Title of Invention

METHOD TO MEASURE DEGREE AND HOMOGENEITY OF ALUMINA CALCINATION

Abstract The present invention relates to a method to measure the degree and homogeneity of calcination of alumina characterized in that said method is carried out with an image analysis device equipped having a sensitive camera in a spectral analysis window corresponding to a wavelength range equal to or in the vicinity o visible light and in that it comprises the following steps: vicinity of a) mixing the alumina to be analyzed with a liquid which has a refractive index that is, in said wavelength range, between the refractive index of a lightly calcined alumina and the refractive index of a strongly calcined alumina; calcined b) preparation of a slide for observation of said mixture in said image analysis device, said mixture being illuminated by stable polychromatic radiation compatible with said spectral analysis window; iation c) reception of the image by the camera and processing of the signal resulting the definition of an image composed of a given number of pixels with in colorimetric components; three d) statistical processing of said pixels using their colorimetric components used to give the calcination degree and the homogeneity of the calcination.
Full Text METHOD TO MEASURE DEGREE AND HOMOGENEITY OF ALUMINA
CALCINATION
Field of the invention
The invention relates to an in-process monitoring method of the degree and homogeneity of alumina calcination.
The majority of technical aluminas involve, in their production process, a calcination step. Said step is determined to define the standard properties of the powder to be used in numerous applications (polishing, refractory concrete, fine ceramics, etc.). It is known that the calcination degree is well correlated with the surface area developed by the calcined alumina and, for the aluminas of interest to us, said surface area, measured using the BET method is typically between 0.5 and 5 m2/g.
However, numerous applications and some target applications (ceramics, electronics, etc.) show the need for a more detailed analysis of these products, an analysis particularly giving information on the calcination homogeneity of the product; it is -clear that two aluminas with a BET surface area equal to 1 m2/g for example but one of which is homogeneous and the other composed of a mixture of two aluminas at 0,5 and 1.5 mVg will not have the same applicative properties.
State of the related art

Through US 3,941,874, a method used to measure the calcination degree of alumina by measuring the loss by weight of said alumina is known.
To estimate the calcination degree achieved, it is also possible to grind the alumina to the crystallite size and perform a granulometric analysis but this method is rarely or never used in practice since it is long and incompatible with the implementation of a method.
It is also possible to measure said calcination degree indirectly from the surface area developed by the alumina using a conventional method, such as the BET method. This method consists of measuring the quantity of nitrogen adsorbed over the entire surface area developed by the alumina. Said quantity increases with said surface area and the porosity of the grain (however, provided that the pores are open). It is used in production but it is long and does not give information on the heterogeneity of the calcination degree of the product.
Of the known analytical methods, image analysis may be used but it gives essentially dimensional information {grain size, shape factor) which is not correlated with the desired information (degree and homogeneity of calcination).
Problem statement
The applicant tried to incorporate in the implementation of the alumina manufacturing method means to measure the degree and homogeneity of calcination.

such means must be more rapid than the BET method in evaluating the average, calcination degree and give information of the heterogeneity of the calcination degree of the alumina produced. Subject" of the invention
The invention, firstly relates to a method to measure the degree and homogeneity of calcination of alumina comprising the use of an image analysis device equipped with a sensitive camera in a spectral analysis window corresponding to a wavelength range equal to or in the vicinity of visible light and comprising the following steps:
a) mixture of the alumina under analysis in a
liquid wherein thfe refractive index is, in said
wavelength range, between the refractive index of a
lightly calcined alumina and the refractive index of a
strongly calcined alumina;
b) preparation of a slide for observation of said
mixture in said image analysis device, said mixture
being illuminated by stable polychromatic radiation
compatible with said spectral analysis window;
c) reception of the image by the camera and processing of the signal resulting in the definition of an image composed of a given number of p"ixels with three colorimetric components;
d) statistical processing of said pixels using their colorimetric components and used to give the calcination degree and the homogeneity of the calcination.
Indeed, the applicant started with the assumption that, by choosing to mix the alumina powder in a liquid

with a mean refractive index between the extreme indices of a strongly calcined alumina and a lightly-calcined alumina, it set up the best conditions to accentuate the differences in responses to a given lighting, typically visible light, and that theses differences in responses could be processed digitally and statistically to characterise the calcination degree of the alumina observed and the homogeneity of the calcination carried out on the alumina.
It obtained very encouraging results which are explained retrospectively as follows:
- The refractive index of a substance varies as a function of the wavelength of the radiation passing through it. In the visible light range, this variation is much more significant for liquids than for solids. When a solid is immersed in a liquid with the same refractive index, no light dispersion or absorption takes place. However, if the refractive indices are different, a dispersion/absorption phenomenon is observed.
- However, it is known that the refractive index of an alumina varies according to its degree of calcination. If an alumina immersed in a liquid is observed and if, for a given wavelength, the refractive index of the alumina and the liquid is not the same, a dispersion/absorption phenomenon takes place which is conveyed by a varying light intensity response.
- When a solid, in this case alumina, is dispersed into a very large number of particles, the multiple facets are oriented in a random fashion with reference to the incident ray of light. Therefore, it is

necessary to illuminate a large number of particles to obtain a significant statistical effect on the reflected (or transmitted) image.
- If a polychromatic light - in the visible range for example - is used for illumination, the intensity response will differ according to the wavelength. The processing of the signal received by the camera makes it possible to obtain a coloured image that can be broken down into a multitude of pixels characterised by three colour components (according to the RGB system recommended in 1931 by Compagnie Internationale de 1"Eclairage CIE, or according to the Lab system (or Hunt system), or according to any three-dimensional colorimetric space reference, etc.).
The applicant observed that the colour of each product, expressed in this way pixel by pixel in a three-dimensional colorimetric space, is characteristics of the product observed and that, by transferring the desired data to a histogram, it is possible to compare said histogram to "stand" histograms of known aluminas wherein the calcination degree can also be measured using the BET process.
In this approach, two steps make use of polychromatic radiation but the distinction that can be-made between the spectral analysis window of the radiation illuminating the mixture and the visible spectrum used to describe the result of the processing of the signal received by the camera is clear.
The spectral analysis window may indeed correspond to the visible light spectrum but it may also correspond to a more restricted, continuous or

discontinuous, wavelength range, or a range extended to the infra-red and ultra-violet ranges. The important factor is that the spectral analysis window is chosen such that the refractive indices of aluminas of varying degrees of calcination remain stable and that the refractive index of the liquid varies in this range, said range being sufficiently broad so that the refractive index of the liquid is between the refractive index lightly calcined alumina and the refractive index of a strongly calcined alumina. In addition, the order of magnitude of the wavelengths must in particular be less than that of the particle size, which makes it necessary to define a spectral analysis window in a wavelength range typically between 1000 A and 10,000 A, visible light (#4000-7000 A) being for this reason well suited and enabling easy control of the measurement method. To facilitate the presentation, we will associate the spectral analysis window below with a visible light window and we will use the terms generally used for the observation of an object illuminated with visible light.
To prepare the observation, the alumina powder under -analysis is mixed in a liquid of a given refractive index. As, in visible light, the refractive index of an under-calcined alumina is in the vicinity of 1.70 and the refractive index of an alumina ex is in the vicinity of 1.76, a liquid with a refractive index in the vicinity of 1.73 is preferentially chosen. Such liquids, such "as methylene iodide, are commercially available.

The mixture is deposited for example on a glass slide such as those used in optical microscopy. The slide is then placed in the observation field of the camera by means of a magnification system such that each particle can be represented by a sufficient number of pixels. Typically, with an image of 640*480 pixels, it is preferable to choose a magnification such that the image contains less than 1000 particles. However, said particles must be sufficient in number, since it is necessary to obtain a significant statistical effect on the image. The magnification system of an image analysis device used within the scope of the invention must make it possible to analyse an image comprising at least 50 particles.
The mixture is illuminated by a stable polychromatic radiation compatible with the spectral analysis window of the camera, i.e. the wavelength range of the reflected or transmitted radiation is included in the spectral analysis window of the camera.
The signal detected by the camera is processed such that an image defined by a given number of pixels with three colorimetric components is obtained. The image may contain a colour corresponding to the radiation detected by the camera but it may also have codified colours arbitrarily characterising the radiations detected - in this case, the term "false colours" is used. However, it is preferable to be able to use the entire range of the colorimetric space to define each pixel of the image. ■
The pixels, characterised by the three components or a colorimetric space, are then processed. Firstly,

it is necessary to be certain that they correspond to particles since they are not contiguous in the mixture prepared on the slide and do not occupy the entire surface area of the image. Sorting may be carried out easily since the part of the image corresponding to background, i.e. occupied only by the liquid, shows a high light intensity. Therefore, it is possible to remove all the pixels wherein the three components are greater than a given value.
The remaining pixels are then represented in the colorimetric space. As the calcined alumina observed does not correspond to a perfectly homogeneous phase and as each pixel may in fact correspond to a plurality of photons of different energy, therefore to a mixture of colours, the representation in the colorimetric space of the pixels of the image correspond to a scatter of points varying in range. The applicant observed that, by choosing a liquid with a suitable diffractive index for the chosen spectral analysis window, it was possible to differentiate between aluminas using the scatters of points representing said aluminas in the colorimetric space, the location of said scatters being related to their calcination degree and the range of said scatters being related to their degree of homogeneity.
Using these representations in a three-component space, several types of statistical processing are possible. The purpose of the processing described here is to obtain a representation that is easier to interpret. It consists of situating the images under analysis with reference to the representative scatters

of two known aluminas which are calcined at most different possible calcination degrees. For example, an a alumina and an under-calcined alumina, preferentially the least calcined but the most homogeneous possible, are chosen.
In principle, the two scatters corresponding to the alpha alumina and to the under-calcined alumina, respectively, are distinct from each other. Otherwise, it is preferable to choose another liquid and/or another spectral analysis window.
If the two scatters are distinct, it is possible to define a sub-space which "passes" through these scatters (an axis or a plane which passes through their centres of gravity, for example) and wherein the estimation of the calcination degree may be carried out using the "distances" between points, highlighted in said sub-space. In this way, it is possible to project the scatter of points of the alumina under analysis onto said sub-space and estimate the distance existing between said projection and that of the a alumina and/or that existing between said projection and that of the under-calcined alumina.
In this way, by illuminating the alumina + liquid mixture with visible white light, it is observed that a particle of a alumina gives a blue image and an under-calcined alumina gives a brown image of varying darkness. If the pixels were defined in the RGB reference system, it is attempted to classify them according to the "B-R" (blue less red) component. 3y classifying the number of pixels with a rising B-R component, a histogram or characteristic spectrum is

obtained: the higher the calcination degree of the alumina, the more its spectrum is positioned to the right and, conversely, the lower the calcination degree of an alumina, the more its spectrum is located to the left.
The mean of the histogram obtained is characteristic of the calcination degree of the alumina analysed.
The standard deviation of the histogram obtained in this way is characteristic of the homogeneity of the calcination of the alumina analysed.
Secondly, the invention relates to the use of the measurement method described above to monitor the degree and . homogeneity of calcination of alumina produced continuously in a ring furnace. Said measurement method is much more rapid that the BET method in determining the calcination degree of the alumina and makes it possible to react more rapidly to a deviation in the parameters of the furnace. This is particularly advantageous for a ring furnace working in continuous mode and producing several tonnes of alumina an hour. Finally, using this method, which is the only one known to date that makes it possible to determine the homogeneity of calcination, it is possible to correct the adverse effects of dust recycling which takes place in an untimely and poorly controlled fashion in ring furnaces and "contaminates" the alumina produced with an alumina of different grain size.
Using an alumina sample taken from the furnace, four slides are prepared. Each slide is observed at three different points using the claimed measurement

method. Results that are reproducible to within 5% are
obtained.

Figures
Figure 1 illustrates the acquisition chain used within the scope of the example described below.
Figure 2 represents three response spectra on the axis (B-R+100) relating to three aluminas (a, b, c) having different calcination degrees.
Figure 3 shows the bijective relationship existing between the calcination index ic measured using the method according to the invention described in the example and the BET surface area expressed in m^/g.
Figure 4 gives the relationship between the BET surface area and the crystallite size TC expressed in μm.
Figure 5 shows this relationship in the range most specifically of interest to ceramists.
Figure 6 shows the spectra obtained by mixing two aluminas of different degrees of calcination.
Figure 7 shows, as a function of the percentage of one of the ingredients, the variation of the homogeneity index in a mixture of two aluminas, one strongly calcined and the other under-calcined.
Embodiment of the invention
Acquisition chain (figure 1)
The acquisition chain, represented in figure 1, is as follows:
1) The slide A on which the alumina / index liquid mixture is deposited must be prepared carefully: it is necessary to prevent the appearance of bubbles and obtain the most homogeneous particle distribution possible, said particles not being attached to each

other. We selected a liquid with an index of 1.73. For industrial calcination monitoring in ring furnaces, this preparation may be automated.
2) Alumina grains being micronic, the use of a
binocular microscope B with variable lenses according
to the alumina is recommended. The aim is to obtain a
sufficient quantity of grains for a valid statistical
analysis (thus avoiding having to increase the number
of tests).
3) The slides are illuminated with visible
spectrum light. The setting and stability of the
lighting are important since they affect the
colorimetric components of the pixels of the image
obtained by the camera C. It is particularly necessary
to avoid saturate light intensity signals.
In this example, the camera is a mono-CCD matrix camera. A tri-CCD camera, which is more precise since it makes it possible to obtain the three basic colours (Red-Green-Blue) separately, is not necessary, but it makes it possible to access more detailed information resulting in a reduction in uncertainty on the measurements made with the colour images.
4) the data processing system D comprising
.a digitisation card used to digitise the information provided by the camera and offering numerous setting possibilities (colour, saturation, contrast, luminosity, etc.),
- the image processing application: it is used to perform acquisitions with the camera and then process the images in different formats (RGB, HSL, greyscale.

etc.). A macro-instruction system enables the automatic acquisition processing operations. We used OPTIMAS.
- the computer used is sufficiently powerful to process the images acquired (64 0*4 80 pixels * 3 components) . In this case, the computer used is a PC with 48 MB RAM.
5) The colour of each product is thus expressed pixel by pixel in a three-dimensional colorimetric space and each shade of colour may be transferred to a histogram using which the desired data may be extracted.
By classifying the pixels according to their components, a spectrum E is defined by projection on a particularly suitable axis to accentuate the differences in response. In this way, according to their degree of calcination, the, grains, appear throughout the axis. The mean of the spectrum can easily be linked with the mean degree of calcination, itself correlated with the BET surface area: in this way, each product range shows a specific spectrum appearance. In addition, the dispersion (spread of the curve) gives a reliable estimation of the heterogeneity of calcination of each product.
Example of application
Chromatic space reference system: RGB, the axis on
which the spectrum is projected being the B-R axis
In this reference system, the greater the blue component of the pixel, the more the corresponding part of the grain is considered to be calcined. Therefore, we calculated for, each pixel the difference between the

blue component (noted from 0 to 255} and the red component (as above} and recorded on the axis B-R+100 (difference between blue component and red component + 100) the pixel populations corresponding to difference levels" grouped in intervals of 10. The histograms obtained in this way have the same appearance as the characteristic spectra in figure 2, relating to three aluminas with different calcination degrees.
Alumina a is a metallurgic alumina, subject to light calcination, and a significant proportional of its grains have a blue component less than the red component: they are rather brown in appearance. On the other hand, alumina c is an alumina calcined in the presence of a mineralising agent. It comprises a majority of blue grains. Alumina b is an intermediate alumina which comprises a spectrum located between both extreme spectra. Alumina a, b and c comprise declining BET surface areas (a: 75 mVg; b: 3.9 mVg; c: 1.1 m2/g). The products thus appear to be classified according to their calcination degree or, in reverse order, according to their BET surface area.
It is noted that some peaks are characteristics of the alumina family:
- peak at 60: under-calcined grains,
- peak at 180: calcined grains.
Determination of calcination index
On the basis of spectra such as those in figure 2, it is possible to determine a calcination index IC;
If i is the classes from 0 to 2 60 (interval of 10} , f1 is the percentage and c1 the value of the

di fference [blue component - red component + 10 0) corresponding to each class i, the calcination index is defined by:
For all the products, a bijective correspondence between the calcination index IC and the BET values (figure 3} is found.
Figure 4 shows a correlation conventionally defined between the BET surface area and the crystallite size TC - characterised here by the fine diameter d50 obtained by means of granulometry after grinding - in the range of interest to ceramists and refractory scientists (0.8 to 5 m2/g). This correlation is of the type:
BET = A/(TC-L)+B
where B (of the order of 0.5 m2/g) represents the analytical limit of the BET measurement and where L (of the order of 0.-4 ^m) represents analytical limit of the granulometric measurement.
Therefore, it is possible to correlate the calcination index IC with the crystallite size TC and, in the range of interest to ceramists and refractory scientists a linear equation: TC = A"»IC+B" (Figure 5) is obtained. Such an equation is refined for each product range.
Determination of homogeneity index
The spectra of some products undergo more spreading than others and "comprise varied grain colours. The homogeneity index was defined to account

for the spreading and dispersion of such curves. The standard deviation of the frequencies was selected for one product family.
To illustrate the representative nature of this index,"we prepared two products in the laboratory: one calcined [BET = 1.4 m2/g) and one under-calcined (BET = 77 m2/g) that we mixed. The first (corresponding to "100% alpha" in figure 6) has a confined curve, the second (corresponding to "0% alpha" in figure 6) is more dispersed. The spectra obtained on different mixtures (corresponding to "x% alpha", where x is successively equal to 95, 90, 85, 80, 60 and 40} are presented in figure 6.
Figure 7 shows, as a function of the percentage of calcined alumina, (symbolised by "% alpha"), the variation in" the homogeneity index IH of the different mixtures obtained. Note that any mixture comprising up to 70% calcined alumina has a lower homogeneity index than that of under-calcined alumina.
Advantages of the method according to the invention - rapid and reliable response
measurement method can be integrated into monitoring of calcined alumina production; in practical terms, the preparation of the sample requires some training but the determination of the coefficients IC and IH is automatic, enabling the operator to react rapidly in the event of a deviation in production parameters. This more rapid action has two beneficial effects: greater furnace flexibility and improved

stability of calcination conditions. In this way, it is possible to gain 20% on out of specification products.


We Claim:
1. Method to measure the degree and homogeneity of calcination of alumina
characterized in that said method is carried out with an image analysis device
equipped having a sensitive camera in a spectral analysis window corresponding to
a wavelength range equal to or in the vicinity of visible light and in that it
comprises the following steps:
a) mixing the alumina to be analyzed with a liquid which has a refractive index that is, in said wavelength range, between the refractive index of a lightly calcined alumina and the refractive index of a strongly calcined alumina;
b) preparation of a slide for observation of said mixture in said image analysis device, said mixture being illuminated by stable polychromatic radiation compatible with said spectral analysis window;
c) reception of the image by the camera and processing of the signal resulting in the definition of an image composed of a given number of pixels with three colorimetric components;
d) statistical processing of said pixels using their colorimetric components and used to give the calcination degree and the homogeneity of the calcination.

2. The method as claimed in claim 1, wherein the spectral analysis window of the camera is the visible spectrum and wherein the refractive index of the liquid is chosen between 1.70 and 1.76.
3. The method as claimed in claim 1 or 2, wherein the magnification system of the image analysis device is such that it makes it possible to obtain images of said mixture containing between 50 and 1000 particles.

4. The method as claimed in claim 1 to 3, wherein the images are defined by pixels with three RGB colorimetric components.
5. The method as claimed in claim 1 to 4, wherein the pixels are classified according to the "B-R" (blue Jess red) component, the histogram obtained in this way having a mean and a standard deviation characteristic of the degree and homogeneity of calcination of the alumina observed in this way, respectively.
6. A monitoring method of the degree and homogeneity of alumina calcination
having sampling of the alumina produced characterized in that the sampled
alumina is analysed according to the measurement method as claimed in any one of
claims 1 to 5.

Documents:

in-pct-2002-0980-che abstract.pdf

in-pct-2002-0980-che claims-duplicate.pdf

in-pct-2002-0980-che claims.pdf

in-pct-2002-0980-che correspondences-others.pdf

in-pct-2002-0980-che correspondences-po.pdf

in-pct-2002-0980-che description (complete)-duplicate.pdf

in-pct-2002-0980-che description (complete).pdf

in-pct-2002-0980-che drawings.pdf

in-pct-2002-0980-che form-1.pdf

in-pct-2002-0980-che form-19.pdf

in-pct-2002-0980-che form-26.pdf

in-pct-2002-0980-che form-3.pdf

in-pct-2002-0980-che form-5.pdf

in-pct-2002-0980-che pct search report.pdf

in-pct-2002-0980-che pct.pdf

in-pct-2002-0980-che petition.pdf


Patent Number 216165
Indian Patent Application Number IN/PCT/2002/980/CHE
PG Journal Number 13/2008
Publication Date 31-Mar-2008
Grant Date 10-Mar-2008
Date of Filing 25-Jun-2002
Name of Patentee ALUMINUM PECHINEY
Applicant Address 7, place du Chancelier Adenauer, F-75218 Paris Cedex 16,
Inventors:
# Inventor's Name Inventor's Address
1 MARTINENT-CATALOT, Valerie 38, Hameau de Biver, F-13120 Gardanne,
2 LAMERANT, Jean-Michel Mas de l'Escaillon, F-13320 Bouc-Bel-Air,
3 ROUIT, Bernard 90, avenue des Chutes Lavie, F-13004 Marseille,
PCT International Classification Number G01N 21/31
PCT International Application Number PCT/FR2000/003227
PCT International Filing date 2000-11-21
PCT Conventions:
# PCT Application Number Date of Convention Priority Country
1 99/14951 1999-11-26 France