NL8902196A - AUTOMATED METHOD FOR IDENTIFYING MINERALS AND CHARACTERIZING STONES. - Google Patents

Description

AUTOMATED METHOD FOR IDENTIFYING MINERALS AND CHARACTERIZING STONES

This invention relates to an automated method for mineral identification and rock characterization.

In general, mineral information is of great significance for many drilling and extraction techniques for oil and gas, for example, to control reservoir quality, and the interaction between rock and fluid during extraction operations, etc.

Such information about the mineralogical composition of a sample, for example a sandstone sample, is usually obtained by examining a thin segment using an optical microscope. Initially, the geologist will only indicate estimated amounts of the minerals present. If more accurate data is required, the sample is usually examined according to a certain point count. This is a labor-intensive technique and involves identifying the minerals that are present at about two or three hundred points per thin segment. This technique analyzes each individual point based on its color and structure. However, such a method still only provides semi-quantitative information, in particular on minerals that are present in small amounts. Since the segment has a limited thickness (30 μια), problems are also encountered when points fall on grain edges. At such points it is difficult to determine which mineral (or which pore) should be counted, which further limits quantitative analysis.

An automatic prior art analysis method has resulted in an automatic point counting method in which a sample is scanned by an electron beam such that a coarse grid of 50 x 50 points is obtained.

A full X-ray spectrum is collected and analyzed at each point. In order to identify the mineralogical composition of each point in the grid, X-ray counts in twelve energy regions are compared to normalized values in said ranges of common reference materials, with the smallest difference between the unknown spectrum and one of the comparison spectra determining the respective point in mineralogy.

Rock characterization methods are also known, particularly those involving the analysis of reservoir pore complexes. The recent state of the art describes an automated scanning electron microscope (* = scanning electron microscope -SEM) imaging method, in which backscattered electrons (= back-scattered electrons - BSE) form an image of a polished mineral sample by developing analog signals displaying time-varying voltages proportional to the brightness on site. In this method, prior to analysis, the pores are filled with epoxy resin, which reacts completely differently from the heavier mineral grains.

The images thus obtained can be processed in various ways. The SEM images are, for example, scanned by a video scanner, after which the obtained analog television signal is converted into a digital form. In such a digital display, the observed field is electronically arranged in an arrangement of grid points or so-called pixels (picture elements) which are spaced on a "gray level" scale. The individual pixels of such a digital image are classified as either rock or pore.

In another example, color filter images of mineral samples are obtained, impregnated with pigmented, in particular blue, epoxy. Such filters also produce pixel array distributions. Values of "one" and "zero" are assigned to each color image to distinguish between pores and other matter.

In further analysis of the images obtained in the above manner, erosion and dilation techniques are used, with which pore complex properties can be determined.

Only limited mineralological information can be obtained using BSE analysis. The problem lies in the fact that there may be a gray level overlap area for the different minerals; Minerals (clays) with a variable composition can also cause a gray level overlap area, which hinders accurate mineralogical identification.

The aforementioned research techniques are troublesome in that they involve rather complicated processing techniques and indirect imaging techniques.

With the above mineral identification method, only a few sample points are analyzed. Intermediate sample areas and gaps are not sufficiently investigated. Grain edge and pore wall properties cannot be determined at all in this way.

With the above rock characterization methods, the necessary characterization data is obtained only after performing intermediate steps such as scanning the BSE image and then filtering the scanned image.

A main object of the present invention is to improve the mineral identification methods described above, in particular to provide a method of viewing the entire sample, and more particularly to provide a fully automated method that allows both mineral classification and geometric characterization of rock and pore network is possible, in which method no intervention of an operator is required.

Another object of the present invention is to improve the above mineral classification method.

Yet another object of the present invention is to perform pore network and rock matrix methods that directly process BSE information data.

In order to achieve the aforementioned objectives, to overcome the shortcomings discussed and to accomplish said objectives, the present invention relates to a computerized method for analyzing and characterizing polished mineral samples, which method comprises the steps of automated collection, memory storage, reading, visual displaying and processing sample data, using electron backscattered electron (BSE) imaging methods from at least one inspection field of said sample, thereby obtaining BSE information data for further processing, and inelastic electron scattering X-ray sampling of said field to determine the mineral composition of said field of said sample, and wherein said BSE information data is digitized and directly converted into gray level information data to obtain information data concerning the rock matrix and pore network, and whereby with a mineral identification method with regard to said field individual objects are determined in an automatic determination step based on said BSE information data, and which specific objects are examined by exposing said objects each in its entirety to X-ray object scanning using high-resolution inelastic electron scattering, which provides X-ray data of atomic elements in said objects, using which ratio classification data is derived to determine the amounts of minerals in said sample.

The invention also relates to a method wherein said gray level information data is further processed by the following successive steps: segmenting gray levels; distinguishing individual objects in the inspection field, i.e. pores or granules, by detecting dark and non-dark phases; separating said objects from each other by sharpening contours; reconstructing said objects; determining both rock matrix and pore network; and evaluating the information matrix and pore network information data thus obtained, in order to obtain parameters concerning the rock matrix and pore network.

In accordance with another aspect of the present invention, a method is proposed in which said ratio classification data is compiled by element ratios of investigated objects, and wherein said element ratios are compared to prototype mineral ratios to determine the mineral composition of the investigated objects.

Yet another aspect of the present invention pertains to a system for analyzing and characterizing polished mineral samples, comprising the use of an accelerated electron microscope (SEM) producing an accelerated electron beam and a backscattered electron detector (BSE) detector, an energy dispersive X-ray (EDX) analyzer for analyzing the mineral composition of said samples by inelastic electron scattering, and an image processing system (IPS).

A particular advantage is the accurate classification of the minerals, even when small changes in the composition of the sample to be analyzed occur.

However, the main feature of the above method is in the combination of the BSE and EDX techniques. With the improved BSE method, the objects to be analyzed are precisely defined. With said EDX technique, a greatly improved mineral classification is achieved, and in addition, imaging of said method leads to an overall characterization of a sample, which is of great benefit to its users, both in the field of drilling and extraction techniques for oil and gas, as well as for mining and soil research.

Further aspects, features and advantages of the present invention will become apparent from the following description and the accompanying drawings, wherein:

Figure 1 shows a system with which the method according to the present invention is carried out;

Figure 2 schematically shows the mineral identification method of the present invention;

Figures 3A through 3D show various processing steps of the mineral identification method of the present invention;

Figures 4A and 4B show results of classification methods known respectively from the prior art (A) and the technique of the present invention (B);

Figure 5 shows a schematic representation of the pore network and rock matrix characterization method of the present invention; and

Figure 6 shows a worked example of a sample analyzed and characterized according to the present invention.

In Figures 2 and 5, the same reference numerals are used for corresponding process steps.

Figure 1 shows a system with which the method according to the present invention is carried out. The system consists of three main parts that are interconnected through several coupling parts (not shown; for clarity, these coupling parts are represented by a single block of electronics for bundling, scanning and imaging): a scanning electron microscope (SEM) 1, a energy dispersive X-ray analyzer (EDX) 2 and an image processing system (IPS) 3.

The SEM 1 comprises a SEM column 10 and uses an accelerated electron beam (up to 30 kV) under high vacuum conditions, to scan the sample to be examined. For the work described here, samples impregnated with resin, especially an epoxy resin, are cut and highly polished to obtain a flat surface. They are then coated with a conductive carbon layer to prevent build-up of electric charge. The different types of electrons and photons generated at the point of impact of the electron beam on the sample surface are observed separately. As the electron beam scans the sample, the response of any of these detectors can be used to build a characteristic image of its surface. The detector commonly used in the work described below is a backscattered electron detector 14.

In Figure 1, several beam conditioning functions, such as magnification, size of the light spot, scanning method, detector and focal point, are represented by a single line 12, and X and Y control by a line 13. Beam stabilizing means are also included in the shown SEM column ( not shown for clarity).

A sample 16 is arranged on a platform 15, or more samples can be arranged, so that the sample (s) can be moved and replaced if desired. At least one inspection field, i.e. a beam-covered area, will be scanned for each sample. More inspection fields can also be scanned for one sample, for example eight fields. In such a way, a more representative sampling is obtained, which leads to a more precise identification of the mineral and / or characterization of the sample. In general, the samples to be examined are minerals, such as rocks.

It is noted that the above functions are controlled by an electronic switching system, indicated in the drawing by a block 17, just electronics for bundling, scanning and imaging.

The energy dispersion X-ray (EDX) detector and analyzer consists of a detector and analyzer 20 also connected to said block 17. The analyzer detects the X-rays generated by the interaction between the electron beam and the sample with a detector, while the x-ray energy is measured to obtain qualitative and quantitative information about the chemical composition of the sample. This is possible because each atomic element in the sample has its own specific energy distribution of emitted X-rays.

The image processing system (IPS) 3 comprises a central processing unit (CPU) 31, based on a high-speed array processor (array processor), and is suitable for high-speed image processing, amplification, reconstruction and analysis. The images to be processed are obtained from the SEM 1, while the X-ray data required to identify the minerals are obtained from the EDX. The SEM, EDX and the IPS are interconnected through an interface. In fully automated operation, the IPS CPU controls all applicable functions of the microscope and X-ray analyzer. As mentioned above, functions such as magnification, beam size adjustment, test method, detector selection, automatic focal adjustment and spiral feed of the SEM can be controlled, as well as those for the EDX system and the collection and processing of X-ray data.

The IPS 3 also includes an operating interface (with display) 32 which is used by operating personnel to initiate and set up the required functions and operations.

It will be clear that with an automated system as described here the operating time will be considerably shortened.

Figure 2 shows the mineral identification method according to the present invention schematically.

Two main techniques are used to determine the mineralogical composition using an SEM / EDX system, namely the imaging method using back-scattered electron (BSE) and the X-ray analysis. First, a rough outline of the basics of these techniques will be given below. Thereafter, a detailed description regarding the various procedure steps will be given with reference to Figures 2 and 3A-D.

As a result of the primary electron beam scan of the sample, backscattered electrons are generated. The detector for these electrons (14 in Figure 1) is mounted in the SEM just above the sample. The signals provided by the backscattered electron detector as BSE information data reflect the average atomic density at the impact site on the sample and thus provide information about the average composition at that location. As a rule, the resolving power of BSE images for mineral samples, such as rocks, is about 1 micron at an accelerating voltage of 20 kV. In SEM studies, the BSE imaging method is mainly used to obtain information about the spatial distribution of minerals, grains and pores.

Figure 3A shows a BSE image of a sandstone sample where the gray level refers to the average atomic number of the mineral. The heavier the mineral, the brighter it will be seen in the BSE image. The pores are dark because they are filled with a low density epoxy resin. Starting from such an image, the image analyzer can recognize certain groups of minerals by detecting only those areas whose gray levels fall within a specific area. The method is called segmentation. However, since mutually different minerals can have substantially the same atomic numbers (for example quartz, dolomite and sodium feldspar), the segmentation is only suitable for making partial distinctions. To further distinguish such minerals from each other, X-ray analysis should be used.

The X-ray analysis system measures the X-ray energy spectrum, i.e. the intensity and energy of the X-rays. In an SEM, X-rays are generated during inelastic scattering of primary (beam) electrons. These X-rays are formed by two distinctly different processes: (1) the delay of the beam electrons in the Coulomb field of the atomic nucleus gives a continuous background spectrum ("Brems-strahlung") and (2) the interaction between a beam electron and electrons of the inner shell lead to characteristic lines in the spectrum. The continuous (background) part of the spectrum is not used in the mineral identification method. Detection and classification of the characteristic X-ray lines allow identification of the atoms (elements) in the excited sample volume. The number of rays emitted by each element is also related to the concentration of that element.

The minimum concentration of an element that can be observed by micro-X-ray analysis is at best 0.1%, and generally less than 1%. For most rocks, the spatial resolution is about 1 micron.

The automated mineral identification method basically involves recognizing individual objects (i.e. granules, diagenetic minerals, clays) in the sample. Each object is subjected to X-ray analysis to obtain X-ray spectrum data from elements contained in the sample, which may be referred to as mineralogy. It is advantageous to obtain X-ray spectra of nine elements. The complete procedure 40 consists of a number of steps 41-50 which are described in detail below and are shown in Figure 2.

In order to obtain an image of the SEM, it takes IPS

various functions of the microscope under control. First, an inspection field on a sample is selected, ie an area to be examined by a scanning light beam, represented by a block 41. Then, in a BSE imaging step 42, the appropriate detector and magnification are selected and the scanning of the electron beam on the sample is selected regularly. While the SEM scans the sample, the IPS converts the incoming BSE information data into a gray level discriminating step 43 into digital form and stores it in a video memory. For example, a stored image has a standard format of 512 x 512 points (or pixels), each point can have a value from 0 to 255, shown as gray level information data. For mineral samples, the gray values advantageously range from 0, which corresponds to black (pores), through gray (quartz) to 25.5, which corresponds to white (heavy minerals). It is noted that more sophisticated image information can be obtained with advanced information processing equipment. Increasing research facilities, for example up to 10 bit x 10 bit pixels per image with 4 byte gray values for each pixel, are now possible.

Once the data is stored in a video memory by the IPS, a large number of image processing functions are available to enhance the image. These functions can be used to filter out noise, improve contrast, sharpen contours, distinguish, separate and identify certain objects in the image, as indicated by a block 44 in Figure 2.

In BSE images of sandstone samples, the edges of grains and the points where grains meet often have a darker color than the interior of the grain (Figure 3A). If this darker edge were present around the entire grain, each grain could be distinguished from the other grain by a simple gray level segmentation method. However, the tangent points between the grains are sometimes not visible over the full length, and therefore a more advanced grain detection method is needed. This again implies segmentation of the BSE image, resulting in a binary image in which, for example, the grains are white and the pore space is black. In this image, groups of two or more grains will still remain that are not "separated" but that touch each other at narrowings, as shown in Figure 3B. These constrictions are removed using different morphological methods, resulting in separate "separated" objects, see Figure 3G.

Once the objects (i.e. separated grains) in the image have been recognized and identified, the coordinates of all points belonging to an object are known. The IPS then takes care of the control of the electron beam in the microscope, and the beam scans the area of each object successively. X-rays are generated during this procedure. A problem may arise when two objects with different composition of elements are in contact with each other. For example, it is possible to detect X-rays generated by an object other than the one being examined. In order to prevent this, all objects are eroded mathematically, i.e., pixels from the object boundaries are omitted, thereby reducing the actual scanning area from which the X-ray pulses are detected, as shown in Figure 3D. Such a procedure, as indicated by block 45 in Figure 2, will lead to the removal of small objects from the image. Therefore, only those objects that do not disappear with this method will be scanned. With the present method, an X-ray spectrum of each object is advantageously obtained in an object scanning step 46 using nine predetermined energy windows, each window representing an element. The X-ray counts are normalized and sent to the IPS as a vector of nine elements. In the IPS, the contribution is removed from the background prior to the classification of the minerals.

Once the X-ray data has been collected and sent to the image analyzer, an attempt can be made to classify any object or grain in a classifying step 47. This is done by comparing the vector of X-ray pulses obtained from an object to vectors previously obtained. of known minerals or "prototypes". This method can be considered an "absolute method", due to the use of absolute values in said classification. When the vector of the unknown mineral is sufficiently similar to that of the prototype, the unknown mineral can be classified.

A method known from the prior art is shown in Figure 4A, where the concentrations of Si and Al (roughly equal to the X-rays at the correct energy) are used only for classifying some silicate-containing minerals. With said classification method, a vector of nine elements (Na, Mg, Al, Si, P / Zr (combined), K, Ca, Ti / Ba (combined) and Fe) is composed in order to distinguish between a larger number of minerals. The elements P / Zr and Ti / Ba are combined due to peak overlap in the X-ray spectrum. This method works successfully for many minerals, but problems are encountered when classifying minerals whose composition varies slightly. Dolomite, for example, can contain small amounts of iron. When the X-ray pulses of such granules are compared to that of a "pure" dolomite, the vectors may differ too much to allow proper classification of the mineral. Defining a proto-type composition for any slightly different mineral would be an extremely difficult if not impossible task. Another problem with this absolute method is the amount of analysis time required to reduce statistical fluctuations in the data, which can lead to incorrect classification of some minerals.

In view of the above problems of the present invention, the ratios between the elements of the unknown grain are compared to those of the known prototype minerals. Figure 4B shows this method, which is to be considered a "ratio method", for the same silicate-containing minerals as those of Figure 4A. In this way, statistical fluctuations have less influence on the classification method, which leads to a more accurate classification. The analysis time per grain is about 1-2 seconds. It is also possible to account for slight variations in composition without the need to define a large number of prototype compositions for each mineral. The results of classifying the grains detected from the BSE image of Figure 3A can be shown in color, the different colors representing the different mineral types, as indicated in Figure 4A, B. Further processing of such an image is necessary before attempting to quantify mineral percentages.

Due to the initial erosion of the grains used to prevent edge influences in the object scanning procedure, the original object boundaries are lost. Before measuring the mineral percentages in the image, the objects must be restored to their original shape in a reconstruction step 48. A similar technique to that used to separate objects is used for this repair procedure. For a homogeneous sample, measuring the area percentage of a mineral on 2-D segments is a good estimate of its volume percentage in the large mass of rocks. The larger the total research area of the segment, the better the estimate. Measuring the area percentages of the pores and the various minerals in an evaluation step 49 is a simple and fast method using image analysis. Finally, the mineralogical composition of said sample can be obtained in a mineralogical determination step 50, said mineralogical determination being significantly improved over that of the prior art using the above "ratio method".

The above methods are all linked together and can be applied for one specific inspection field. However, in order to allow analysis of more inspection fields, it is necessary to move the sample. This is performed by a motor driven scanning platform (as shown in Figure 1) mounted on the SEM which can be remotely controlled by the IPS. Before starting this procedure, the coordinates of the required analysis positions are stored in the IPS memory. This means that automatic analysis of a number of inspection fields or samples can be performed in one method, without the need for an operator (e.g. overnight).

In addition to the analysis of mineral quantities, it is also important to determine various parameters regarding grain structure (eg size, shape, etc.) of the samples. This can be done for the complete attrition collection or only for certain minerals (e.g. quartz). Different parameters of size and shape are available to the user (e.g. area, maximum or minimum diameter, elliptical shape factor (i.e. ratio between small and large ellipse axes)) and other parameters can be defined if desired. Before these analyzes can be performed, it is necessary to remove those grains from the image that touch the edge of the inspection field, because their geometric parameters cannot be accurately determined.

Another method that can be easily performed after mineral identification is to determine the pore wall mineralalogy. The determination of the grain structure parameters and the pore wall mineralology add little to the overall analysis.

Figure 5 shows schematically the method according to the invention for characterizing the pore network and the rock, indicated by the number 60. Said method also uses a BSE image of an impregnated and polished rock sample, collected and transferred as said BSE information data from the SEM to the CPU. In order to obtain sufficient statistical accuracy in the analysis, a number of inspection fields are collected for each sample at an appropriate magnification. In particular, the chosen magnification should be small enough to avoid statistical scattering, but large enough to define small pores and small mineral particles. As discussed above with regard to the mineral identification method, in particular with regard to steps 41, 42 and 43, a BSE image provides a certain amount of mineralogical information, with heavier (high atomic density) minerals being brighter. The pores are filled with low density epoxy resin and in contrast look very dark. The BSE image is improved by using the so-called gray level processing technique, such as local noise filtering.

The pores are then distinguished from the mineral constituents in the image by detecting only the dark phases. Mathematical morphology techniques, proposed by successive steps 61, 62 and 63, allow to delineate individual pores, defining the pore necks as constrictions in the 2-D pore network (pores hitting the edges of the analysis excluded because their geometrical properties cannot be accurately determined).

A similar procedure, proposed by successive steps 64, 65 and 66, is used to derive the rock matrix from the BSE image, distinguishing the grains from the pores by detecting non-dark phases only. Pore area and pore neck length distribution are measured, as well as the two-dimensional coordinate number (i.e. the average number of pore necks per individual pore). Starting from the rock matrix image, grain diameter (grain size), grain smoothness and asparagus ratio (grain shape) are measured.

In the scheme shown in Figure 5, the two approaches for the characterization method are clearly indicated. When data on pore network and rock matrix have been obtained, images will be taken and the data on rock matrix and pore network must be displayed separately. Then a "final." imaging step 67 performed to define these images individually.

The last two steps of this method include evaluating the image data in an evaluation step 68 and measuring pore network and rock matrix parameters in a step 69, as indicated in Figure 5.

As an example, the results of identifying and characterizing the sample, as noted above, i.e. the results collected over eight inspection fields, shown in Figure 6, illustrating the great advantage of combining the above methods for taking BSE images and taking X-ray samples.

It will be apparent to those skilled in the art that the example elaborated above, and the methods employed, may be extended to, respectively, other parameter measurements and other methods not currently used, such as neutron scattering methods and 3D analysis.

It is hardly necessary to mention that the method discussed here and the system followed can be modified and modified without departing from the scope of the present invention.

Claims (6)

A computer-aided method for analyzing and characterizing polished mineral samples, which method comprises the steps of automated collection, memory storage, reading, visual display and processing of sample data, using electron backscatter imaging methods ( back-scattered electron (BSE) of at least one inspection field of said sample, thereby obtaining BSE information data for further processing, and using inelastic electron scattering X-ray sampling of said field to determine the mineral composition of said field of said sample, and wherein said BSE information data is digitized and directly converted into gray level information data for obtaining information data regarding rock matrix and pore network, and with a mineral identification method regarding said field separate objects n are determined in an automatic determination step based on said BSE information data, and which particular objects are examined by exposing said objects each in its entirety to X-ray object scanning using high-resolution inelastic electron scattering, thereby obtaining X-ray data from atomic elements in said objects, using which ratio classification data is derived for determining the amounts of minerals in said sample. The method of claim 1, wherein said gray level information data is further processed by the following successive steps: segmenting gray levels, distinguishing individual objects in the inspection field, i.e. pores or granules, by detecting dark and non-dark phases; separating said objects from each other by sharpening contours; reconstructing said objects; determining both rock matrix and pore network; and evaluating the information matrix and pore network information data thus obtained, in order to obtain parameters concerning the rock matrix and pore network. The method of claim 2, wherein said parameters include: pore area, pore neck length, distribution of said pore area and pore neck length, and coordination number. The method of claim 2, wherein grain characteristics are calculated from the following parameters: grain diameter, grain smoothness and aspect ratio. The method of claim 1, wherein said ratio classification data is constituted by element ratios of the examined objects, said element ratios compared with prototype mineral ratios, to determine the mineral composition of said examined objects. System for analyzing and characterizing polished mineral samples, comprising a scanning electron microscope (SEM) that produces an accelerated electron beam, and a detector for scanning backscattered electrons (BSE), an energy dispersion X-ray ( energy dispersive X-ray = EDX) analyzer for analyzing the mineral composition of said samples by inelastic electron scattering, and an image processing system (IPS).