretation it is
necessary to find a mathematical model which can be supported by low
performace computers, in order to equip a portable instrument.
Then a CrankNicholson implicit method was applied, which allows the direct
use of the model on the same computer that performs the test.
Examples of correlation and entrapment parameters calcuation are shown
in figure 6. Obviously a large number of high binding energy traps guarantee
an high fishscale resistance.

figure 6  Experimental curves and model fitting 
Conclusion
The approach proposed can mix together the reliability of the Devanathan
test together with the fast response given by a new Hydrogen sensor in order
to predict the fishscale susceptibility of an enamelling steel.
The model which analyses the hydrogen permeation can individuate the
different mechanisms which can help to avoid the fishscale defect both in
traditional and in new I.F. steels.
