GB2179472A - Photolithographic mask prepared using a computer-generated original - Google Patents

Abstract

A photolithographic mask is made by a computer generated two-dimensional array of light and dark pixels being photoreduced to provide the mask in which the pixels remain resolved. In use the mask is imaged on to a workpiece such that individual pixels are not resolved, thus providing a varying grey scale which closely approximates to a true grey scale. Problems with non-linear response of photographic materials to illumination intensity are avoided. The pixel array is divided up into resolution cells or pixel sub-arrays each providing a respective optical transmission coefficient. The computer assigns each cell a respective pixel pattern appropriate to the respective transmission coefficient. Pixel patterns may be prearranged and stored or computed using a random number generator. Transmission coefficients may be input to the computer as data, or alternatively computed from a mathematical function where the required mask transmission profile corresponds to such a function.

Description

SPECIFICATION Photographic masks This invention relates to photolithographic masks having varying grey scales and to methods for making them.

The production of scales or shades of grey in photolithographic processes is known. Illustrations in newsprint for example are produced in shades of grey derived from a dot pattern of varying density. It is, however, very difficult to produce an accurately controlled grey scale variation over a photolithographic mask because of the non-linearity of photolithographic material response to illumination intensity.

It is an object of the invention to provide an alternative method of manufacturing grey scale photolithographic masks.

The present invention provides a method of making; a photolithographic mask including the steps of: (1) arranging a computer to produce two-dimensional pixel array information corresponding to the required optical transmission properties of the mask, and (2) reproducing the pixel array to provide a photolithographic mask in which individual pixels are resolved.

When used in a photolithographic process, the mask is imaged onto a workpiece in a slightly defocussed manner so that individual pixels are not resolved. This provides the advantage that a mask is produced simply and rapidly with well-resolved pixels and accurately known transmission properties. Moreover, in use the defocussed image is equivalent to that produced by a mask having an actual grey scale. The problems of phootlithographic non-linearity of grey scale with respect to illumination intensity are avoided.

The pixel array may be reproduced photographically from a computer display or printout.

Alternatively, it may be reproduced in the form of a mask by a numerically controlled machine positioning mask defining means adjacent successive mask pixels. The mask defining means may be a finely focussed light beam or a spark erosion electrode for example.

The computer may produce the pixel array as an assembly of individual resolution cells each consisting of a respective pixel sub-array. Each resolution cell is then furnished with a respective pixel patten corresponding to the optical transmission coefficient required for that cell's position in the photolithographic mask.

Transmission coefficient data may be input to the computer as a function of cell position in the mask. Alternatively, the computer may compute transmission coefficients from a mathematical function specifying the mask transmission profile.

Pixel patterns each corresponding to a respective transmission coefficient may be prearranged and stored in the computer for subsequent selection as appropriate. They may alternatively be computed using a random number generator.

Pixel arrays produced by the computer may be replicated to provide a mask having a plurality of such arrays and suitable for making device arrays.

The invention also extends to photolithographic masks made as previously indicated and to processes employing such masks.

In order that the invention might be more fully understood, an embodiment thereof will now be described by way of example only, with reference to the accompanying drawings, in which: Figure 1 is a flow diagram of computer operation, Figure 2 illustrates use of a mask of the invention in photolithography; Figure 3 illustrates resolution cell pixel patterns for providing prearranged transmission coefficients; and Figure 4 shows an example of a mask pixel pattern suitable for producing a lens.

Photolithographic optically transmissive masks are produced in accordance with the invention as follows. Such a mask is required to have a variation in grey scale over its surface. This variation corresponds to the light transmission characteristics required in, for example, a photochemical etching process using the mask. Such a process is described by Ostermayer et al.

Appl. Phys. Lett. 43(7), 1st October 1983 (Reference 1). The photochemical etching rate is directly proportional to light intensity. The steps in making the mask are as follows: (1) The light intensity variation required over the workpiece surface is determined. In photolithographic semiconductor processing for electronic or other devices, the variation is defined by the degree of etching required at each point on the workpiece and is obtainable from the device design requirements.

(2) A computer is employed to generate a graphics display or print-out consisting of ON or OFF pixels. This produces the discrete pixel equivalent of a varying grey scale corresponding to the light transmission characteristics required for processes such as photochemical etching.

(3) A photographic negative is produced from the computer output pixel pattern.

(4) Finally, the negative is photoreduced to a size appropriate for use as a photolithographic mask. At this stage the individual pixels remain well resolved as individual clear or dark regions.

The mask is employed in a photolithographic process involving an optical system to image the mask on to the workpiece. The image is slightly off focus so that individual pixels are not resolved. The advantage of this procedure is that it is relatively straightforward to produce a well-resolved pixel pattern which accurately reproduces predetermined optical transmission characteristics within digitisation limits. High contrast techniques may be used. It is, however, photolithographically very difficult to produce an accurate grey scale in a smoothly varying or analogue manner. In particular, accuracy of reproduction is reduced by temperature variation and compositional non-uniformity over the photographic film.

The procedure for producing the computer-generated pixel pattern will now be described in detail. A Hewlett-Packard 9845 computer with graphics software and a graphics printer is used to generate pixel patterns. The computer has an available plotting area of 560X455 addressable pixels. In this example a square area of 448X448 pixels is employed. The area is subdivided into 56X56 resolution cells or pixel sub-arrays each containing 8X8 pixels. A resolution cell is the smallest unit of pattern definition, and is used to define the optical transmission required for the eventual mask in the latter's corresponding region. The average grey scale provided by a cell is determined by the ratio of clear to dark pixels in that cell.

It is required to produce a photolithographic mask in which the individual pixels are sufficiently close to avoid their being resolved when imaged on to a workpiece. To achieve this, a minimum of 8 and a maximum of 56 pixels per 64-pixel resolution cell are arranged to be dark. A choice of 49 optical transmission factors will accordingly be available.

For each of the 49 cell transmission factors, a respective pixel pattern is required to give an unresolved effectively constant grey scale when a mask is used. Such patterns are predesigned and stored in the computer. Patterns 1 to 24 are equivalent to 26 to 49 with light and dark pixels exchanged. The computer is arranged to output the relevant stored pattern for pattern number p from 1 to 25. For from 26 to 49, the computer outputs the pattern for 50-p with light and dark pixels interchanged. Each pixel is located relative to its cell by x and y coordinates each in the range 1 to 8. This requires the storage of (7+p) pairs of dark pixel coordinates for each p=1 to 25, is 500 pairs. Each coordinate has 4 bits, so storage is required for 103 4-bit numbers.

In electronic device manufacture, the variation in illumination intensity required over the device surface in a photolithographic process step is known or can be calculated from device design data. In the present example, 56X56 or 3136 cell transmission factors are required. These are obtained from design data and fed into the computer. The latter assigns each cell in turn the respective pixel pattern appropriate to its required transmission factor. Having carried this out for all cells, the computer prints out or displays on a screen the resulting full 448 by 448 pattern of light and dark pixels. The full pattern is then photographed and photoreduced. This provides a transmission mask in the form of a photographic negative of the appropriate size for photolithography.

Referring to Fig. 1, a flow diagram 10 of computer operation is shown, this being compatible with the Hewlett-Packard computer and associated graphics package previously described. The mask size is first defined at 12, 448X448 pixels in the present example. This is followed at 14 by setting the 8X8 pixel array size of each resolution cell. At 16, cell transmission data is input for each cell in succession. The cell number n is set to zero at 18, and incremented by 1 at 20.

At 22, cell number n is assigned from computer memory the respective pixel pattern appropriate for its transmission factor specified by the data input at 16. Pixel-in-cell coordinates x and y are converted at 24 to pixel-in-array coordinates X and Y from the expressions:

where

in the integral part of (n-1) 56 The pixel values (light or dark) are then plotted at (X, Y) for each (x, y). These expressions assume that the computer begins at the extreme left lowermost cell for which n= 1, n= 1 to 56 provides the first lowermost row, n=57 to 112 the second row and so on up the array. In addition the equivalent approach is assumed for pixels in a cell; ie x=y= 1 corresponds to the lowermost extreme left cell and x=y=8 the uppermost extreme right.This is the approach used in the computer graphics package previously referred to.

Cell number is tested at 26 for equality to 3136, the final cell number, if not3136, steps 20 to 26 are repeated by virtue of loop 28 until n=3136, when print out of the entire pixel array occurs at 30.

Referring now to Fig. 2, an example is shown of use of a mask produced in accordance with the invention.

A lamp of 40 illuminates a mask 42 of the invention via a condenser lens 44. The mask 42 is imaged on to a workpiece 46 by a projection lens 48, and is slightly off forcus to a degree sufficient to avoid individual pixels being resolved. The workpiece 46 is mounted on a rubber 0ring 50 inset (not shown) into a glass cell 52 containing photochemical etchant fluid (not shown). The fluid communicates with the sample 36 via a hole 54 in the cell 52. Clamping means (not shown) hold the sample 46 against the O-ring 50 forming a seal.

Referring now to Fig. 3, there are shown enlarged views of four adjacent 8X8 resolution cells 60a to 60d; these have 8, 12, 22 and 42 dark pixels corresponding to transmission factors of 7/8, 13/16, 21/32 and 11/32 respectively.

Fig. 4 shows a mark pixel pattern 70 produced in accordance with the invention and suitable for producing a convex spherical profile lens. Although photoreduced from a computer print-out, the pattern is not in the final mask size which would be much smaller. In reference 1, the lenses are shown to be about 60 microns in diameter (see Fig. 3). The pattern 70 is about 15cm from top to bottom, so would be reduced by a further factor of 2500 to produce a 60 micron lens.

This reduction would be done largely photographically, but to an extent also by image reduction in the Fig. 2 optical system.

For some specific applications of the invention, the required mask may have a transmission coefficient T which varies over its surface in accordance with a known mathematical functon. In particular, Reference 1 describes the making of an array of light emitting diodes furnished with a respective lens. A photochemical process is used in which the etch rate of the material in contact with the etchant is directly proportioned to the illuminating intensity. In making an individual lens, the required lens profile for a light transmission mask might be cosine, spherical, elliptical, hyperbolical, Gaussian etc. It would then be unnecessary to input to the computer data on cell transmission properties.Each mask may be arranged to provide a single lens; the transmission factor for each resolution cell is then calculated by the computer from the cell's position in the array by inserting the cell number in the formula. In particular, for a convex spherical lens, using previously defined parameters, mimimum transmission is required at the centre of the mask for minimum etching, and maximum at the outer edges. Taking the centre of the mask at X=Y=224, mask transmission factor T at pixel coordinates (X, Y) for a convex spherical lens is given by:

The terms 81 and l in the above expression normalise Tc, to extreme values si and -, as required for the minimum and maximum numbers of pixels per cell to be 8 and 56.Similar expressions may be employed to calculate Gaussian or other profiles. The Fig. 1 computer flow diagram becomes modified by amending box 16 to read "Input etch profile function". In addition boxes 22 and 24 are replaced by the following four: Convert each (x, y, n) to (X, Y); Calculate T from profile function; Look up pixel pattern for T in memory; Plot each pixel at a respective (X, Y).

Reference 1 describes the production of an array of lenses. This can be carried out in accordance with the invention by replicating the single lens mask as necessary. This is done by exposing successive small regions of a sheet of photographic material to the pixel array image until the required array mask is built up. The image position is moved between exposures by optical scanning apparatus well known in the art of optics. The overall image movement may be equivalent to a raster scan to build up a mask for a two-dimensional array of lenses. When appropiately photoreduced, the pixel resolution in the mask may be very small, in the order of micron dimensions for example. A similar procedure may be used to produce arrays of other devices.

It is to be noted that the cell pixel patterns chosen for a particular mathematical function or profile are not necessarily optimum for a different function. A cosine profile for example would preferably employ a different pattern set to a spherical profile. To avoid the need to generate and store large numbers of pattern sets, a random number generator could be employed as follows. The generator would produce pairs of numbers each in the range 1 to 8 subject to the constraint that the same pair were not to be selected twice. For each cell transmission factor a previously known number of pixels would be required to be dark, and these would be assigned coordinates by the random number generator. Computational constrainst could be inserted to avoid undesirable clustering of dark or light pixels.Such a generator has been found to produce adequate results from a number of mathematical expressions. It is not as good as a predesignated pattern set which is optimum for a given expression. It is, however, superior to a pattern set inappropriate to an expression.

Although the invention has been described in terms of photographic reproduction of a pixel array to produce a photolithographic mask, non-photographic methods of mask-making may also be employed. These will now be described.

The pixel array data from the computer is output to a numerically controlled mask making machine, instead of being displayed or printed out and photoreduced as previously described.

The machine incorporates a finely focussed light beam, such as a laser beam for example. The beam is controlled by the machine to expose the final photolithographic mask directly. This produces the mask (when developed) as an array of opaque and transparent pixels or spots. The final photographic development stage could be omitted if a laser were to be employed to burn holes in the mask. The mask produced by either of these methods is virtually indistinguishable from that generated by the process described with reference to Figs. 1 to 4. The mask differs from that described earlier in that it is exposed pixel by pixel by the light beam, the latter being controlled directly by the machine from computer output data.

In a further embodiment, a procedure equivalent to that previously described is employed, except that a spark erosion technique is used to make the mask. The light beam is replaced by a very fine spark erosion electrode, and an equivalent numerically controlled machine is employed to position the electrode at successive pixels for erosion to the required degree in each case. The mask is in the form of a metal layer evaporated on to a glass substrate, and erosion is implemented by striking an arc between the electrode and a point on the metal layer positioned nearest to the electrode. The spark opens a hole in the metal layer, producing a transparent pixel. Pixels required to be opaque are left uneroded.

Further embodiments of the invention employ computer generated data fed to a numerically controlled positioning machine arranged to position means for defining a mask. The mask defining means may be a fine stylus or an array thereof, or a very fine ink jet printer.

In the foregoing examples, an additional photoreduction procedure may be employed if the mask or pixel pattern produced by the numerically controlled machine is too coarse for the intended use. In all methods of mask making in accordance with the invention the individual pixels remain resolved in the final mask. The mask itself is in all cases employed in a photolithographic process to produce an image in which individual pixels are not resolved. This may be achieved by defocussing the image slightly, or by reduction of the image to beyond the resolution limit of individual pixels.

Claims (12)

1. A method of making a photolithographic mask including the steps of: (1) arranging a computer to produce two-dimensional pixel array information corresponding to the required optical transmission properties of the mask; and (2) reproducing the pixel array to provide a photolithographic mask in which individual pixels are resolved.

2. A method according to claim 1 wherein the pixel array is reproduced by photoreduction of a computer output display or print-out.

3. A method according to claim 2 wherein the pixel array is reproduced by a numerically controlled machine arranged to- receive pixel array information and to control the positioning of mask defining means in accordance with the information to reproduce pixels in succession.

4. A method according to Claims 1, 2 or 3, wherein the computer is arranged to: (1) generate the pixel array as an assembly of individual resolution cells or pixel sub-arrays; and (2) furnish a respective transmission co-efficient required for that cell's position in the photolithographic mask.

5. A method according to Claim 4 wherein the computer is arranged to receive input transmission co-efficient data as a function of resolution cell position in the mask.

6. A method according to Claim 5 wherein the computer is arranged to compute transmission co-efficients from a mathematical function.

7. A method according to Claim 4, 5 or 6 wherein the computer is arranged to store a set of pixel patterns each corresponding to a respective transmission co-efficient.

8. A method according to Claim 4, 5 or 6 wherein the computer is arranged to provide pixel patterns derived from random number generation.

9. A method according to any preceding claim wherein the pixel array is replicated to provide a photolithographic mask having a plurality of pixel arrays.

10. A photolithographic mask produced by the method of any preceding claim.

11. A photolithographic process employing the mask of Claim 10 wherein the mask is imaged on to a workpiece such that individual pixels are not resolved.

12. A method of making a photolithographic mask substantially as herein described with reference to the accompanying drawings.