CA1275846C - Process of forming a negative pattern in a photoresist layer - Google Patents

Abstract

ABSTRACT OF THE DISCLOSURE:

A process of forming high resolution negative patterns in a photoresist layer, comprises the steps of (a) coating a substrate with a layer of a photosensitive resin comprising a polymer, preferably a phenolic polymer, mixed or bound to a photoactive compound such as a diazoquinone, (b) exposing the layer to ultraviolet or visible light through a mask, (c) treating the layer with a silicon compound (e.g. hexamethyldisilazane) and (d) dry developing by plasma etching (e.g. an oxygen plasma) to remove the non-irradiated portions of the layer. The silicon compound is able to diffuse selectively into the irradiated portions of the layer and fix in these portion. By dry etching, a silicon oxide etch mask is formed which protects these irradiated portions efficiently throughout the process. The process is useful in the manufacture of semiconductor devices, including integrated circuits.

Description

:~275~

The present invention relates to a new microlithographic process.
~ ore particularly, it relates to a process of forming a negative pattern in a photoresist layer which enables the production at industrial scale, ~ith high yield, of very large scale integrated microcircuits (VL~I), the linewidth of which can go down to the submicrometer level. The invention also relates to the use of this new process in the manufacture of integrated se~icon~uctor circuits.
The continuing trend towards miniaturization in the field of integrated semi-conductor circuits gives rise to the need to accomodate more active circuits per unit area on the surace of a semiconductor cry~tal.
As an illustration of this, reference is made to the commercial development of metal oxide semiconductor (MOS) random access memory (R~) devices from 1 Kbyte in 1975, via 16 Kbyte in 1977 and 64 Kbyte in 1979 to 256 Kbyte in 1982. It is expected that this trend will continue over the next two decades. As a result, the mini-num ~eature size of semiconductor devices is expected to continue to decrease from 8 micrometers for 1 Kbyte MOS RA~ devices,over 2 micrometers for 256 Kbyte ~IOS R~M devices in 1982 to below l micrometer before the end of the 1980's.
Microcircuit ~abrication requires the select;ve d;ffusion of controlled, small quantities of impurities into specific regions of the semiconductor sur~ace to produce the desired eLectrical characteristics of the circuit, such as transistor/e~ement~ of which a large scale integrated circuit contains several tens of thousands of individual units that are interconnected in complex ways by conductors, such as aluminium or highly doped polycrystalline silicon.
The technique used in the commercial production of integrated circuits to obtain specific patterns is called photolithography or microlithography. The optionally oxidized silicon substrate (wafer), is coated with a photosensitive layer (also called photoresist) and exposed through a mask to ultraviolet light The chemical and physical properties of the irradiated regions are different from the non-irradiated regions and create the possibility, in a development step, of removing the exposed (in a positive resist) or unexposed part (in a negative resist).

.
? ) ., ~2~7~ii8~Ç;
Development processes are mainly based on solubility differences and are carried out by a wet technique. After removal of part of the resist layer, the uncovered substrate surface can be treated (etched, doped, oxidized, nitride treated, plated, etc.).
This photolithography process is repeated (to more than 10 times) before the-~hree~imensional circuit geometries necessary for a completed metal oxide semiconductor (MOS) or b;polar device are achieved. The structure of an integrated circuit is complex, both in the topography of its surEace and in its internal composition. Each element of this device has an intricate three-di~ensional structure that must be reproduced exactly in every circuit. The structure is made up of many layers, each of which is a detailed pattern. Some of the layers lie within the silicon wafer and others are stacked on the top.
The process is described in detail in the book of L.F. THO~PSON, C.G.WILLSON
and M.J~ BOWDEN "Introduction to Microlithography", A~erican Chemical Society Symposium Series 219, Amer.Chem.Soc., Washington D.C., 1983.
As a result of the increasing density of very large scale integrated circuits, the minimum feature size of semiconductor devices decreases and the production processes become more difficult. Achieving a high resolution with good linewidth control on substrates with surface topography is a serious problem. As a result of optical reflections and variations of resist thickness over high steps, ~inewidth control becomes very difficult and relatively thick resist layers are necessary. Since, for these small features, lateral dimensionsshrink more rapidly than vertical dimensions, higher height-to-width aspect ratios of resist imagcs are required. In addition, dry etch techniques require relative thick and stable resist patterns. However, thick resist layers limit resolution and give depth of Eocus problems for projection printing. Furthermoreswelling of negat;ve resists in organic solvent developers also makes them unsuitable for high resolution microLithography and restricts the choice to the positive-acting diazoquinone resists. However, even these high contrast and high resolution resists become insufficient when linewidths decrease to the micrometer and submicrometer regime.
Several approaches to obtain higher resolution and better linewidth control have been studied during the past few years. New exposure techniques are being investigated, such as electron beam writing, X-ray and deep W
exposure. Electron beam lithography requires very costly equipment, suffers from low throughput and is unable to produce h;gh aspect ratios as a result oE backscattering of electrons, i.e. the so-called proximity effect. ~-ray exposure has ~ery high resolutioll capabilities but equipment and resist materials are s~ill in the developmental stage. The manufacturing of ~-ray masks is also a difficult operation. The use of deep W light decreases diff-ract;on effects, but resolution is still lim;ted by swelling of the negative resists and positive deep W resists with sufficient sensitivity are not yet available.
In order to eliminate the adverse effects of solvents during wet developing, plasma development of photoresists has been considered (cf.
J.N. S~ITH, ~I.G. HUGHES, J.V. KE~LER, ~.R. GOODNER and T.E. ~IOOD, in Semiconductor International~2,(1979,n10, December),p.41-4~. The dry development of resists is an important step towards achieving a totally dry microfabrication process, offering better process control, reproducibility and cleanliness. [n addit;on, dry development solves problems such as resolution loss by swelling (particularly in the case of negative resists) and handl;ng of large amounts of inorganic or organ;c solvents. It also has a much better adaptability to automa~ic l;ne fabrication processes. However, most dry cievelopment res;st systems have a low contrast and undergo a serious reduction o~ resi~t thickness during clevelopment.
~ nother approach to obtain high resolution and high aspect ratio consists ;n the use oe the multi-layer resist systems. tn these processes, excellent step covernge alld clry etc~l resistance oE a thick ptanarizing bottom layer arecombiaed with the high resolution oE a thin top imaging Layer. Both layers can bë optimized ~or their specific requirements. After exposure and development of the thin top resist, the patterns obtained are transeerred almost vertically into the thick underlying layer either by deep W exposure and development or by oxygen-reactive ion etching. For the latter method, a third interMediate layer is sandwiched between the imaging and the planarizing layer. Usually, a thin layer of plasma-deposited or spin-on silicon oxide is applied on top of the planarizing layer. The exposed and developed patterns in the top resist layer are first transEerred into the intermediate (silicon oxide) layer by etching techniques. After removal of the resist, the thin oxide patterns act as a very effective mask for oxygen-reactive ion etching of the thick bottom layer, resulting in profiles with almost vertical sidewalls. Aluminium or _ .4 _ silicon nitride have also been used as intermediate layer.
With these techniques, high resolution and high aspect ratios can be realized on substrates with surface topography (c~. J.M. MORAN and D. I~YDAN
in The Be~l System Technical Journal,58,(1979,n5,May-June, p.1027-1036).
However, the multi-layer resist systems have also several serious draw-backs. Thus, formation of interfacial layers occurs when one poly~er material is spun on another. Hard bake steps may somewhat reduce this phenomenon but even then additional treatmen~s are required to remove these layers before pattern transfer. However, baking may induce film stress. Spun-on silicon oxide films are very susceptible to cracking when the baking temperature is too high.
On the other hand, a sufficiently high baking temperature is required to avoid interfacial mixing between the oxide and the imaging layer.
Another problem is the optical interference effects. Even with a highly absorptive bottom layer, reflections can occur at the interfaces, resulcing in standing waves. A post-exposure bake of the imaging layer can reduce this effect but only at the expense of a contrast loss.
Finally, the thin top resist layers which are used in multi-layer systems to produce better images make the system more susceptible to pinhole problems.
Multi-layer resist systems are discussed in depth by B.J. LIN in the above-mentioned book 'IIntroduction to ~;crolithography"~ p.~87-350.
Another major disadvantage of multi-layer systems is process complexity.
Several layers have to be applied and b~ked and each of them has to be exposed, developed and/or etched. ThereEore, efforts have recently been made to simplify these multi-layer systems, without losing cheir advantages. One example is the two-layer reactive ion etching (RIE) process in which the top imaging layer is a resist containing both organic and inorganic components (for example, poly(dimethyl siloxane) and poly(vinyl-methyl siloxane) (cf.
M. HATZAKIS et al., in Proceedings of the International Conference on Microlith-graphy, 1981, p.386-396), copolymers of chloromethylstyrene and trimethylsilyl--styrene (cf. M. SUZUKI et al., in J.Electrochem.Soc.130,(19R3),p.1962-1964) and copolymers of vinyltrimethylsilane and sulfur dioxide (U.S.Patent Specification No. 4,357,369). The thin top resist Layer is exposed and developed, followed by etching of the thick bottom layer by oxygen RIE. During this process, the inorganic components in the resist form refractive oxides which act as an in situ-formed etch mask. In this way, the top imaging layer and the intermediate silicon oxide layer of the three-layer resist system are combined into one organo-l~etallic resist layer. Processing of these syst~ms is easier than the th~ee-layer resists but is still more complex than single layer resist systems. rhey have also only been developed so far for eLectron beam and deep l;V exposure and s[ill require wet development in organic solvents. Another example of a simplified multi-layer system consists in the use of a contrast enhancement layer. In this process, a thin photo-bleachable layer is coated over a standard positive photoresist. During exposure, the dye in the coating bleaches, forming a new mask in intimate contact with the resist surEace. After exposure, the layer is removed in a solvent system and the resist is further processed by standard techniques.
Although this method improves contrast and adds little to the process complexity, it does not overcome the standing waves problem and linewidth variations over steps. In addition, extended exposure times are required.
The ideal would obviously be to have at one's disposal a single layer resist system affording the same technical advantages as multi-layer resist systems, but which does not have their drawbacks. In this way, the formation of interfacial layers, optical interference effects and film stress problems would disappear and process complexity would be drastically reduced.
~towever, the difficuLty is to find a single layer resist system, which can be dry developed and wnich is char.lcterized at the same time by high resolution, high aspect ratios and excellent linewidth control on substrates with surface topography, ;n order to obtain hig~l quality resist patterns which enable the production with a good reproducibility of very large scale semiconductor circuits.
That is the reason why various attempts have been made to improve the single layer resist systems. As an example, in Japanese patent application n 23937/82, a process is described in which a positive photo-resist, coated on a substrate, is exposed to an atmosphere containing an organic silicon compound~ more particularly hexamethyldisilazane. After exposure to ultraviolet light, the photoresist is developed in a conventional manner by dissolving the exposed areas with an alkaline liquid developer. The treatment with the hexamethyldisilazane vapors has as effect to reduce the developing rate of the photoresist in the alkaline developing solution, tnus * published 8th February 1982 ~1 permitting the formation of a resist pattern having a higher contrast (more vertical edges).
It must be emphasized that, apart from the treatment with the hexamethyl-disilazane vapor, this process does not differ in any way from the conventional photolithographic process. The image formation is still based on soLubility differences produced by the creation of carbo~ylic groups in the photoresist under the effect of the ultraviolet light and development is exclusively carried out by wet development in an alkaline developer. Even if the contrast is enhanced by this process, the well known drawbacks of wet development are not suppressed. The light exposure has to be effected in depth in the resist layer, with the consequence that effects such as reflections and standing waves cannot be suppressed. This can give rise to linewidth variations (especially on a reflective topograyhy).
D. FOLLET et al. ("Polarity reversal of PM~A by treatment with chlorosila-lS nes" - The Electrochemical Socie~y Extended Abstracts,82-2 (19~2, October 17-21), p.321-322) describe a process in which a poly(methyl methacrylate) resin (hereafter called PMMA) is electron beam irradiated, then sequentially treated with dichlorodimethylsilane and water vapor, to form polysiloxanes, and Einally,developed in an oxygen plasma. According to these authors, there would be selective difEusion oE the dichlorodimethylsilane in the irradiated and partial'ly degraded areas of the resist, foLlowed by polymeri~ation oE the chlorosilane upon expo~ure to the water vapor. They find a polarity reversal oE the res;st Erom positive into negative: the irradiatecl and treated areas of the PMM~ oEEer indced a higher resistance to the oxygen plasma. A pattern is obtained consisting of 2 /um Lines and ~ ~n spaces and abrupt edge profiles.
This process has nevertheless several important drawbacks. Electron beam exposure, eve~l iE it allows a high resolution, requires a very expensive equipment, and the irradiation times per silicon waEer are too long. Because of these limitations, up to the present electron beam lithography is only used for direct writing of patterns for devices intended for research. Moreover,PMMA is a material which ofEers a very weak resistance to dry etching operationsbecause it degrades very rapidly in the plasmas used to ~his end (5 times fasterthan the aromatic polymers). Finally, owing to the fact that dichlorodimethyl-~2~7S~

silane is not immediately fixed in the irradiated areas of the PMMA layer(since water vapor is necessary for the conversion in polysiloxanes), it is conceivable that this compound can easily go out again from the layer by diffusion. This will necessarily have a detrimental effect on the reproducibi-lity of the characteristics of the obtained patterns, since the concentrationof the dichlorodimethylsilane in the layer will be a function of the time which elapses between the treatment with this compound and the treatment with the water vapor.
T.M. WOLF et al., in J.Electrochem.Soc.131,(1984,n7),p.1664-1670, propose still another process intended to improve the single layer resist systems. The photoresists used by these authors are negative photoresists conventionally used in photolithography. They consist of a partially cyclized polyisoprene containing a bis-azide as photosensitizer (co~mercial WAYCOAT IC-43*and SELECTILUX N-60~).The proposed process comprises the steps of ultraviolet ;rradiation or electron beam exposure, treatment with a volatile inorganic halide, followed by development by oxygen-reactive ion etching. In the described experiments, these authors use silicon tetrachloride (SiC14), tin tetrachloride (SnCl4) and dichlorodimethylsilane ((CH3)2SiCl2) as volatile ;norganic halides, since these compounds are able to react with the secondary am;nes created during the photolysis of the resist. They hoped indeed that this reaction could be used to incorporate the inorganic halides selectively into the exposed areas of the resist, thus allowing the formation by oxidation of a th;ck irlorganic oxide protecting l~yer in said exposed areas during the following development step by oxygen-reactive ion etching. Only the unexposed areas of the photoresist would thus be removed selectively by oxy~en-reactive ion etching.
Nevertheless, they find that the inorganic compounds are sorbed quickly, not only in the exposed areas, but also in the unexposed areas of the resist.
Moreover, they note that, contrary to all expectations, by oxygen-reactive ion etching, the unexposed areas of the photoresist are orotectedby an oxide layer and etched at a significantly lower rate than the exposed areas, whereas these latter are selectively removed during the dry etching development step. In other words, the photoresist behaves as a positive tone resist. The authors explain this phenomenon by the fact that, in the unexposed areas, complexes between the inorganic halide and the azide group of the photosensitizer are also formed, while in the exposed areas, the * trade mark ~7~

anticipated reaction with the photolysis products of the resist takes place.
Moreover, in the unexposed areas, the formed complexes are readily hydrolyzed by the water vapor present in the ambient atmosphere and converted to refractory inorganic oxides, thus forming an in situ protective masking layer, whereas the reaction products formed in the exposed areas are more slowly hydrolyzed and therefore read-ily removed as volatile compounds during oxygen-reactive ion etching.
Nevertheless, this process still has important drawbacks.
First, according to the statements made by the authors themselves, th;s process is satisfactory only provided very strict light exposure times and at the same time very stric~ treatment times with the inorganic halide are observed. Indeed, in order to obtain a satisfactory result, it is necessary that the exposure of the photoresist to light takes place during 16 seconds and that the treatment with the inorganic halide takes place during 7 seconds.
Shorter or longer exposure and/or treatment times give unsatisfactory results.
Thus, for example, after a treatment with the inorganic halide for 10 seconds, it is no longer possible to develop the resist by etching, regardless of exposurtime. In other words, the operating conditions are very critical, which can only be ~etrimental to the reproducibility of the results.
Secondly, ev~n ulld~r the optm~l oX~it~G cibed hereinbefore, only 70% of the initial thickness oE the photoresist remains after development by dry etching.
Thirclly, as shown in figure 11 of page 1669 of the publication,there remains aEter ~evelopment~ an important residue in the uncovered/, which residueis very difficult to remove (this fact shows at the same time that the selectivity of the process is insufficient).
To conclude, it can be seen that till now, a microlithographic process has not yet been developed in which a single layer resist system is used which is ent;rely satisfactory.
It is for these reasons that we have carried out research work to develop a single layer microlithographic process, which would be free from the drawbacks of the hitherto known processesl particularly of those described in Japanese patent application n 23~37/32 and in the above mentioned publications of D. FOLLET et al., and of T.~. WOI.F et al.
This object is fully achieved by the process described nereinafter, which has all the advantages and the simplicity of single layer microlitho-graphy, which allows the use of dry etching techniques, which can be used on standard projection printing equipment and wafer stepper equipment, and yet gives equal or better suilmicrometer re~olution than the above-mentioned multi-layer processes.
Thus, the present inven~ion provides a new improved microlithographic process. ~ore specifically, the present inVentiOn provides a process of forming a negative pattern in a photoresist layer comprising the following steps:
(a) coating a substrate with a layer of a photosensitive resin comprising a polymer mixed or bound to a photoactive compound, said layer enabling a silicon compound to diffuse selectively into irradiated portions thereof, on exposure of said portions to visible or ultraviolet radiation;
(b) exposing the photosensitive resin layer to ultraviolet or visiole light through a mask to expose only selected portions of the layer;
(c) treating the photosensitive resin layer with a silicon compound, whereby said compound is selectively abeorbed ;nto the irradiated portions of the coating and is caused to react with said irradiated portions; and (d) dry developing the thus treated photosensitive resin layer by plasma etching to remove selectively the non-irradiated portions thereof in order to obtain the desired negative pattern.
Accordin6 to a preferred embodiment of the present invention, the substrate is a silicon wa~er, the photosensitive resin used comprises a phenolicpolymer and the photoactive compound which is mlxed or bound to it is a diazoquinone, whereas the silicon compound is an easily vaporizable silylatin~
agent.
According to a particularly preferred embodiment of the present invention, the phenolic polymer is chemically bound to the dia~oquinone.
On the other hand, according to the present invention, the treatment with the silicon compound is preferably carried out after exposure of the photo-sensitive resin layer to ultraviolet or visible light. However, those skilledin thè art will understand that, for the purpose of simplicity and readiness of e~ecution, it is possible to carry out the treatment with the silicon compound already during the exposure oE the photosensitive resin layer to light. In other words, it is possible to conduct steps (b~ and (c) of the process simultaneously.

34~;

According to a particularly preferred embodiment, the silicon compound is volatilized and brought in~o contact in vapor form with the photosensitive resin layer.
~ ccording to a particularly advantageous embodiment of the invention, dry S development is carried out by oxygen-reactive ion or oxygen plasma eeching.
The invention further comprises the use of the new microlithographic process in the manufacture of in~egrated semiconductor circuits.
The process according to the invention is based on the surprising discovery that positive acting photosensitive resins, comprising a polymer associated with a photoactive compound such as a diazoquinone, undergo substantial modifications of their properties under the influence of a visible or ultraviolet radia~ion. We have found indeed that the exposure to tne radiation modifies to a considerable extent the permeability properties of these resins, and we have taken advantage of this particularity in order to better differentiate the irradiated portions from the non-irradiated portions of a layer of these resins coated on a substrate, and in consequence, to develop an improved single layer microl;tnographic process. To this end, according to the invention, the irradiated photosensitive resin layer is subjected to a ~reatment with a silicon compound, in order to allow, thanks to the particularity just merltioned hereinbeore, this compound co penetrate sel~ctively in the irradiated portions of the layer, preferably in the top part thereof, and consequently fix selectively in these irradiated portions by reaction with the functional groups of the photosensitive resin.
E~perience has shown that the silicon compounds actually diffuse selectively in the irradiated portions of the layer, whereas such a diffusion practically does not occur in the non-irradiated portions, or only to a minor extent.
Thus, unlike the process of T.M. WOLF et al. above described, the silicon compounds do not difEuse in all the regions of the photoresist layer, but only in the irradiated portions of this layer.
The process of the invention is thus based on selective diffusion oE che silicon compolmds in the irradiated portions of the photoresist, unlike the process of the state of the art, which is exclusively based on a selective .~
;

~75i~

reaction of the silicon compounds in either the exposed or the unexposed regions (the silicon compounds being incorporated in every region of the photoresist).
Frorn the technical point of view, this difference has considerable repercussiOns on the results because, in the process according to the invention, after development by dry plasma etching, superior quality patterns are obtained. Moreover, the above-mentioned disadvantages of the known single layer processes are definitively suppressed. By dry etching, the non-irradiated portions of the photoresist are completely removed and, in the irradiated portions, a silicon oxide mask is very quickly formed in situ, which remains and protects very efficiently these portions throughout the etching process. After development, the thickness of the obtained pattern is practically identical to the initial thickness of the photoresist; the residual thickness generally represents more than 95% of the initial thickness of the resin layer coated on the substrate. In the uncovered parts, no residue remains. Moreover, the reproducib:ility is considerably improved since the light exposure times and the treatment times with the silicon compound do practically not have any effect on the results (contrary to the process of T.M. WOLF et al. above described).
Thus, the present invention provides a single layer resist system which can be dry developed and which has all the advantages of the multi-layer resists: planari~.ation, high resolution, high aspect ratios with excellent retention of the initial thickness of the photosensitive resin layer coated on the substrate, good linewidth control over the steps, good reproductibility of the pattern characteristics and suppression of the interference with light reflected at subscrate topography. In addition, the described system has several advantages over the multi-layer resist systems. Formation of interfacial layers does not occur, interference with reflected light does not occur, film stress problems do not exist and process complexity is drastically reduced.
Even if the process of the invention provides a negative acting resist system, it does not need wet development, which definitively suppresses the problems of resolution losses caused by the swelling in developrnent solvents.
Moreover, since no solvents are used in the development step, adhesion is no longer a critical parameter. The completely dry processing of the resist ~%~
results in improved process control and makes this system particularly suitable for automatic line fabrication processes.
The invention will now be further described with reference to the accompanying drawing where;n like reference numerals refer to same parts throughout the several views, where~n the layers are greatly exaggerated.
Figure 1 is a partial cross-section of a silicon wafer coaced with a photosensitive resin layer before the exposure step.
Figure 2 is a partial cross-section of a silicon wafer coated with a photosensitive resin layer during its exposure eo light through a mask.
Figure 3 is a partial cross-section of a silicon wafer coated with a photosensitive resin layer after the treatment with the silicon compound.
Figure 4 is a partial cross-section of a negative pattern obtained after development by oxygen-reactive ion or oxygen plasma etching.
A wide range of ?olymers can be used in order to prepare the photosensitive resin used in the process according to the invention. However, preferably a phenolic polymer is used, which is selected from:
- the condensation products of a phenol, a naphthol or a derivative thereof ring-sub~tituted by ~n aLlcyl or aryl radical or by a halogen atom, with an aliphatic or aromatic aldehyde, which can be substit~lted by a halogen atom, - the poly(vinylphenols), the phenolic ~roup of which can be substituted by an all~yl or aryl radical or a halogen atom, - the copolymers of a vinylphenol with an ethylenically unsaturated compound, and - mixtures of the aforesaid polymers between them or with other aromatic polymers, such as polystyrene or poly(N-vinylcarbazole).
As illustrative but not limitative examples of phenolic polymers, there may be mentioned: phenol-novolacs, cresol-novolacs, ccndensation products of formaldehyde with alkylphenols (p-tert-butylphenol; p-n-propylphenol; p-ethylphenol; octylphenol and the like), condensation products of benzaldehyde with cresols or naphthols (e.g. l-naphthol), poly(p-vinylphenols), copolymers of p-vinylphenols with p-chlorostyrene, and the like.
The photoactive compound mixed or hound to the polymer is preferably a diazoquinone such as those used in traditional positive photo-f~

~7~
resists (see ~ritish Patent Specification No. 711,626). As non-limitative examples of these dia~oquinones, there may be mentioned: 5-diazo-5,6-dihydro-6-oxo-1-naphthalenesulfonic acid, 6-diazo-5,6-dihydro-5-oxo-1-naphthalene-sulfonic acid, 3-diazo-3,4-dihydro-4--oxo-1-naphthalenesulfonic acid, 4-diazo-3,4-dihydro-3-oxo-1-naphthalenesulfonic acid, j-diazo-3,4-dillyaro-4-oxo-1-benzenesulfonic acid, the corresponding carboxylic acids, derivatives thereof and mixtures of at least two of the aforesaid compounds.
As an example of a derivative of the aforesaid acids, the condensation product of 3 moles of 6-diazo-5,6-dihydro-5-oxo-1-naphthalenesulfonyl chloride with 1 mole of 2,3,4-trihydroxybenzophenone may be mentioned.
The photoactive compounds such as diazoquinones can easily be chemically bound to the aforesaid poly~ers by conventional synthetic processes (see U.S.Patent Specification No~ 3,046,119); a typical well-known reaction is the reaction of the derivatives of these photoactive compounds carrying a sulfonyl chloride subst;tuene (or carbonyl chloride) with the hydroxyl groups of the phenolic polymers, with formation of sulfonates (or carboxylates).
As shown in Figure 1, in the first step of the process o~ the invention, a thick photosensitive resin layer (1) is coated on a silicon wafer or substrate (2). 'rhis coating is generally effected by spin-coating from a dilute solution of the photosensitive resin in a suitable solvent. The thickness of the photosensitive resin layer may vary depending upon the topography Oe ~hc substra~e. The presence of the polymer in the photosensitive resirl facilitates formation of smooth, even-levelled surfaces on the substrate.
When the photosensitive resin is coated on the substrate from a dilute solution in a solvent, the solvent may be selected from a wide range of products, depending on solubility parameters, evaporation rate at processing temperatures and desired rheological characteristics. E~amples of solvents which can be used include ketones, escers such as acetates, propionates, and the like, ethers, ether esters, polyethers, alcohols, aliphatic and aromatic hydrocarbons, tetrahydrofuran, dimethylformamide, pyrrolidones, and the like.
The nature o~ the solvent is irrelevànt to the subject of the present invention as long as a solution is obtained with film-forming properties. The concentra-tion of the solution depends upon the required thickness of the photosensitive resin layer.

",, ~

~27~

After having been coated on the substrate, the photosensitive resin layer (1) is generally dried by baking to remove the solvent.
The conditions used during this baking step are not critical, as long as solvent is efficiently removed and a tack-free surface is obtained. In practice, temperatures of about 50 to about 150C, and preferably of ~bout 80 to 120C are used. The baking time can be shortened when the baking temperature i9 increased. This time will also depend upon the ease of evaporation of the solvent, or of the solvent mixture used. Baking time used in practice varies between a few seconds and 1 hour. After baking, the thickness of the photosensitive resin layer (1) is generally between 1 and 2.5 ~ .
As is shown in Figure 2, the substrate (2) coated with the photosensitive resin layer ~1) is then exposed to visible or ultraviolet radiation (3) through a photoimaging mask (4). The wavelength of the applied radiation czn lS vary within a range of from about 100 to 600 nm, preferably of from about 350 to 450 nm. Under the eEfect of visible or ultraviolet radiation, the chemical and physical propert;es of the irradiated portions (5) of the photosensitive resin layer undergo considerable modifications as compared with the non-irradiated portions.
~0 Thoueh mechanisms of the reaction and/or decomposition caused by the exposure to visible or ultraviolet light have not been fully elucidated yet, it ;s well known that, when the photoactive compound is a dia~oquinone, nitroger;g evoLvecl and Ill indenecLIrbo~ylic acid is formed, in the ?resence o~ water.Ilowever, as alrellcly explained above, a very substantial mo~ification of the permeability oro?erties o~ the photosensitive resin Layer occurs simultaneously.
The visible or ultraviolet exposure energy can be adjusted so as to induce this modification of the permeability in the photosensitive resin layer at the desired depth. According to a preferred embodiment of the process of the invention, this modification is only induced in the top part, near the surface of the layer.
Typically, but not in a limiting manner, the applied e~posure energy may vary from 10 to 130 millijoules per square centimeter, measured at a wavelength of 400 nm; the quantity of energy to be applied depends obviously on the nature of the photosensitive resin and of the silicon compound used in the following step of the process.

., .,~ . ., ~7~
Light absorbance of the photosensitive resin layer can possibly be increased by adding specific dyes, as described by M.M. O'TOOLE, E.D. LIU and M.S. C~NG in Proc. SPIE Int.Scc.Opt.Eng.,275,(1981) (Semicond.Microlithogr.
VI),p.128-135 and by ~. CHEN et al., in U.S. Patent Specification No. 4,362,809. These dyes should absorb significantly at the wavelengths used to expose the photosensitive resin (100 to 600 nm, preferably 350-450 nm), but they have to be transparent in a part of the visible region to make alignment possible during printing. A proper concentration of such a dye will restrict exposure to a thin upper layer in the photoresist. Photosensitizers or other similar agents can also be added to the photosensitive resin.
As already explained above, the treatment of the photosensitive resin layer with the silicon compound can be effected during the exposure to light, but preferably after this exposure, for simple reasons of ease of handling.
Furthermore, this treatment can be carried out in the liquid or in the vapor phase. According to a preferred embodiment of the invention, this treatment is carried out in the vapor phase. This latter process is preferred since it enables the use of silicon compounds and more particularly of silylating agents easily vaporizable at appropriate temperatures and pressures.
As non-limitative examples of useful silicon compounds, the following silylating agents can be mentioned: tetrachlorosilane and alkyl- and aryl-halosil~nes ~e.g. trime.thylchlorosilane, dimethyldicnlorosilane, methyl-trichlorosilane, crimethylbromosilane, trimethyliodosilane, triphenylchloro-silane), dis;lazanes (e.g. hexame~hyldisilazane, heptamethyldisilazane, hexaphenyldisilazane and 1,3-bis(chloromethyl)-1,1,3,3-tetramethyldisilazane), N-trimethylsilylimidazoLe, N-trimethylsilylacetamide, N-trimethylsilyl-dimethylamine, N-trimethylsilyldiethylamine, hexamethylsilaned;amine, N,O-bis(triethylsilyl)acetimide~ N,N'-bis(trimethylsilyl)urea, N,N'-diphenyl-N-(trimethylsilyl)urea and mixtures of at least two of these compounds.
The treatment of the photosensitive resin layer with the silicon compound can be carried out in a separate reaction chamber or in a wafer stepper equipment.
The treatment with the silicon compound produces, as already explained above, the selective diffusion of the silicon compound in the irradiated portions (5) of the layer; this compound is then fixed in these portions by reacting with the functional groupg of the resin. On the other hand, in the non-irradiated portions, there is practically no diffusion of the silicon compound or only to a minor extent at the surface. Thus, the incorporation of the silicon compound occurs only in the irradiated portions of the layer, and not in the non-irradiated portions thereof. This fact has been ascertained by analysis with AUGER electron spectroscopy (see e~ampLe 12).
~ s is shown in Figure 3, the silicon compound can advantageously be built into the top part (6) of the irradiated portions (5) of the photosensitive resin layer. The thickness of the silicon-containing layer is governed by the laws of diffusion and will vary with the nature of the silicon compound and its concentration and the composition of the photosensitive resin. For a given system7 it will depend upon the exposure dose, the time, the temperature and the pressure of the treatment, and, therefore, can be directed by readily controllable parameters.
Since imaging can occur in a very thin planar layer at the top part of the photoresist, high quality images can be obtained by standard projection printing with sharply focussed light. It will indeed be sufficient to limit light exposure to a thin layer at the top part of the photosensitive resin, in such a way that light is sufficiently extinguished by absorption in the remaining part of the layer, before interference of reflected light can occur.
~0 ~s a result, proble~ns such as standing waves and linewidth variations over highly reflective 8teps, such n9 metal interconnectors, ~re eliminated.
The treatment oE the photosensitive resin Layer with the silicon compound is carried out at a suitable temperature, which depends upon the nature of the resin and upon the silicon compound u ~ ; ~s temperature is selected in such a manner that a seLective diffusion of the silicon compound occurs in the irradiated portions of the resin; it can be selected between the volatilization temperature of the silicon compound (which depends itself of the pressure prevailing in the system, which may vary from a relative vacuum to several bars)and the thermal decomposition temperature of the components of the photo-sensitive resin. Practically, the temperature of this treatment can varywithin a range of from about -20 to 150C, and preferably of from 60 to 140C.
The time of the treatment with the silicon compound is not critical; it depends essentially upon the temperature used, and accessorily upon the nature of the photosensitive resin and of the silicon compound. Practically, this time can vary between a few seconds and one hour and preferably between I and ~Z7~846 45 minutes.
After treat~en~ of the photosensitive resin with the silicon compound, excess thereof may be removed for reasons of further ease of handling; this removal is preferabLy carried out by vacuum evaporation. Nevertheless, it is not compulsory to remove excess silicon com?ound, since the qualities of the patterns obtained by the process of the invention are qui~e as good without removal of excess silicon compound as with its removal; this can be explained by the fact that the silicon compound pratically does not diffuse in the non-irradiated portions of the photosensitive resin and that these portions are thus easily removed by ~urther development by dry etcning by means of oxygen gas plasma or oxygen-reactive ion etching.
In the last step of the process of the invention, the photosensitive resin layer treated as described above, is dry developed using preIerably an oxygen plasma or oxygen-reactive ion etching technique. These techniques, as well as lS devicesused for their execution, are well kno~n to those skilled in the art and need not be described in detail.
As is shown in Figure 4, development by dry etching has as consequence that the non-irradiated portions of the photoresist are quickly removed, whereas the irradiated portions (5) resist perfectly owing to the Ln situ formed silicon oxide etch mask.
When the non-irradiated portions are completely removed up to the surface of the substrate (2), excellent quality negative patterns are obtained. Indeed, as is further shown in the examples, these patterns have vertical side-walls, high resolution, whereby lineg of less than ~lE a micrometer wide and spaces near half a micrometer wide can be obtained, a perEect retention of the initial resis~ layer thickness, no measurable variation in linewidth over steps and high aspect ratios, with heights over 2 micrometers. Furthermore, the uncovered regions contain no residue.
The following examples are given only for the purpose of illustratin~
the present invention. In these examples, the values given for the UV exposure energy are measured at a wavelength of 400 nanometers.
Example l.
Silicon wafers with thermal oxide of approximately 120 nanometers thickness are primed with hexamethyldisilazane as adhesion promoter. As photosensitive resin, the partial esterification product of 6-diazo-5,o-dihydro-;-oxo-l-~`' ~27~ 6 naphthalenesulfonic chloride wit'n the condensation product of p-tert-butyl-phenol and ~ormaldehyde, is used. This resin is dissolved in a solvent mixture contain;ng ~0% by weight of 2-ethoxyetllanol, 10% by weight of xylene and 10~ by weight of ~-butyl acetate, in order to obtain a 25% by weight solution.
This solution is spin-coated on the silicon wafers at a spin speed of 3000 rpm. In this way, a resin layer of 1.7 /um thickness is obtained on each wafer.
The thus coated wafers are baked in a convection oven at 90C for 30 minutes. Then they are exposed through a pattern mask to UV rays, in a conventional equipment producing W light, the wavelength of which lies between 350 and 440 nanometers; the exposure energy is 60 mJ/cm .
The exposed wafers are then treated wit.h hexamethyldisilazane vapor at 91C for 4 minutes.
After development by oxygen-reactive ion etching, negative patterns with vertical side-walls are obtained in the exposed parts. Furtherr.lore, the thickness of the patterns is practically identical to the initial thickness of the resin layer coated on each wafer.
Example 2.
.

; 20 The procedure o~ ExampLe I is foLlowed, but the photosensitive resin ; used consis~ ~f a mixture of 100 g of a co~lercial cresol-formaldehyde novolac and 25 g of the condensation product of three moles of 6-diazo-5,6-dihydro-5-oxo-l-naphthalenesulfonic chloride wiCh one mole of 2,3,4-trihydroxybenzo-phenone. rhis mixture is dissolved in 250 g oE a solvent mixture containing ~0~ by wei~ht oE 2-ethoxyethanol, 10% by weight of xylene and 10% by weight of n-butyl acetate. The Eormed resin layer has a thickness of 1.4 ~m. After prebaking, it is exposed to UV rays through a mask, the exposure energy being 70 mJ/cm2, then treated with hexamethyldisilazane vapor during 30 minutes at 65C. After development, negative patterns having a residual thickness of 1.2 ~m are obtained, i.e. 86% of the initial resin layer thickness.
Example 3.
The procedure of Example 1 ;s followed, but the photosensitive resin used is prepared by partial esterification of a commercial cresol-formaldehyde novolac with 6-diazo-5,6-dihydro-5-oxo-L-naphthalenesulfonic chloride. 25 g of this resin are dissolved in lO0 g of 2-ethoxyethyl acetate. The formed resin ~.~

layer has a thickness of 1.5 ~m. After prebaking, it is exposed to UV rays through a mask, the exposure energy being 85 mJ/cm , then treated with hexamethyldisilazane vapor Eor 10 minutes at 80C. After development, negative patterns with a residual thickness of 1.4 ~m are obtained, i.e. 93~ of the initial resin layer thickness.
Example 4.
The procedure of Example 1 is followed, butthe photosensitive resin used is the partial esterification product of 6-diazo-5,6-dihydro-5-oxo-1-naphthalenesulfonic chloride with poly(p-vinylphenol). 25 g of this resin are dissolved in 100 g of 2-ethoxyethyl acetate. The formed resin layer has a thickness of 1.7 ~m. After prebaking, it is exposed to UV rays through a mask, the exposure energy being ~5 mJ/cm2, then treated with hexamethyldisilazane vapor for 3 minutes at 125C. The negative patterns obtained after development have vertical side-walls;they have a residual thickness of l.oS ~m, i.e. ~7 of the initial resin layer thickness.
Example _.
This ExampLe shows that the presence of the photoactive compound is essential to obta;n a selective reaction in the exposed areas of the photo-resin layer; it proveg, moreover, that the silicon compound is able to react with the phenoLic poLymer.
The procedure Oe Example I is ~olLowed, but a non-photosensitive resin is used consisti~g soleLy of poly(p-vinyLptlenol~ i.e. without photoactive compoulld. 25 U oE this resin are dissolvecl in lO0 g of 2-ethoxyethyl acetate, and this soLu~.;on i~ spin-coated on siLicon wafers 90 as to obtain a resin Layer oE L.7 ~1111 thickness. ~fter prebaking, this layer is exposed to UV rays through a mask, the exposure energy being L30 mJ/cm2~ The samples are then treatecl in two clifferent ways with hexamethyldisilazane vapor:
a) at 130C during 30 minutes. In this case, development by oxygen-reactive ion etching results in complete removal of the resin layer, without patterns being obtained;
b) at 155C during 10 minutes. In this case, the resin layer is completely resistant to oxygen-reactive ion etching. No noticeabLe decrease of its thickness occurs9 even after prolonged etching. No patterns are obtained.
Example 6 The procedure of Example 1 is followed, but the photosensitive resin used Iq ~ %~
is prepared by partial esterification of the condensation product of p-n-propylphenol and formaldehyde, with 6-diazo-5,6-dihydro-S-oxo-l-naphthalene-sulfonic chloride. 30 g of this resin are dissolved in L00 g of 4-methyl-2-pentanone. The formed resin layer has a thickness of 2.3/um. After prcbaking, it is exposed to UV rays through a mask, the exposure energy being 85 mJlcm , then treated Eor 8 minutes at 115C with hexamethyldisilazane vapor.
Development affords negative patterns with high resolution (lines o~ 0.45 ~
wide and spaces 0.85 /um wide), with vertical side-walls; they have a residual thickness of 2.15 /um, i.e. about 94% of the initial resin layer thickness.
Example 7.
The procedure of Example 1 is followed, but the photosensitive resin used consis~ of a mixture of 10 g of polystyrene and 20 g of the partial esterifica-tion product of 6-diazo-5,6-dihydro-5-oxo-1-naphthalenesulfonic chloride with the condensation product of p-ethylphenol and formaldehyde. This mixture is dissolved in 100 g of cyclohexanone. A layer of 1.8 lum thickness is formed, which after prebaking, is exposed to UV rays chrough a mask, the exposure energy being L10 mJ/cm . This layer is then treated with trimethylchlorosilane vapor for 10 minutes at 100C. After development, high resolution negative patte ms with vertical side-walls and a residual thickness of 1.5 ~m are obtained, i.e. about 85% of the initial resin layer thickness.
E m ~ , The procecture of Example 1 i9 followed, but the photosensitive resin consists of the partial egterification product of 6-diazo-5,6-dihydro-5-oxo-1-naptlthalenesulonic chloride with the conclensation product of cresols and benzaldehyde. 25 g of this resin are dissolved in 100 g of bis(2-methoxyethyl)-ether. The formed resin layer has a thickness of 1.7 lum. After prebaking, it isexposed to UV rays through a mask, the exposure energy being 90 mJ/cm2, then treated for 8 minutes at 125C with hexamethyldisilazane vapor. After development, high resolution negative patterns with vertical side-walls and a residual thickness oE 1.6 /um are obtained, i.e. 94% of the initial resin layer thickness.
Example 9.
The procedure of Example 1 is followedS but the photosensitive ~esin consistsof a mixture of 3 g of poly(N-vinylcarbazole) and 20 g of the partial esterification product of 6-diazo-5,6-dihydro-5-oxo-1-naphthalenesulfonic chloride with the condensation product of l-naphtol and benzaldehyde. This , ;

~2~

mixture is dissolved in 100 g of bis(2-methoxyethyl)ether. The formed resin layer has a thickness of 1.8 ~m. After prebaking on a hot plate at 9;C for 45 seconds, it is exposed to UV rays through a mask, the exposure energy bein~
115 mJ/cm2, then treated for 10 minutes at 1~0C with hexamethyldisiLazane vapor. After development, high resolution negative patterns with vertical side-walls and a residual thickness of 1.o5 lum are obtained, i.e. 92~ of the initial resin layer thickness.
xample 10.
The procedure of Example 1 is followed, but the photosensitive resin is prepared by partial esterification of a copolymer of p-vinylyhenol and p-chlorostyrene, with 5-diazo-5,b-dihydro-5-oxo-1-naphthalenesulfonic chloride.
25 g of this resin are lissolved in L00 g of 5-methyl-2-hexanone. The formed resin layer has a thickness of 1.7 ~m. After prebaking, it is exposea to W rays through a mask, the exposure energy being 85 mJ/cm , then treated for 10 minutes at 125C with hexamethyldisilazane vapor. ~fter development, high resolution negative patterns with vertical side-walls and a residual thickness of 1.6 /um are obtained, i.e. 94% of the initial resin layer thickness Example 11.
The procedure of Example 1 is followed, but the photosensitive resin is prepared by partial esterification of 6-diazo-5,6-dihydro-5-oxo-1-naphthalene-sulfonic chloride with the condensation product of octylphenol and formaldehyde.30 g of this resin are d;ssolved in 100 ~ of cyclohexanolle. The formed resin layer has a thickness of 1.~ ~m. AEter prebaking at 95C for 45 seconds on a hot plate, it is exposed to UV rays through a mask, the exposure energy being 2S 50 ml/cm2, then treated for 10 minutes at 125C with hexamethyldisilazane vapor.
After development, high resolution negative patterns with vertical sidé-walls and a residual thickness of 1.7 /um are obtained, i.e. about 90% of the initial resin layer thickness.
Example 12.
This Example shows that the differentiation which occurs between the irradiated portions and the non-irradiated portions of the pnotosensitive resin,during its exposure to light, is essentially caused by an increase of the diffusion rate of the silicon compound in the irradiated portions of the resin.
This example shows at the same time that the silicon compound penetrates mGre deeply in these irradiated portions the more the applied UV exposure ., .

~7~

dose i~s increased.
Silicon wafe~s with thermal oxide of approximately 120 nm thickness are primed ~ith hexamethyldisilazane as adhesion promotor. As photosensitive resin, the partial esterification product of 6-diazo-5,6-dihydro-5-oxo-1-napnthalene-sulfonic chloride with the condensation product of p-tert-butylphenol and formaldehyde~ is used. 25 g of this resin are dissolved in 100 g cyclohexanone.
The thus obtained solution is spin-coated on the silicon wafers so as to obtain a photosensitive resin layer of 1.5 /um thickness.
The coated wafers are prebaked on a hot plate at 95C for 45 seconds. They are then exposed to ultraviolet radiation, applying diEferent exposure energies on each sample. Samples are thus obtained, which have received an exposure dose respectively of 0~ 13, 25, 3~ and 50 m~/cm2. The exposed wafers are then treated with hexamethyldisila~ane vapor at 125C for 10 minutes.
~he samples prepared in this way are subjected to AUGER Electron Spectroscopy. ~rgon ion sputtering of the resist layer at a rate oE 3 nm/minute enables the dcpth profile of the relative sîlicon concentration in the layer to be obtained. To this end, the intensity of the silicon peak, which is proportional to the concentration of silicon in the resist layer, is monitored a9 a function of sputter time.
It is found that the silicon concentratioll reaches a certain value, whatever the applied exposure dose may be.. ~fter a certain sputter time, the intensity of the silicon p~alc decreases abruptLy and the peak einaLly disappears. This ti~e is clearly a Eullction of the applied UV exposure ~ose. Thus, at exposure doses of 0, 13, 25, 38 and 50 ~/cm2, the silicon peaks disappear respectively after approximately 5, 20, 60, l10 and 160 minutes.
On the other hand, similar samples are prepared, in order to obtain patterns in the layer, subjecting them to the same treatment as described above, but followed by oxygen~reactive ion etching rather than AUGER spectrosco--py. It thus appears that an exposure dose in between 13 and 2- mJ/cm~ is sufficient for good pattern for~ation. Taking into account the sputter rate of 3 nm/minute, it may be concluded that a penetration depth of the silicon compound in the resist layer f approximately lOO nm slows down the etch rate sufficiently to allow complete clearance of the non-irradiated portions. This correlates well with the residual thicknesses of 90 to 95% obtained according the process of the invention.

~2~7~84~

The analysis by AUCER electron spectroscopy thus shows that the diffusion of the silicon compound in the resist is negligible, or insignificant, when it is not subjected to UV radiation (radiation energy = O). This analysis shows also that when the applied e~posure dose increases, the silicon compound penetrates more in depth in the resist layer, whereas the concentration of fixed silicon is independent of the applied exposure dose. This fact is best explained so that the conversion of the resist layer under the effect of the exposure, changes its per~eability and makes it possible for the silicon compound to selectively diffuse in the irradiated portions.

Claims (15)

1. A process of forming a negative pattern in a photoresist layer comprising the following steps:

(a) coating a substrate with a layer of a photosensitive resin comprising a polymer mixed with or bound to a photoactive compound, said layer enabling a silicon compound to diffuse selectively into irradiated portions thereof, on exposure of said portions to visible or ultraviolet radiation;

(b) exposing the photosensitive resin layer to ultraviolet or visible light through a mask to expose only selected portions of the layer;

(c) treating the photosensitive resin layer with a silicon compound, whereby said silicon compound is selectively absorbed into the irradiated portions of the coating and is caused to react with said irradiated portions said silicon compound being at least one silylating agent; and (d) dry developing the thus treated photosensitive resin layer by plasma etching to remove selectively the non-irradiated portions thereof in order to obtain the desired negative pattern.

2. A process according to claim 1, wherein the photoactive compound is a diazoquinone.

3. A process according to claim 2, wherein the diazoquinone is selected from the group consisting of 5-diazo-5,6-dihydro-6-oxo-1-naphthalenesulfonic acid, 6-diazo-5,6-dihydro-5-oxo-1-naphthalenesulfonic acid, 3-diazo-3,4-dihydro-4-oxo-1-naphthalenesulfonic acid, 4-diazo-3,4-dihydro-3-oxo-1-naphthalenesulfonic acid, 3-diazo-3,4-dihydro-4-oxo-1-benzenesulfonic acid, a corresponding carboxylic acid, a derivative thereof and a mixture of at least two of -the aforesaid compounds.

4. A process according to claim 1, wherein the polymer is a phenolic polymer.

5. A process according to claim 4, wherein the phenolic polymer is selected from - the condensation products of a phenol or a naphthol or a derivative thereof ring-substituted by an alkyl or aryl radical or by a halogen atom, with an aliphatic or aromatic aldehyde, which can be substituted by a halogen atom, - the poly(vinylphenols), the phenolic group of which can be substituted by an alkyl or an aryl radical or a halogen atom, - the copolymers of a vinylphenol with an ethylenically unsaturated compound, and - mixtures of the aforesaid polymers between them or with other aromatic polymers.

6. A process according to claim 1, wherein the photosensitive resin further comprises a dye.

7. A process according to claim 1, wherein the silicon compound is an easily vaporizable silylating agent.

8. A process according to claim 1, wherein the silicon compound is a silylating agent selected from tetrachlorosilane, trimethylchlorosilane, dimethyldichlorosilane, methyltrichlorosilane, trimethylbromosilane, trimethyliodosilane, triphenylchlorosilane, hexamethyldisilazane, hepta-methyldisilazane, hexaphenyldisilazane, 1,3-bis(chloromethyl)-1,1,3,3-tetramethyldisilazane, N-trimethylsilylimidazole, N-trimethylsilylacetamide, N-trimethylsilyldiethylamine, N-trimethylsilyldimethylamine, hexamethyl-silanediamine, N,O-bis(triethylsilyl)acetimide, N,N'-bis(trimethylsilyl)urea, N,N'-diphenyl-N-(trimethylsilyl)urea and mixtures of at least two of these compounds.

9. A process according to claim 8, wherein the silylating agent is hexamethyldisilazane.

10. A process according to claim 1, wherein the treatment of the photosensitive resin layer with the silicon compound is carried out at a temperature of from -20 to 150°C.

11. A process according to claim 10, wherein said treatment is carried out at a temperature between 60 and 140°C.

12. A process according to claim 1, wherein the time of the treatment of the photosensitive resin layer with the silicon compound is in the range of a few seconds to one hour.

13. A process according to claim 12, wherein said treatment time is between 1 and 45 minutes.

14. A process according to claim 1, wherein the dry development is carried out by oxygen-reactive ion or oxygen plasma etching.

15. An integrated semiconductor circuit made by means of the process according to claim 1.