GB2293459A - Method for printing of a pattern of features - Google Patents

Abstract

A method for printing a pattern of features that includes providing a mask defining the pattern of features, recording the pattern in a TIR hologram 9, printing the pattern from the TIR hologram into a photosensitive layer on a substrate by illuminating the TIR hologram with a scanning laser beam having a constant intensity and a constant speed, measuring the departures of the linewidths of the printed features from their desired values, and printing the pattern of features from the TIR hologram into a photosensitive layer 13 on another substrate by illuminating the TIR hologram with a scanning beam 17 and modulating the exposure energy density imparted by the scanning laser beam, either by varying the intensity of the scanning laser beam or by varying its speed, such that the dimensions of the features printed into this photosensitive layer have their desired values. <IMAGE>

Description

METHOD FOR PRINTING OF A PATTERN OF FEATURES The present invention relates generally to the field of microlithography as employed for the manufacture of electronic and other types of device comprising highresolution features. It relates in particular to microlithography based on the imaging properties of total internal reflection holography.

Most high-resolution ( < 1.5 Mm) lithographic processes for fabricating microdevices start with a mask, or a set of masks, defining the pattern of features to be formed on a substrate surface, for instance, on a silicon wafer. The pattern may be transferred to the substrate either by contact printing the mask or by imaging the mask through a lens or mirror system. The latter is often preferable, especially for large production quantities, because it avoids damage to the mask. The main shortcoming of highresolution imaging systems is that optical aberrations limit the size of the exposure field to typically 1.5 x 1.5 cm2, and consequently if a larger area has to be printed, such as an 8" diameter silicon wafer, a multi-exposure step-andrepeat procedure has to be employed.Not only is this size of exposure field restrictive for many types of device (eg.

CCDs and DRAMs) but the stepping motion requires very sophisticated and expensive mechanics in order to achieve good layer-to-layer registration and high throughput.

Device performance is dependent on how accurately the features with the smallest dimensions (critical dimensions, or CDs) can be realised and located with respect to other such features. For this reason the pattern in the mask is best fabricated using electron beam (e-beam) lithography.

However, although very precise, e-beam lithography also has its limitations. CD errors are introduced mainly by the spin processing of the mask substrate following the e-beam lithography and so the magnitude of these errors varies slowly over the mask area. An approach that is therefore commonly used to ensure high accuracy of the printed features is to fabricate the mask at a scale five times larger than the pattern required and then, during pattern transfer, to image the mask through 5x reduction optics. In this way the CD errors present in the mask are reduced to an acceptable level. Unfortunately, the optical aberrations (eg. astigmatism and coma) of reduction optics distort the features and limit CD accuracy.

An alternative lithographic technique is based on the imaging properties of total internal reflection holography Total internal reflection (TIR) holography has been demonstrated to be powerful technique for sub-micron lithography [1-3] . TIR holographic imaging is free of the off-axis aberrations of lens and mirror imaging systems and so allows near diffraction-limited resolution over an unlimited field.

The main principles of TIR hologram recording are illustrated in Figure 1. A holographic plate 1 comprising a holographic recording layer 2 on a substrate 3 is placed in optical contact with a surface of a prism 4. An object in the form of a mask 5 is located in proximity to the recording layer 2. Two mutually coherent laser beams illuminate the system. One, the object beam 6, passes through the mask 5 to the recording layer 2 and the other, the reference beam 7, is directed through another face of the prism 4 so that it is totally reflected from the surface of the holographic layer 2. The optical interference of the two beams 6 and 7 is recorded by the photosensitive material in the layer 2 to form the TIR hologram.

A more uniform recording of the mask may be obtained by scanning the object and reference beams across the mask and holographic layer.

Reconstruction of the hologram is performed by illuminating the hologram with a laser beam directed in the opposite direction to the reference beam 7 that recorded it. This generates an image that is an accurate reproduction of the pattern contained in the original mask 5. This image can be printed onto a substrate such as a silicon wafer.

A particularly advantageous procedure for printing the image from a TIR hologram onto a substrate surface is to employ a scanning technique, as disclosed in U.S. patent 4,966,428, whereby the time-integrated energy density from the illumination beam over the substrate surface is made highly uniform. Furthermore, by employing a dynamic focus system that measures the separation between hologram and wafer where exposure is taking place and adjusting the position of the wafer so that the local separation remains constant as the beam scans, the complete pattern is printed in focus, even over poor flatness wafer. This technique is described in detail in copending European Patent Application No.

90310375.2.

TIR holography, however, has one disadvantage : it is intrinsically a lx process; there is no demagnification of the mask pattern. Consequently, any CD errors contained in the original e-beam mask are accurately recorded in the TIR hologram and then transferred to the device, degrading device performance.

It is therefore an object of the present invention to provide a method and apparatus of performing lithography using TIR holography such that the CD, or linewidth, errors contained in the mask are not transferred to the device. The method may also be used to correct linewidth errors introduced by non-ideal hologram recording.

The method may also be employed to deliberately introduce a particular variation in the linewidths of the printed features over the printed area.

SUMMARY OF THE INVENTION According to the present invention there is provided a method for printing a pattern of features, which method comprises: A method for printing a pattern of features, including: a) providing a mask defining the pattern of features; b) recording a TIR hologram of the mask; c) printing the pattern of features recorded in the TIR hologram into a first layer of photosensitive material on a first substrate by illuminating the TIR hologram with a first scanning beam having a constant intensity and using a constant scanning speed; d) measuring the dimensions of at least two features in the pattern printed in the first layer of photosensitive material and determining the deviations of these dimensions from their desired values; e) printing the pattern of features recorded in the TIR hologram into a second layer of photosensitive material on a second substrate by illuminating the TIR hologram with a second scanning beam and modulating the exposure energy density from the second scanning beam so that the dimensions of the features printed in the second layer of photosensitive material have their desired values.

Preferred embodiments of the various aspects of the present invention will now be described in detail with reference to the accompanying drawings.

BRIEF DESCRIPTIONS OF THE DRAWINGS Figure 1, already described, illustrates the main principles of total internal reflection holography.

Figure 2 schematically shows a system for printing a pattern of features from a total internal reflection hologram in accordance with the invention.

Referring to Fig.2, a 6" square glass substrate 8 with a TIR hologram 9 on its lower surface is located in optical contact with the base of a 450, 450, 90 glass prism 11 by way of a layer of refractive-index matching fluid 10.

The TIR hologram 9 is a recording of an array of nominally identical device patterns each with an area 1 cm x 1 cm and a critical dimension 1 Am. The hologram was recorded according to the prior art using high-uniformity illumination beams and a separation between mask and holographic recording layer of 100 m. The recording material was a layer of HRS-352, a high-resolution, lowscatter photopolymer manufactured by Dupont de Nemours & Co.

A scanning technique may alternatively have been used to record the hologram in order to further improve the uniformity of the recording.

The orientation of the hologram substrate 8 on the prism 11 is such that a beam normally incident on the hypotenuse face of the prism 11 illuminates the hologram 9 in the reverse direction to the reference beam that recorded it.

A test wafer 12 coated with a 1.2 m layer of photoresist 13 is loaded onto a vacuum chuck 14 located below the hologram.

The parallelism between the layer of photoresist 13 and hologram 9 and their separation are measured, and the chuck 14 adjusted using stepper motors 15 until the photoresist layer 13 is parallel to the hologram 9 and at a distance of 100 m from it. Measurement of the degree of parallelism and separation between hologram 9 and photoresist layer 13 is preferably obtained interferometrically by, for instance, introducing a laser beam or laser beams through the vertical face of the prism 11 so that they are totally internally reflected off the hypotenuse face of the prism 11 and illuminate the hologram 9 and wafer 12 at normal incidence, and then detecting the interference of the beam reflected from the hologram 9 with the beam reflected from the photoresist layer 13.In order that the reconstructed image is printed accurately in focus over the entire wafer surface irrespective of the flatness of the wafer, it is further desirable that the separation-measurement laser beam can scan simultaneously with the illumination beam so that a local measurement of the separation can be made and the gap continuously adjusted to a constant value using the stepper motors 15. (This gap measurement apparatus is however not shown in the figure since it is well described in the background art. Multiwavelength phase-step interferometry is a particularly useful technique for determining the separation. See for instance EP-A-02421645 and reference 6).

The beam 17 for reconstructing the hologram 9 is produced by an argon ion laser 18 operating at a wavelength of 363.8 nin.

The laser system has a computer interface card 19 allowing the output power of the laser to be controlled from a computer 20 via an RS232 link 21.

With the shutter 22 open, the beam from the laser 18 passes through a beam expander 23 producing a collimated beam with a Gaussian intensity profile and diameter -1.5 cm (at the l/e2 intensity points). The beam is deflected by a mirror 24 so that it travels parallel to the x axis. Whereas the diagram shows, for the reason of ease of illustration, the output of the laser 18 to be travelling in the z direction and the mirror 24 to be oriented such that the beam is directed in the x direction, in actual fact the output of the laser 18 is in the y direction and the mirror 24 is oriented appropriately to deflect the beam in the x direction.

The beam is incident on a scanning system 25 that is controlled by the computer 20 and is configured so as to generate a raster scan of the laser beam across the mirror 30 and thence the hologram 9. The scan passes of the raster pattern are produced by the stage 26 travelling in the x direction with a speed of -5 cm/s whilst the steps between scan passes are produced by the stage 27 that translates the stage 26, input mirror 24 and output mirror 28 in the y direction in steps of -3mm (so providing a high uniformity of exposure energy). During the scan the computer 20 knows the positions of the stages 26 and 27 at all times.

The pattern recorded in the TIR hologram 9 is printed into the layer of photoresist 13 on the wafer 12 by exposing the hologram 9 to the scanning laser beam 17. During the exposure, the intensity and speed of the beam 17 are kept constant and the intensity level is such that subsequent development of the photoresist causes the printed features to be etched through to the underlying wafer 14.

Following development of the photoresist layer 13 the printed pattern is analysed using either a scanning electron microscope or an atomic force microscope. In this analysis, an imaginary grid is laid over the wafer surface such that each device is centred on a grid square. A CD measurement is performed at the same relative place (for instance, at the centre) in each of the grid squares. (Alternatively, a number of CD measurements could be made in each of the grid squares and the measurements averaged to give a more representative value). The resulting "map" of CD measurements is then entered into the computer 20, the values being stored in a dedicated array.

Determination of the CD errors in the pattern recorded in the hologram may alternatively be determined directly from the original photomask. The drawback of this approach however is that it does not take account of any additional linewidth errors that may be introduced during the hologram recording process.

The computer 20 is also provided with data describing the dependence of CD on exposure energy. Such data can be obtained either from the photoresist manufacturer or can be empirically determined by carrying out standard process latitude tests.

A second wafer bearing a 1.2 pm layer of photoresist is next loaded onto the wafer chuck 14 and positioned as before so that its surface is parallel to the hologram 9 and at a distance of 100 Sm from it.

This second wafer is now printed. On this occasion, as the laser beam 17 scans the hologram 9, the computer 20 modulates the output power of the laser 18 via the RS232 link 21 and interface card 19. The instantaneous laser power is calculated from the location of the beam 17 on the hologram 9, from the map of CD measurements entered in the computer and from the relationship between linewidth and exposure energy. A number of possibilities exist for the exact detail of the algorithm employed by the computer but a representative one is as follows : applying the same imaginary grid as before to the second wafer, when the centre of the beam enters a grid square, the laser beam intensity is adjusted to compensate for the difference between the linewidth measurement made for the respective grid square on the test wafer and the desired value. Another possible algorithm that is applicable if the beam size is large compared to the grid square, is to take account of all the grid squares being instantaneously illuminated by the beam and then to calculate the average value of the respective linewidth measurements made on the test wafer. A more sophisticated variant of this is to calculate a weighted average to take account of the Gaussian intensity distribution of the beam. The laser power is then adjusted to compensate for the difference between the average value calculated from the test wafer and the desired value.

Instead of modulating the intensity of the laser beam as the beam 17 scans the hologram 9, the computer 20 may alternatively modulate the speed of the beam 17. The instantaneous speed of the beam is determined in an analogous manner to the determination of the instantaneous intensity described above.

It should be noted that higher accuracy compensation of the CD errors can be obtained by using a finer measurement grid and by using a scanning beam with a smaller diameter (necessitating also a smaller stepping distance between scan passes).

Following the printing operation, the wafer is removed the chuck 14 and the photoresist developed.

Claims (8)

1. A method for printing a pattern of features, including:

a) providing a mask bearing the pattern of features; b) recording a TIR hologram of the mask; c) printing the pattern of features recorded in the TIR hologram into a first layer of photosensitive material on a first substrate by illuminating the TIR hologram with a first scanning laser beam having a constant intensity and a constant speed; d) measuring the dimensions of at least two features in the pattern printed in the first layer of photosensitive material and determining the deviations of these dimensions from their desired values; e) printing the pattern of features recorded in the TIR hologram into a second layer of photosensitive material on a second substrate by illuminating the TIR hologram with a second scanning beam and modulating the exposure energy density imparted by the second scanning beam such that the dimensions of the features printed in the second layer of photosensitive material have their desired values.

2. A method according to claim 1 wherein in step e) said modulation of the exposure energy density is produced by varying the intensity of illumination of the second scanning beam.

3. A method according to claim 1 wherein in step e) said modulation of the exposure energy density is produced by varying the speed of the second scanning beam.

4. A method according to claim 1 wherein in step c) the local separation between the TIR hologram and the first layer of photosensitive material where the TIR hologram is being illuminated by the first scanning beam is measured and adjusted to a constant value.

5. A method according to claim 1 wherein in step e) the local separation between the TIR hologram and the second layer of photosensitive material where the TIR hologram is being illuminated by the second scanning beam is measured and adjusted to a constant value.

6. A method for printing a pattern of features, including a) providing a mask bearing the pattern of features; b) measuring the dimensions of at least two features in the mask and determining the differences of these dimensions from their desired values; c) recording a TIR hologram of the pattern of features; d) printing the pattern of features from the TIR hologram into a layer of photosensitive material on a substrate by illuminating the TIR hologram with a scanning beam and modulating the exposure energy density imparted by the scanning beam such that the dimensions of the features printed on the substrate have their desired values.

7. A method according to claim 6, wherein in step d) said modulation of exposure energy is produced by varying the intensity of illumination of the scanning beam.

8. A method according to claim 6, wherein in step d) said modulation of exposure energy is produced by varying the speed of the scanning beam.