US3900734A

US3900734A - Scanning electron-beam instrument

Info

Publication number: US3900734A
Application number: US462295A
Authority: US
Inventors: David Kynaston; Peter Irving Tillett; Richard Stephen Paden
Original assignee: Cambridge Scientific Instruments Ltd
Current assignee: Cambridge Instruments Ltd
Priority date: 1973-04-19
Filing date: 1974-04-18
Publication date: 1975-08-19
Anticipated expiration: 1992-08-19
Also published as: DE2418279A1; JPS5031770A; DE2418279C2; GB1441824A

Abstract

In a scanning electron beam instrument such as a scanning electron microscope, including a variable-focus final lens following the scanning coils and consequently causing rotation of the scanning raster, electronic circuitry is incorporated between time base generators that generate the line and frame scanning signals and the scanning coils themselves, this circuitry mixing the signals according to known principles but being coupled to means controlling the final lens excitation so that the raster orientation at the specimen surface is independent of working distance and furthermore the circuitry includes attenuating means (or amplifying means) in at least one of the scanning signal channels such as to allow correction of distortion of the image introduced by tilting the specimen surface, this correction being consequently unaffected by changes in working distance.

Description

United States Patent 11 1 Kynaston et al.

$746,855 7/l973 Hilditch 250/396 3,753.034 3/!973 Spicer .4 250/396 LT w ANALOGUE MULTlPLIER FT 1 M2 muocue MULTIPUER ANALOGUE uuuwusa M3 sin 9 SCANNING ELECTRON-BEAM INSTRUMENT Inventors: David Kynaston; Peter Irving Tillett;

Richard Stephen Paden, all of Cambridge, England Assignee: Cambridge Scientific instruments Limited, Cambridge, England Filed: Apr. 18, 1974 Appl No.: 462,295

Foreign Application Priority Data References Cited UNITED STATES PATENTS cos 9 ANALOGUE MULTIPLIER Aug. 19, 1975 Primary Examt'nerJames W. Lawrence Assistant Examiner-T. N. Grigsby Attorney, Agent, or Firm-Scrivener Parker Scrivener and Clarke 57] ABSTRACT in a scanning electron beam instrument such as a scanning electron microscope, including a variable focus final lens following the scanning coils and conse quently causing rotation of the scanning raster, electronic circuitry is incorporated between time base generators that generate the line and frame scanning signals and the scanning coils themselves, this circuitry mixing the signals according to known principles but being coupled to means controlling the final lens excitation so that the raster orientation at the specimen surface is independent of working distance and furthermore the circuitry includes attenuating means (or amplifying means) in at least one of the scanning signal channels such as to allow correction of distortion of the image introduced by tilting the specimen surface, this correction being consequently unaffected by changes in working distance.

2 Claims, 5 Drawing Figures ATTENUATOR R R PATENTEI] AUG] 9 I975 3. 9 00 T 34 SHEET 1 OF 2 FIG. 1 PRIOR ART C Y A W L r Rotation Lcos+Fsln Mi/ F|G.3, PRIOR ART PATENTEU mm 9 I975 SHIT 2 BF 2 mwou o SCANNING ELECTRON-BEAM INSTRUMENT a slower frame scan in a manner very like a television.

raster. Looked at in the Z direction (i.e. along the electron-optical axis) the raster has a certain orientation in relation to that axis at the region of the scanning coils, and in the straightforward case the X and Y directions, i.e. the line and frame deflections, coincide with the axes of the respective pairs of scanning coils, although there is known, and indeed we have used, deflection circuits which allow the X and Y directions to be rotated through any angle up to 360 about the Z-axis, while maintaining the orthogonal relationship between the line and frame vectors.

However, regardless of the orientation of the raster at the regions of the coils, this raster will undergo rotation about the electron-optical axis before it reaches the specimen, by virtue of the action of the electromagnetic final lens. The amount of rotation will depend on the excitation of the lens, which in its turn determines the focal length. Thus the rotation will depend on the spacing of the specimen along the axis from the exit pole piece of the lens, as this spacing determines the focal length to which the lens has to be adjusted.

Where the surface of the specimen under examination is normal to the electron-optical axis (which we will call the Zaxis), the rotation of the scanned raster caused by this final lens effect is inconvenient as it upsets the alignment of the specimen stage movement (the specimen stage being normally movable in two orthogonal directions, i.e. along an X and a Y axis with the line and frame scanning vectors.

Where the rotation is more serious, however is in the case of a tilted specimen. It is common in electron microscopy to incline the scanned specimen surface at an angle to the beam (the Z-axis) other than 90, chiefly in order to obtain an improved secondary electron signal or X-ray signal from a detector necessarily placed on one side of the axis. This introduces distortion in the scanned raster due to foreshortening of the dimension that coincides with the steepest slope (i.e. in a direction on the specimen surface lying in a plane containing the Z-axis and the normal to the surface at the point of impact). Where either the X direction or the Y direction (i.e. the line or the frame scan) of the raster coincides with this direction there is no great problem as the appropriate deflection signal can be multiplied by a correction signal dependent on the cosine of the angle of inclination. However this is only true for one value of the focal length of the final lens. At other values the raster is rotated more or less and so the distortion will produce a parallelogram or rhombus type of distortion of the raster.

According to the invention in its broadest aspect the problem of distortion resulting from inclination of a specimen surface is overcome by introducing a correction for the final lens effect for this purpose which is independent of any provisions that may or may not be present for rotating the scanning raster at will under control of the operator.

This correction of the distortion is achieved by applying an appropriate attenuation factor electronically along the direction of steepest slope regardless of the orientation of the raster in relation to that direction (or by applying an amplifying factor in a direction across the slope) and at the same time making orientation of the raster automatically independent of the working distance.

The invention will now be further described by way of example with reference to the accompanying drawings, in which:

FIG. 1 illustrates diagrammatically the scanning coils, final lens and specimen in a scanning electron beam instrument;

FIG. 2 is a diagrammatic view looking along the Z- axis (the axis of the electron beam);

FIG. 3 shows diagrammatically a known correction circuit;

FIG. 4 is a diagram that illustrates the distorting effect of tilting the plane of the specimen surface with respect to the Z-axis', and

FIG. 5 shows a circuit in accordance with the present invention.

For simplicity FIG. I shows only the final part of the beam-forming system of an electron probe instrument which may be for example a scanning electron microscope or a scanning X-ray micro-analyzer. A beam B of electrons is brought to a final focus on a specimen S. the surface of which is to be examined, by a final electromagnetic lens shown diagrammatically in section at L. To keep aberrations small, the focal length of the final lens (and therefore the distance from its lower pole to the specimen surface) should be as short as possible but it has been exaggerated in FIG. I in the interests of clarity.

In a scanning instrument the beam is caused to scan back and forth in a raster, usually like a television raster, that is to say in a number of lines making up a rectangular area, by means of line scanning and frame scanning coils FS and LS mounted in the back bore of the lens L. In practice, for reasons which need not concern us here, there are two sets of coils for each scanning direction, spaced apart along the axis of the beam (which we will call the Z axis), deflecting the beam first one way and then back again but for the purposes of the present invention we can treat each pair of axially spaced coilss as a single coil.

In the simplest case the line scanning coils LS will deflect the beam in one direction perpendicular to the Z axis, which we will call the X axis, and the frame scanning coils will deflect the beam in a third direction, the Y axis, perpendicular to the other two. For this purpose the coils are supplied with sawtooth timebase signals, from line-scan and frame-scan timbase generators LT and FT.

The same generators also control the X and Y deflections of the beam of a cathode ray rube CR or equivalent recording device, the brightness of the beam being controlled by a detector D receiving secondary electrons, reflected electrons, or X-rays emanating from the point of impact of the electron beam B on the specimen surface, so that there appears on the screen of the cathode ray tube CR a rectangular image, corresponding to the scanned area of the specimen surface, of

which the contrast is representative of the variations in the nature or topography of the specimen surface over that area.

The specimen S is usually mounted on a specimen stage (not show) that is movable along two mutually orthogonal axes in a plane parallel to the specimen sur face so as to allow different regions of the surface to be brought under the beam. For convenience these axes should coincide with the directions of the X and Y axes at the specimen surface. However the orientations of the X and Y axes at the specimen surface are not the same as those of the coils FS and LS because the focus sing action of the final lens L causes the electrons to move in a spiral-helical path that effectively rotates the whole raster about the Z axis between the coils and the specimen surface, If the working distance, i.c. the distance of the specimen surface from the final lens is fixed, and therefore the excitation of the lens is constant, it is possible to allow for this by correctly orientating the scanning coils FS and LS in relation to the specimen stage movements when initially setting up the instrument. However in practical instruments it is desirable to be able to alter the working distance.

FIG. 2 shows a raster C and a raster C rotated counterclockwise with respect to the raster C through an angle 4) about the Z axis. Now it is known that the coordinates of any point P with reference to one set of axes X, Y can be represented in the other set of coordinates (xfly'), where Therefore it is known to mix the signals (of magnitude L and F) from the two timebase generators LT and Fl to produce new signals L' and F where F F cos d; L sin (b and a suitable known circuit is shown diagrammatically in FIG. 3. The value of the angle d may be manually controlled, allowing the operator to rotate the raster as much as he likes, so that the line scan is in any direction he selects. This allows him to orientate the raster with the specimen stage movements (or with any other feature) without moving the positions of the scanning coils themselves.

Where rotation is a more serious problem, however, is where the plane of the surface of the specimen S is no longer perpendicular to the axis of the beam B. This tilting of the specimen is often desirable in order to increase the signal in the detector D. It introduces distortion in that, if the dimensions of the raster are left unchanged, that dimension of the raster which is along the line of steepest slope (i.e. in a direction on the specimen surface lying in a plane containing the axis of the beam and the normal to the surface at the point of impact) is foreshortened and the resulting image (which is still in itself rectangular) on the screen of the cathode ray tube CR is distorted. If either the X direction or the Y direction (i.e. the line or frame scan) of the raster at the specimen surface coincides with this direction there is no great problem as the appropriate deflection signal can be multiplied by a correction signal dependent on the cosine of the angle of tilt.

However this is only true for one given working distance. As soon as the distance of the specimen from the final lens is altered (and therefore the excitation of the lens is also altered) the raster is rotated.

lt is true that the operator could rotate the raster back into alignment with the slope. but he is then no longer free to rotate the raster to take account of other features. Any subsequent rotation will introduce distortion, which will be ofa kind making a given scanned region, intended to be square or rectangular, a rhombus or kite shape. lt will be appreciated that the image on the cathode ray tube CR will still be square or rectangular, but will be a distorted representation of a rhombus-shaped or kite-shaped scanned region of the specimen surface.

FIG. 5 shows a circuit according to the invention which corrects this electronically by introducing a correction that remains true regardless of the orientation of the raster in relation to the line of steepest slope and regardless of changes in the working distance.

Referring to FIG. 5, the line and frame signals L and F from the saw-tooth timebase generators LT and FF are fed unchanged to the deflection circuits of the cathode ray tube CR as before. For the scanning coils LS and FS the signals are first mixed, in the manner described above, with each other and with signals of variable magnitude sin 0 and cos 6 to produce derived signals L and F where:

F'=Fcos0Lsin6 This is done in a manner that need not be described in detail as the necessary circuits elements are readily commercially available and comprise analogue multipliers Ml-M4 to form the factors L cos 6, F sin 6 and so on, and sum and difference amplifiers S and DI respectively to combine the factors appropriately. For example operational amplifiers such as those sold by Texas Instruments under the description 702 may be used. The value of 6 is under the control of the operator and can be made adjustable throughout the range from 0 to 360.

These two derived signals L and F' are fed to separate independently adjustable attentuator networks A. They are then again mixed, this time using operational amplifiers in a slightly different way. The signal L is fed to the input of a summing amplifier SA! through a resistor of magnitude R/cos d), the feedback resistor of the amplifiers being the value R, so that the signal is effectively multiplied by cos (b, the derivation of which is described below. Similarly the signal F is fed to the summing input of the same amplifier through a resistor of value R/sin 12. The output of the amplifier is therefore L" where:

L" L cos qb-i- F sin d) Similarly the inverted value of L (i.e. L') obtained from an inverting amplifier 1A is fed to a second summing amplifier SAZ together with the signal F, again through input resistors of appropriate value, to produce an output F" where:

F" =F' cosd -L dz These two outputs L" and F" are fed to the scanning coils LS and FS respectively.

Each of the input resistors of value R/cos d), and R/sin d) of the two summing amplifiers is not a single resistor but one of a number of resistors of different values, corresponding to different values of the angle (it, and selected by a multi-way switch ganged to a switch that selects different values of the excitation current of the final lens L. The values are arranged so that for any value of the excitation current the correction introduced by the summing amplifiers SA] and 8A2, i.e. the

derivation of the signals L" and F" from the signals L and F, is just the amount required to cancel out the rotation d) of the raster introduced by the final lens L. In a typical example there are twenty-four separate values capable of being selected for each of the input resistors. In an alternative arrangement, with appropriately wound potentiometers the variation could be stepless. although the winding of the potentiometers would be difficult, bearing in mind that the variation of focal length with excitation is non-linear.

The angular position of the raster is therefore independent of the working distance. The specimen stage has its X and Y movements aligned with the line and,

scan directions of the coils LS and FS during the initial settingup of the machine. If now the specimen is tilted with its line of steepest slope either in the X direction or the Y direction. the distortion introduced by the tilting can be corrected empirically by adjustment of one or the other of the attenuators A (or instead of attentuators one could use amplifiers).

Once this has been done, and if the attenuators A are left undisturbed thereafter, it will be found that the angle 8 can be altered as much as one likes, i.e. the raster can be rotated with complete freedom, without affecting orthogonality of the raster, and so the image on the screen of the cathode ray tube CR (which has remained rectangular throughout anyway) remains a true and undistorted image of the scanned region on the specimen surface. This is unaffected by changes in the working distance.

It is true that some trapezium distortion is also introduced by the tilting of the specimen but this can also be corrected, if desired, by other means, which are not shown and do not form directly part of the present invention. and which are based on automatically attenuating the across-the-slope component of the signal by an amount which varies with the magnitude of the down-the-slope component.

If, for any reason, the angle of tilt of the specimen is altered, or if the direction of tilt is altered, distortion is introduced but again can be corrected by temporarily turning the raster (by control of the angle 6) until the X axis or the Y axis coincides with the direction of steepest slope and then re-adjusting one or both attenuators A. Again. once this has been done. the user can therefore turn the raster to any position he desires without introducing any distortion.

We claim:

1. A scanning electron beam instrument comprising means for forming a fine probe of electrons along an axis, a variable-focus electromagnetic final lens acting on said probe, line and frame time base generators capable of generating line and frame scanning signals. line and frame scanning coils disposed to act on said probe ahead of said final lens whereby to cause the formation of a raster on a specimen placed at the focus of said final lens, variable control means controlling the focal length of said final lens, signal mixing means connected to said frame and time base generators and operative to mix said signals and feed derived line and frame scanning signals to said line and frame scanning coils respectively, said derived signals being such as to rotate said raster about said axis as compared with a raster produced by said first-mentioned signals, an interconnection between said signal-mixing means and said variable control means such as to vary the rotation of said raster by said signal-mixing means in opposition to rotation of said raster introduced by said final lens, whereby the orientation of said raster at the focus is independent of the focal length of said final lens at least over a range of focal lengths, and relative attenuating means interposed between at least one of said time base generators and said signal mixing means, capable of varying the dimension of said raster at the focus in a given direction transverse to said axis.

2. The instrument set forth in claim 1 including second signal-mixing means interposed between said time base generators and said first-mentioned signal-mixing means ahead of said relative attenuating means, said second signal-mixing means being manually controllable to vary the orientation of said raster independently of the focal length of said final lens.

* i I. f

Claims

1. A scanning electron beam instrument comprising means for forming a fine probe of electrons along an axis, a variable-focus electromagnetic final lens acting on said probe, line and frame time base generators capable of generating line and frame scanning signals, line and frame scanning coils disposed to act on said probe ahead of said final lens whereby to cause the formation of a raster on a specimen placed at the focus of said final lens, variable control means controlling the focal length of said final lens, signal mixing means connected to said frame and time base generators and operative to mix said signals and feed derived line and frame scanning signals to said line and frame scanning coils respectively, said derived signals being such as to rotate said raster about said axis as compared with a raster produced by said first-mentioned signals, an interconnection between said signal-mixing means and said variable control means such as to vary the rotation of said raster by said signal-mixing means in opposition to rotation of said raster introduced by said final lens, whereby the orientation of said raster at the focus is independent of the focal length of said final lens at least over a range of focal lengths, and relative attenuating means interposed between at least one of said time base generators and said signal mixing means, capable of varying the dimension of said raster at the focus in a given direction transverse to said axis.