JPS59163549A - Automatic analysis of electron-ray diffraction figure - Google Patents

Abstract

PURPOSE:To carry out automatically on-line the identification of a sample from an electron-ray diffraction figure obtd. by an electron microscope or the like, by obtaining distances from a central spot to a point on which the intensity distribution of a picture element on the line passing through the central spot becomes maximum, determining the lattice-lattice distance value of the sample, and checking them with the data of various substances. CONSTITUTION:The intensity of a picture element corresponding to the coordinates of (x), (y) is taken as Uxy and the coordinates (Uo, Vo) of a central spot are determined by obtaining the coordinate when each SIGMAUxy on (x), (y) becomes maximum. As to the intensity distribution of the picture element on the straight line passing through the central spot, the intensity becomes maximum at the point corresponding to diffraction rings. The distances from the central spot to each maximum point and the radii of respective diffraction rings are required. The coordinates (u, v), the distances Rs and (d)-value Ds of the spots are required, and the radii and (d)-value Dr of the rings are obtd. The data stored on an outer storage device 7 are read, and checked with the (d)-value data of samples by an electronic computer 4.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field 1] The present invention relates to a method for automatically identifying a sample on a Δn-line from an electron beam diffraction image obtained by an electron microscope or the like.

[Prior art] Using an electron microscope, a diffraction image of a sample is captured in the 1z: mode or scanning transmission image mode, and this diffraction image is analyzed to identify the sample or to determine the index {vl[j or The orientation of the crystals in the glass 1 was measured, but the conventional method
Well, I had no choice but to do it manually.

That is, when J3 is set from 1;
Then, find the value of d{li'1, which is the lattice spacing of the crystal.
j Leeru d {11'i is recorded and compared with 6 cards, select the card with the most matching set of d values, and play Mue (
The index of the diffraction image is calculated by using the control chart (h, ni, l) written on the selected card.

However, such conventional methods require knowledge and experience, involve uncomfortable dark work, and take a long time.

[Objective of the Invention 1] The present invention solves such conventional drawbacks and provides a method for analyzing an electron beam diffraction image by automatically analyzing a diffraction image online and identifying a sample. It is intended to provide.

[Structure of the Invention] The present invention converts one electron beam diffraction image into an electrical signal, and converts one electron diffraction image into an electric signal,
When the intensity of the pixel corresponding to y is expressed as UXV, the intensity distribution of the pixels on the straight line is the maximum.By finding the distance between the point and the center spot, the four corners of the lattice plane iii' o) By comparing and overwriting the set of lattice spacing values with the data that does not represent the lattice spacing values of various materials, it is possible to perform a trial) L'iJ determination. It is a characteristic feature.

"Action of the Invention" Below, the basic idea of the present invention will be explained in 7+g.

When the diffraction image is converted into pixels with a two-dimensional distribution of 1-, let the intensity of the pixel corresponding to the coordinates x and y be Uxy, and the coordinates of the spot at the center of the diffraction ring be (uO2Vo),
The central spot is the one that is maximum in the plurality of diffraction phosphor marks IJQ and VQ. The stage where the coordinates of the central spot have been determined - (゛, on the straight line passing through this central spot (
When examining the intensity distribution of the falling pixels, the intensity should be at a maximum at the point corresponding to the diffraction ring. Therefore, if the distance from the center spot 1 to each maximum point is determined, the radius of each diffraction ring can be determined.

"Example" The present invention is based on this idea, but
In order to make the embodiment easier to understand, the following points will be explained in advance.

As is well known, each diffraction image is shown in FIG. 1(a).

As shown in (b) and (c), one consists only of a sporamil-like diffraction image based on an electron beam diffracted from a relatively large crystal, and the other consists of a ring-shaped diffraction image based on an electron beam diffracted from a powdered crystal. There are three types: one consisting only of a diffraction image, and one consisting of a mixture of both. Therefore, when performing manual analysis, it is necessary to perform separate processes for spotting, spotting, and ringing, but in the case of manual analysis, intuition ('easy to judge') is necessary. In order to analyze the distinction between the three types in A-line, which is automatically published, it is necessary to use too many complex
■Spelling l41<,? X! It is essential to be able to distinguish it mechanically. In the case of 1 to 1, only the ring, or both, the part corresponding to the brightness occupying the image plane (■11 et al. The ratio of γil!min) is 5-"/,j-runoJ, and the time IJr image is changed to the degree of brightness of the groove inner element.
(Pixel detection game)) Maro and Toide Jtl (((l+
) i51, spot 1 ~ only, ring I no 2 ministry/
NoX11? White Shinoko ↓ straddles white (゛(Well, the second picture smells)
, dotted line A, dashed line 1-1. Two points g4' J line c (acceptance -
There is a clear difference between J and U.13However, in Figure 2 there is a clear difference between A3 and
The horizontal axis of C1 indicates the intensity I, and the vertical axis indicates 1-1 (1), which is indicated by a solid line. The difference between
For example, the value σ given by the following formula (1), which is usually used to find the variance, is calculated, and this value corresponds to the cases of spot l-, ring, and both cases. The type of diffraction image is determined depending on where the diffraction image falls within the area. However, in the first equation (, I is the weighted average given by the following equation 0, 1 = (Σl-1
(1) I l) / (ΣI-1(I))...
(2) d.
Therefore, in the figure, 1 is an electron microscope, and 1 is an electronic meter.
For the deceptive mirror 1, set the electron microscope 1 to transmission image observation mode,
An electron microscope 1 is supplied with a 1-il-line bow for transmissive image observation. Da, buy 2 or take (=j(,:J), and when the electron microscope 'I or transmission image observation becomes one
, Tsu-(3! Analog time series signal corresponding to the 3 images (j, ΔD☆(kuki 3 j,
When the electron microscope 1 is in the scanning transmission mode, the transmission radiation detector 5 is inserted on the optical axis. A detection signal of Δ excess electrons (when the sample is scanned with an electron beam) is converted into a digital signal by an AD converter 6 and is supplied to an electric power calculator 4.

The electronic measurement rule I is based on the image signal A''J supplied from the imager 2 or the transmission electron detector j5, the results of identifying the sample 31, and the set of d values for various substances for search. It is connected to an external storage station 7 for storing data, etc. of C.

The electronic computer 1 also includes a CRT 8 for monitoring a diffraction image and a keyboard 9. A CRT 10 for console is connected.

Using this J, Una'rM device, we will be able to manipulate it (
31 Give a command to the electronic meter 4 using the keyboard 9,
For example, set the transmission image mode 1 of the sample to be projected onto the imaging device 2. When the command is given to start the computer 1, the computer 1 proceeds with the analysis work according to the flowchart shown in FIG.

That is, first, the diffraction image projected onto the screen of this imaging device by the 11i imager 2 is converted into a digital signal by the AD converter 3, and taken into the electronic calculator 1 and 4 by the electronic converter 3. 512x in the storage section of Mr. 4! 1i Store and read the stored image data as 12 pixel data (step A) 13 Next, the electronic computer 4 outputs the stored image data 'C1 The diffraction images of this image are respectively shown in FIGS. ),
It is determined whether the car consists of only spots as shown in (c), or whether there is a mixture of -6 car consisting of only rings and b (step 13). This determination is performed as follows.1. As shown in FIG.
1! , the X coordinate is x, July j,
El 7J bow! (Assuming that the intensity of a certain pixel is expressed by UXy, the number of pixels whose intensity is iU I on the back side of all tjxy is expressed as 1 (!!' - 1, but the error t+ < 1 for intensity ■) ).Next, 1-1
Using the value of (1), a weighted average summary expressed by the -C equation (2) is obtained. Therefore, the variance σ expressed by equation (1) is calculated using the output value of this weighted average. Furthermore, this calculated time is 88. ．． 000 or less (Case 1
) or is above 105,0OOLJ (ris=-s2)
or between 88,000 and 105,000 (Case 3). However, in this case, the total number of pixels is 512 x 512, and the gradation of UXV brightness 8 (intensity) is taken as C (Y255).So, if l) - S1, the diffraction image is spora1 .

Now, based on this judgment, we can use the example (, J, diffraction (tI consists only of spora j) and η, then we can move to the suit block C shown in Figure 1'84 (-). There is a central spora h (up, vo) of the electron beam that has undergone no diffraction (up, vo) of the electron beam that has gone straight and the other diffraction spora 1 (?,
j (Ll.

V), and further calculate the distance m Rs from the center spora 1 to the valley diffraction sub-bottom 1 and the distance m Rs and each diffraction spora I ~ 1!
juru (1 (1 white DS)
). This Sue 1゛fC4,L edition 1111 is shown in the flowchart of Figure 6.
constant strength IαJ, small black level) and constant strength jσ
The following is divided into 1 (white level) (step C-1 in which the so-called 21Lj conversion process is performed), and the maximum 6 of the white level pixel groups in the bowl are treated as central spora (.) :
'J: L <, this central spora 1~ locus iM(uo.

Step C-2). (fl L, the coordinates (U, V) of a group of pixels with a white level are determined by the following calculation.

u = (umax+umin) /2-(3-1>v
= (vmax 10 vmin) / 2-
(3-2) Here, as shown in FIG. 5, umax and umin are the maximum X coordinate value and the minimum X coordinate i9"HC" of this pixel group, respectively, and the same is true for the ■ coordinate.Therefore, -CUQ is the average of the maximum and minimum values of the X coordinates of the largest pixel group, and vo-t) is the same as 3, followed by
The coordinates of the pixels of other clusters are also determined, and the coordinates (u, v) of each diffraction spora 1- are determined (step C-
4).

Next, from the center spora] ~ (uo, Vo) to each spora 1 (
Distance to u, v)! i1. Calculate Rs by 4.
Ask for one thing.

Rs -(u -110) 2-1- (V-\'o)
2=・(/l) In this case, for example, multiple distance ship data whose distances are within 5% of each other, what should be the same value for Kiri? It is thought that this is the case, so l'L Wj K! ! Using the average value of distance ERA (jll'j), C one distance 11 (
1. Determine l data 1. Sogo C1 Based on the determined distance value R3, the lattice spacing value of the crystal, dIi, is determined (step C-5). This d value is determined by the following general formula.

d-(1-・λ>/R...(5) However, in the above equation, R is the distance between the center spot l~ and the diffraction spot (or diffraction ring), and (L・λ) is the camera constant Since the camera constants vary depending on the state of the apparatus, those fixed using data measured using vapor-deposited gold particles or the like whose lattice 101 spacing is known are used.

Next, when it is determined that the ring consists only of rings, the process moves to step 1], and the ring radius Rr and d111'l[
) Find r3. This step [ ] is further shown in detail in Figure 7. Now, taking as an example the case of processing the d-J and U-shaped images shown in Fig. 8, the electronic needle stem (43 in section 4, ΣUXV for each pixel coordinate X) is output. Similarly, the coordinate y of each pixel is 34'S'U,
It becomes maximum based on the position of the center spot d.

Therefore, if we find the x and y coordinates of curves A and B that have the maximum values, these will be the coordinates of the center spot (uo,
Vo). In this way, the center spot 1~)・1
Find the reference mark (Ljo, V'o) (step i-) -2
), the eighth point parallel to the X axis passes through the center spot F-.
In the figure, the line segments 1) and 1) parallel to the Y-axis and the line segments 1) and 1), indicated by m'r, and the pixels on S are read out.

The values of the pixels on each of these line segments are aligned with the coordinate axes facing toward the center spot 1~ (as shown in Figure 9), respectively (a), (b>, (c), <( +
) Indicated by J, sea urchin, and H, there are +3, which have a maximum at the position of the diffraction ring. Therefore, in the case of electronically scheduled stoppage, Fig. 9 <a>, (+1>,
(c).

By adding up the intensities shown in (d), perform a performance to obtain the result shown in Fig. 9 (e), and use the coordinates of the center spot 1 ~ obtained earlier and this Fig. 9 <e>. If we calculate the distance from the coordinate where the J value is maximum, then we can calculate r for each ring radius (from this). 21 Once the ring radius Rr is calculated, we can calculate the distance from the coordinate where the J value is maximum. The d value is determined using equation (5).

Next, in step B shown in Figure 4(), spot 1-
A case will be explained in which it is determined that both a ring and a ring exist. When the determination is made in this way, the process moves to step 13 in FIG. 4 (step 13). 'j 1) In exactly the same way as in the case of -1, enter the projection operation to the X axis and the projection operation to YIIIll to find y of the center spot Jj)) J vote (U,
).

Vondoriru (step [--1). 1. This is because Riko's easy-to-contain diffraction image contains six spots.

1 for the intensity projected onto the X #+ and Y axes. Lu, ' J in Figure 11, a local maximum G based on the sea urchin spot 1 ~ occurs, but since the area occupied by the spot is -C smaller than the area of the ring, the local maximum G is There will be no harm to the camp. Next, the image is binarized in line 4r (step E-3), and the image is binarized according to this binarized image information.
Li2(R) spot and ring are discriminated in the image (step F-4). That is, for a pixel group of -15ν, the coordinates (U, \l ).If the pixel group is a ring, its coordinates are either the same as the coordinates (Llo, Vo) of the center spot found first, or it is located within a circle with an extremely small radius r'Q centered on these coordinates. (First, there is C. Then, for each pixel group found, it is determined whether the coordinates fall within this circle, and those that fall within this circle ~ b j) are regarded as rings, and the spot [Exclude from analysis processing as ~.If you mix up the ring with ~ and process it, many diffraction spots with small 4f distance IRs [- will become (f), In the above formula (5), there is a large number of people.
We have concluded that there exists a d value that makes the value right for flll, and therefore the correct judgment of trial 81 is (θ<'
L ri'tKru.

After finding the coordinates of each pixel group excluding the ring in this way, the coordinates (uo, vo) of the center spot have already been found, and f
fj Steps C=-3 to C-5 shown in 61'l C. Perform similar processing to find lj distance [[dIIl, I] for each spot 1~ (sup[- 5)
. Next (Step D in Figure 7)
-3 to slap D -5ff: By performing exactly the same process as in line 4I, the radius of each ring is 1 (1
(+(, i Dr call/search 1, (second), u(thus, sdep C(, L ] When S is obtained, these data are stored in the storage area F of the memory section of Denryo Kei 1 and 4, respectively.] XY, R8 and σDS+
I remember it. Similarly, when the ring radius) or 1゛ and d value Dr are determined in step D or ``, j,'', they are stored in the storage areas RR and DR, respectively. Perform the step ``external storage)'' and read out the data stored in the computer 7 and store it in the computer 4 (
[Already found] Check the data of this sample. In this case, the external storage device contains the name of the item, the crystal system to which it belongs +aO+110, as shown in Figure 12(a).
It is expressed as CO (8 expected reasons and multiple d value data (
1,,d2. ...(I n and each d value finger C (h,
ni, l)+, (11, ni, l)2. ...(11゜l(
, l) The original wave search data in which n is stored is stored, but since it is not possible to efficiently identify substances for a long time from this source wave search data, this data is d as shown in Figure 12 (b). In other words, the search for identification is carried out efficiently with the value as the center. In other words, this rearranged searched data is The d-value data of the other substances in the list are arranged in the order of 1 (for example, from the smallest one), and for each of the 11 numbers, the name of the substance that has this value is stored with an address corresponding to 1i15. 1. Compare this searched data with the d-value data already obtained by measurement according to 4.
For each d value, pick up the name of the substance that cuts the groove, and the name of the substance that has been mentioned the most times is the correct name -c'. ) as shown on the CRTIO. This identification result is shown in Figure 13 (
(a) is the crystalline island that was cited the most (although it may be possible to display the crystalline islands that were cited most often, J shown in Figure 13 (b) The temperature should be displayed along with the percentage.

In this way, when the identification of the step ``Identification is completed'' at J3 in FIG. (Step G
). The indexed C of this surface index is J at electron 1:1 time 1,
(J reads out the original limb search data stored in the external storage device and searches for data that records the relationship between d (i (j) and index of the corresponding (j) substance. row 4
F. ) 4. Only one frame of the crystal strength is also the index f・().
I decided to carry out the electronic meter C1"34 smell-C, -]-
(Automatically i-1, the gathered unity is CI'< T
10', shown in Q.

Note that the present invention is not limited to the embodiments described above, and can be modified in many ways. For example, in the above-mentioned embodiment, for each prefecture that covers the diffraction and the diffraction, calculate the quantity Therefore, we used a formula for calculating the so-called variance and a formula in FIj format, but it is also possible to calculate it using other formulas. Alternatively, after smoothing the curve representing the frequency distribution shown in Figure 2, calculate the slope of the curve 4C71, and use J,
4 [Effect] According to the present invention, the radius of each diffraction ring can be automatically determined. -C1 Based on this radius, the data to be searched can be retrieved.In particular, the sample can be analyzed. Therefore, even if you have no knowledge or experience for C1 analysis, and without having to do uncomfortable mental work, you can perform the analysis of Exam J31).

[Brief explanation of the drawings] Figure 1 shows a comparison between a case where the diffraction image is spot I-shaped and a case where the diffraction image is ring-shaped. Fig. 2 is a comparison of the detection power valve box of the pixel of J, where the diffraction image is a mixture of spot 1 and ring 2. Third
Figure (Mao ('fr! box for implementing the invention -1
Diagram to show 911, body) 4 diagram automatically tries 1 to 1
1-1 shows a break in the processing flow for constant j 3121 is a flowchart for determining the d value when the diffraction image is spot-like. Figure 8 (
, 1 pixel nl shadow operation (157 Diagram for explaining the bullet hole created, Figure 9 1jl Explaining the process of calculating the distance from center slot 1 ~ to the ring Diagram for covering, 7110
The figure shows the case where the diffraction image consists of spots and rings.
A diagram to explain the flow of the first step of calculating i+,
Fig. 11 is a diagram for illustrating the projection intensity when the diffraction 12; consists of spots (- and rings), and Fig. 12 explains the searched data created from the original wave search data and the ji searched data. Figure 13 is a diagram to show an example of the d31 pro display of the analysis results on CR King. 1: Electron microscope, 2: Image carrier, 3: AD converter, /'
1: Electron calculator, 5: Transmission electron detector, 6: AD converter, 7: External storage device, 8: CRT for monitor, 9: Keyboard, 10: OR, T for -1 channel. Patent applicant Foundation Electric Power Central 1ill Research Institute Representative SeifLl? i'1 IE1 Honden Electronics Co., Ltd. Phantom Representative 1)) l Attachment - Grand Agent Patent Attorney Keiji Tsukishima Figure 1 Figure 3 Figure 4 Figure 5 Deception Figure C

Claims (1)

[Claims] Converting the electron beam diffraction image to electrification 8, converting the electron beam diffraction image to gon/moki: x, y'
i (l IIi: Take the strong electric current of Ruru's drawing scale, Jxy and Tsutsugi, ΣUxy and ΣUxVh'X are maximized +1 Find Q and vo - (center By determining the distance vllI to the center surface where the strong one-decay distribution of both surfaces is extremely thick on a straight line passing through 1f between lattice planes
i, ·) value set for the lattice spacing of various materials 1+l! I
A method for automatic analysis of electron diffraction images characterized in that it attempts to identify (1) (1) by comparing it with data representing (1)