Radiation chemistry and the lithographic performance of chemical amplification resists formulated from poly(4-epoxystyrene-stat-styrene) and a photoacid generator Richard G. Jones,* Gerard P-G. Cordina and Julian J. Murphy Centre forMaterials Research, Department of Chemistry, University of Kent, Canterbury, Kent, UK CT2 7NH Copolymers of styrene and 4-epoxystyrene in formulations with triphenylsulfonium hexafluoroantimonate as a photoacid generator undergo a crosslinking reaction by a chain mechanism when irradiated with 20 keV electrons and hence act as negativeworking electron-beam resists.The copolymers have been prepared by a free radical mechanism over the entire composition range and the lithographic performance of the materials has been evaluated. The sensitivities of the resist formulations are shown to correlate with the epoxystyrene content of the copolymers in accordance with a simple model of the radiation chemistry of the system.Contrast variations are explained in terms of the statistical structure of the copolymer and it is demonstrated that at low epoxystyrene contents the systems behave in accordance with an unsensitised single-stage crosslinking mechanism.The copolymers with epoxystyrene contents greater than ca. 8% have such high lithographic sensitivities and show such a good tolerance of processing variations as to commend the optimisation of their performance with a view to their subsequent application as electron-beam resists. The use of so-called chemical amplification resists is becoming with good resistance to oxygen plasma, and excellent tolerance to process variation.The measured contrasts for these systems more commonplace in microlithography as their enhanced response, achieved through some radiation-induced chain tend to be low but this has little effect upon the resolution that can be attained, as features with dimensions as low as mechanism, leads to the sensitivity necessary for the economic exploitation of many modern writing tools.However, many of 0.1 mm have been demonstrated.12 One drawback of the Hatzakis systems is that it is not those presently available are multicomponent systems which have little tolerance of process variations. Typically, even a possible to prepare a range of polymers with structures sufficiently controlled as to allow a structure process optimisation.very small change in the post exposure bake (PEB) temperature or time, required to realise the chemical amplification, might Novalac resins are not amenable to molecular mass control and the extent of their subsequent epoxidation cannot be drastically alter the performance of the resist.1 Some systems are also sensitive to environmental contaminants such as controlled. Here we report the microlithographic performance of electron-beam resist formulations based on a series of airborne amines.2 Such problems make for great difficulty in achieving reproducibility.Positive-working systems commonly poly(4-ethenylphenyloxirane-stat-styrene) copolymers. These can be viewed as epoxy novalac analogues, but their synthesis depend on the deprotection of a functional group by radiationgenerated acid or base, which renders the exposed regions of by radical polymerisation allows polymers of controlled molecular mass and epoxy content to be produced.the resist soluble in an aqueous-based development process. Negative-working systems usually rely on a chain crosslinking process which again is most commonly initiated by acid or base formed during exposure.Thus, two components, a polymer and a photoacid generator (PAG) such as an onium salt, are the minimum required in such formulations but it is by keeping to this minimum that processing difficulties can be contained. Amongst the useful functional groups that might be employed in negative-working systems is the oxirane group.This group, which can undergo a radiation-induced chain crosslinking reaction, confers a high sensitivity upon polymeric resists that contain it. Oxirane functionalised polymers such as epoxidised poly(butadiene) or poly(isoprene)3 and poly(glycidyl methacrylate)4,5 were first investigated as electron-beam resists over twenty years ago. Unfortunately, the poor thermal stability of these aliphatic materials at the temperatures commonly encountered in lithographic processing has limited their exploitation.In terms of processibility, even the GMC series of resists (copolymers of glycidyl methacrylate and 3-chlorostyrene) with their considerable aromatic content fall far short of other copolymer resists based on poly(styrene) structures.6–9 More recently, Hatzakis et al.have reported the lithographic performance of epoxy novalac resins in formulations with triphenylsulfonium hexafluoroantimonate as a PAG, as photo-, electron-beam and X-ray resists10,11 as depicted in Scheme 1. Being aromatic materials, these display a number of the desirable properties associated with poly(styrene)-based resists Scheme 1 such as excellent film forming ability, thermal stability allied J.Mater. Chem., 1997, 7(3), 421–427 421DSC-7 differential scanning calorimeter operating at a scan rate of 10°C min-1. Resist solutions were formulated by dissolving the polymers in propylene glycol methyl ether acetate to produce 15% solutions. Unless otherwise stated, the PAG, triphenylsulfonium hexafluoroantimonate, was also dissolved to 4%.The resist solutions were filtered through 0.5 mm Millipore filters and spun directly onto 3 inch† silicon wafers using a Headway EC-101 spinner to produce films of thickness in the range 0.5 mm. Lithographic assessment was accomplished using a CambridgeInstruments EBMF10.5 electron-beam lithography tool operating at a 20 kV accelerating potential. After exposure the wafers were cut into segments which were then either developed immediately, or baked in an oven for 3 min at 90°C prior to development.For process latitude studies, the postexposure bake (PEB) times and temperatures were increased to up to 9 min and 150 °C respectively. Pattern development was accomplished by immersing the wafer segments in methyl isobutyl ketone (MIBK) for 60 s, rinsing in isopropyl alcohol‡ and drying in a stream of nitrogen.Film thickness measurements before and after exposure were measured using a Nanospec/AFT 210 Film Thickness System. All thicknesses were normalised to the original spun thickness. Sensitivities were estimated both as the gel dose (D0) and as the dose corresponding to 50% thickness remaining after development Scheme 2 (D0.5).Lithographic contrasts, c, were calculated from D0.5 and D0 using eqn. (1). Experimental c=1/[2log(D0.5/D0)] (1) Materials Conversion of the incident radiation dose in mC cm-2 to absorbed radiation dose in Mrad was carried out in accordance Styrene (S) was supplied by Aldrich Chemical Co. Ltd. and with the method developed by Novembre and Bowmer.15 For distilled under reduced pressure at 40–50 °C prior to use. 4- a resist film having an approximate thickness of 0.5 mm and a Ethenylphenyloxirane (4-epoxystyrene, ES) was prepared in density of 1 g cm-3, it can be shown that a conversion factor accordance with the method of Truxa and Suchopa�rek13 shown of about 2 Mrad per (mC cm-2) for 20 keV electrons applies. in Scheme 2, and distilled under reduced pressure (65°C at This is the value that has been used in this study. 0.1 mmHg) from sodium hydroxide prior to use. Copolymerisations initiated by 2,2¾-azoisobutyronitrile were carried out in toluene solution inside stoppered boiling tubes Results at 70°C, under dry argon. The polymers were precipitated in The effect of copolymer composition a large excess of chilled methanol, reprecipitated from toluene solution, filtered and dried under vacuum.Precise experimental All of the 4-epoxystyrene/styrene copolymers have glass trans- details of similar copolymer preparations have been reported ition temperatures that are sufficiently high to meet the elsewhere.14 demands of lithographic processing. These, together with the structural and lithographic parameters of a series of resists Apparatus and procedures formulated from the copolymers and triphenyllfonium hexa- flouroantimonate, are recorded in Table 1.Representative con- Copolymer compositions were established from integration of trast curves are depicted in Fig. 1. With the exception of P1, 270 MHz 1H NMR spectra obtained at room temperature, in the number average molecular masses (M� n) of the copolymers CDCl3 solution, using a JEOL JNM-GX270 spectrometer.were controlled at about 15000. The asymptotic departure of Chemical shifts are relative to SiMe4 . Molecular masses and polydispersities were determined as linear polystyrene equivalents using HPLC equipment supplied † 1 in=2.54 cm. by Polymer Laboratories Ltd. and equipped with a mixed-bed ‡ The solvent development system and development times have not 5 mm PLgel column.Glass transition temperatures were deter- been optimised and were chosen simply because they are regularly used with negative-working resists based on poly(styrene) derivatives. mined under a nitrogen atmosphere using a Perkin-Elmer Table 1 Structural parameters and lithographic properties of 4-ES/S copolymer resists in formulations containing 4% triphenylsulfonium hexafluoroantimonate baked resists unbaked resists 4-ES polymer (%) M2 Ma PD Tg/°C D0.5 D0 c Gn D0.5 D0 c Gn P1 100 15 500 9700 1.56 140 0.190 0.045 0.80 160 0.263 0.051 0.70 110 P2 30 30 400 17000 1.79 118 0.70 0.18 0.85 33 0.77 0.18 0.85 28 P3 18 28 000 15100 1.85 113 1.14 0.35 0.97 24 1.45 0.41 0.91 19.3 P4 12 25 800 14500 1.76 104 2.55 1.00 1.23 9.1 4.51 1.30 0.92 9.1 P5 10 28 400 16300 1.75 104 3.84 1.27 1.0 7.3 7.93 2.67 1.1 5.7 P6 8 26 100 14100 1.83 97 6.3 2.5 1.2 3.9 14.0 7.5 1.8 3.1 P7 4 25 100 14500 1.72 93 56.5 33.5 2.2 1.0 65.5 41.0 2.5 1.0 P8 2 24 700 14400 1.71 93 94.3 66.5 3.3 0.9 99.6 72.4 3.6 0.9 422 J.Mater. Chem., 1997, 7(3), 421–427Table 2 The lithographic properties of 4-ES/S copolymer P6 in resist formulations over a range of triphenylsulphonium hexafluoroantimonate loadings baked resists unbaked resists %PAG D0.5 D0 c D0.5 D0 c 0 — — — 37.9 23.1 2.30 0.25 17.1 3.3 0.70 23.5 10.5 1.43 0.5 14.8 3.6 0.81 22.4 10.6 1.54 1.0 10.1 2.6 0.85 21.6 11.0 1.71 2.0 7.6 2.0 0.86 21.5 10.6 1.63 4.0 6.3 2.5 1.25 14.0 7.5 1.84 Fig. 2 Variation of lithographic sensitivity with copolymer composition for the single-stage crosslinking resists.The progressive inclusion of up to 30% epoxystyrene in this copolymer causes the contrast to drop to 0.85 from an initial value of 3.6 for the homopolymer, polystyrene. The contrast of the corresponding chlorostyrene/methylstyrene system of comparable molecular mass, over the same composition range, holds at ca. 2.5.The effect of post exposure bake From Fig. 1, it is evident that PEB temperatures greater than 90°C and durations longer than 3 min are unnecessary. The effect of baking the exposed images is to decrease the sensitivity parameters, corresponding to an increased sensitivity, but not to an equivalent extent for both of D0 and D0.5. Thus, though the chemical amplification within these systems is evident it is not uniform at all compositions.For resists containing up to 10% epoxystyrene, baking results in a lesser contrast. At higher epoxystyrene contents the contrasts are marginally enhanced, baking having little or no effect on D0 at the higher epoxystyrene contents. At an epoxystyrene content of 30% baking no longer has any noticeable effect, though one is again apparent for the formulation involving the poly(4-epoxystyrene) homopolymer.Only those resists containing less than 10% epoxystyrene show PEB effects that are comparable to the amplifications observed for the epoxy novolac systems.16 Thus, for copolymer compositions that are lithographically useful, P2 to P6, the present resist system is notably more tolerant of processing variations than are the epoxy novolac Fig. 1 Representative contrast curves of PAG sensitised resists: formulations. Whilst not asserting a PEB to be unnecessary, (a) polymer P3 baked at 120 °C for 0 (&), 180 ($), 360 (+) and 540 s it is nonetheless apparent that to all intents and purposes the (,); (b) polymer P6, unbaked (&) and baked at 90 ($), 120 (+) and 150 °C (,) for 180 s maximum attainable level of crosslinking is achieved within the unbaked systems.the contrast curves from the dose axis at low doses, a character- The effect of variations in PAG loading istic of resist systems that undergo a chain crosslinking process, is discernible for all of the copolymers P1 to P5 and for the Fig. 3 depicts the variation of the sensitivity of the copolymer P6 with PAG concentration.A sharp increase in sensitivity baked system of P6. The contrast curves for the remaining copolymers display a more acute departure from the dose axis, results from the inclusion of even the smallest amounts of the PAG, but for the unbaked systems the effect then seems to be similar to that which is shown later in Fig. 5. The lithographic sensitivities of the resists increase sharply with the epoxide more or less invariant up to a content of about 2%.In contrast, for the baked systems a steady increase in sensitivity is evident content but there is an accompanying loss of contrast. A plot of sensitivity against mole fraction of 4-epoxystyrene within over the composition range investigated. Accordingly, the greatest chemical amplification is evident at a PAG content of the copolymers, shown in Fig. 2, levels off acutely at an epoxide content of ca. 10%; the sharpness of the knee in the curve is 2%. The corresponding contrast variations are shown in Fig. 4. For both the unbaked and baked systems, a small but steady striking, being far more pronounced than those observed for the one component, single-stage crosslinking resists such as increase with PAG loading is evident following the initial sharp drop from the contrast of the unloaded system.the chlorostyrene/methylstyrene copolymer systems.14 Such a sensitivity variation might be expected for a resist system that The relatively insensitive system in which no PAGis included has a contrast value of 2.3. This is characteristic of the optimum crosslinks through a sensitised chain reaction.However, the contrast variation is also notably different from that observed that can be obtained from a single-stage crosslinking resist J. Mater. Chem., 1997, 7(3), 421–427 423the chemical amplification principle.16 s=e-ar (1) The sol fraction remaining in the irradiated polymer is s, and r is the absorbed dose in Mrad.The parameter a is given by eqn. (2) in which G is the radiation chemical yield for chain initiation (assumed to be the number of protons deriving from the photoacid generator that are effective in initiating the crosslinking process per 100 eV of energy absorbed), n is the number of polymer chains that are linked in the ensuing crosslinking, and M� n is the initial number-average molecular mass of the polymer.a=1.04×10-6GnM� n (2) Rearranging eqn. (1) gives eqn. (3). ln s=-ar (3) This simple model predicts that plots of ln s vs. r would be Fig. 3 Variation of lithographic sensitivity of polymer P6 with PAG linear with zero intercept, and knowledge of M� n allows the loading, ($) unbaked and (&) baked for 180 s calculation of the product parameters, Gn, from their slopes.These values are listed in Table 1 and the chemical amplifi- cation effect is evident from a comparison of the values for the baked and unbaked resists. The effect is much more pronounced for the systems of lithographic interest in which the epoxystyrene content exceeds ca. 10%. It is usual, when discussing the lithographic response of negative-working resists, to employ equations expressed in terms of the gel fraction, g, since this can be related directly to the thickness, t, of the resist layer that remains after development, normalised to its original spun thickness, t0.Thus, since s=1-g and g=t/t0 it follows that s=1-t/t0. Thus, the lithographic data can readily be plotted in accordance with eqn. (3). Fig. 6 shows representative plots derived from the primary data of the contrast curves shown in Fig. 1 for the Fig. 4 Variation of lithographic contrast of polymer P6 with PAG loading, ($) unbak&) baked for 180 s with anormal (most probable)molecular mass distribution.17,18 The contrast curve for this resist is shown in Fig. 5. Discussion We have previously adapted the Charlesby theory of polymer network formation by an uninhibited radiation-induced chain reaction19 to derive a simple model, shown as eqn.(1), for the lithographic performance of negative-working resists based on Fig. 6 Charlesby plots of data of contrast curves represented in Fig. 1 Fig. 5 Contrast curve for the unsensitised resist based on polymer P5 424 J. Mater. Chem., 1997, 7(3), 421–427resist formulations P2 and P6.The curves are linear up to content of the copolymers. A linear regression analysis of the log 10D0.5 vs. log10x plot of Fig. 7 reveals a slope of -1.99. normalised thicknesses remaining in excess of 0.7, beyond which they show a positive deviation from linearity. The onset The plot of D0.5 vs. x-2 of Fig. 8 shows the accuracy of the correlation over the data points of lithographic interest. of this deviation generally appears at lower doses for the unbaked systems. It is attributed to localised reductions in It was previously shown that, according to the simple model represented by eqn.(1), the contrast, c, should always be 0.8 epoxide groups to levels at which the active centres of crosslinking become effectively trapped at voids. whilst, at the same time, it was acknowledged that for the epoxy novolac systems this apparently represented a minimum With the exception of the plot for the poly(4-epoxystyrene) homopolymer, P1, the plots for copolymers with epoxystyrene value.16 From Table 1 it is apparent that this same minimum value applies to these systems.However, it is clear that the contents greater than 8% displayed small positive intercepts ranging from about 0.05 to ca. 0.25. Within this range there epoxystyrene homopolymer is the only one to which the value 0.8 actually applies, and that with decreasing epoxystyrene was no apparent order to indicate a correlation with the epoxystyrene content of the copolymers, but it can nonetheless content within the copolymers, the contrast steadily increases.If eqn. (4) is used as the premise for the derivation of an be inferred that the model for the copolymers should be modified to accommodate a small correction factor, K, as expression for contrast, taken to be the slope of the contrast curve at the dose required for 50% gelation, then it is readily shown in eqn. (4), the magnitude of which would be no greater than 1.5. The significance that can be attached to the need for shown that eqn.(7) holds. this correction is that the apparent gel dose of the copolymer dg/d log10r=2.303 Kar e-ar (7) resists is marginally suppressed, an effect that can be attributed to the irregular distribution of crosslinking points along the By substituting r=R0.5, it follows that contrast is given by eqn. (8). polymer chain that results from a statistical copolymerisation process.The irregular distribution is itself a source of structural c=0.8+1.15 ln K (8) voids at which active centres might be trapped. This concept accords with the Charlesby models of radiation-induced Comparison of eqn. (8) with Fig. 9, a plot of c vs. the reciprocal of the mol fraction, x, of epoxystyrene in the copolymer, crosslinking reactions in which the processes are eventually terminated at centres from which further growth is inhibited indicates that there is a rough correlation between K and composition.The slope of this plot when it is equated to 1.15 for any one of several reasons, and which are represented by the generalised formula of eqn. (4). ln K leads to eqn. (9). K=e0.047/x (9) s=Ke-ar (4) For ease of reference whilst following the ensuing kinetic argument, an adaptation of Scheme 1 to represent radiationinduced crosslinking in the copolymer resists is shown in Scheme 3, within which Ox represents an oxirane group, C+ propagating carbonium ions and i is the associated extent of crosslinking, which in the limit assumes the average value n, i.e. 1in. PAG e-T H+ H++Ox C0+ CA G Ci-1++Ox Ci+ CA Scheme 3 Although values of G have not been determined it is reasonable to assume that they are proportional both to the PAG loading and to the concentration of epoxide groups since the addition of a proton to an oxirane has a significant energy of activation of 25 kJ mol-1.20 The PAG loadings have been Fig. 7 Log–log plot of the variation of lithographic sensitivity with maintained at 4% throughout the series of resists represented copolymer composition in Table 1.G is therefore proportional to the epoxystyrene content within the copolymers. The length of the ensuing chain crosslinking reaction, n, is again proportional to the concentration of epoxide groups. It follows that the product parameter, Gn, should vary linearly with the square of the mol fraction, x, of epoxystyrene in the copolymers, i.e.Gn=kx2 where k is a proportionality constant. Although the above conclusions can be applied directly to the values of Gn listed in Table 1 it is more informative to relate them back to the lithographic parameters. Thus, substituting s=0.5 at r=R0.5=2D0.5 into eqn. (4) and rearranging leads to eqn. (5). D0.5= ln(0.5/K) 2.08×10-6Gn (5) If K is taken to be 1 then eqn.(5) reduces to eqn. (6). D0.5=3.33×105/Gn=(3.33×105/k)x-2 (6) Eqns. (5) and (6) both indicate that the lithographic sensitivity, Fig. 8 Variation of lithographic sensitivity with copolymer composition plotted in accordance with the form of eqn. (6) D0.5, should vary inversely with the square of the epoxystyrene J. Mater. Chem., 1997, 7(3), 421–427 425mer with a normal molecular mass distribution.19 The broken line represented in Fig. 5 applies to a system which undergoes a radiation-induced chain crosslinking process but which is of the same sensitivity. The more satisfactory fit of the data points to the former model than to the latter is clear. Conclusions It has been demonstrated that statistical copolymers of styrene and 4-epoxystyrene, in formulations with an onium salt photoacid generator, make very sensitive electron-beam resists, even when the epoxystyrene content of the copolymers is quite low.The mathematical model previously developed to correlate the lithographic data for the so-called ‘Hatzakis’ resist based on formulations of epoxy novolacs and a PAG has been extended for application to these systems and a possible explanation for contrast variations has followed. At one Fig. 9 Plot of the variation of contrast with the reciprocalof copolymer extreme of copolymer composition, the systems have been composition shown to behave as chemical amplification resists in an exactly analogous manner to the ‘Hatzakis’ systems, whilst at the other end of the composition scale, their lithographic perform- For copolymers containing 8–30% of epoxystyrene, values of ance is more in accordance with that of an unsensitised, single- K calculated from eqn.(8) are of about the same magnitude step crosslinking system. The chemical amplification effect as those estimated from the intercepts of the Charlesby model within those resists that are of lithographic interest is not as plots though, as the epoxystyrene content drops to lower pronounced as has been observed for the ‘Hatzakis’ systems, values, the value of K, estimated from eqn.(9), increases by an attractive feature with regard to the reproducibility of the an order of magnitude. However, at such low contents of effects of any post exposure bake that would be applied during epoxystyrene it is arguable that the system bears little resem- processing.The wide potential for the further development of blance to the epoxy novolac system upon which it was the resist has been demonstrated. The optimisation of its modelled. Copolymers of M� n approximately 15000 have lithographic performance will need to take into account mol- degrees of polymerisation lying between 100 (epoxystyrene ecular mass considerations and the need to seek an appropriate homopolymer) and 145 [poly(styrene)], corresponding to num- solvent development system but through the present study it bers ofxirane groups per molecule in the range 100 to 11 for has again been shown that resists designed around copolymers copolymers in the lithographically interesting composition offer a flexibility that cannot be obtained from homopolymer range.The remaining copolymers of the series, P7 and P8, systems. contain even fewer oxirane groups, ca. 5 and 2 respectively. Not surprisingly these systems are likely to behave in a similar fashion to poly(styrene) for which sensitisation by a PAG is We thank the EPSRC for the award of a Research Studentship an irrelevance; indeed their lithographic parameters resemble (J.J.M.) and of a Postdoctoral Research Fellowship (G.P-G.C.) those of low sensitivity single-stage crosslinking systems.From We also gratefully acknowledge the assistance of the staff of Table 2, it is evident that the resist based on copolymer P6 the Central Microstructure Facility of the Rutherford Appleton without the inclusion of PAG displays a greater sensitivity Laboratory, in particular Mr Ejaz Huq, for facilitating the than either of these resists.The three resists actually make a microlithographic evaluations. series in which the sensitivities of the unbaked systems increase with increasing epoxystyrene content. Fig. 10 represents a plot of the contrast curve of Fig. 5, which is for the resist P6 References without PAG, in accordance with the Charlesby model for a 1 F.M. Houlihan, E. Reichmanis, L. F. Thompson and radiation-induced single-stage crosslinking reaction of a poly- R. G. Tarascon, in Polymers in Microlithography, ed. E. Reichmanis, S. A. MacDonald and T. Iwayanagi, ACS Symp. Ser., Am. Chem. Soc., 1989, 39, 412. 2 S. A. MacDonald, N. J.Clecak, H. R. Wendt, C. G. Wilson, C. D. Snyder, C. J. Knors, N. B. Deyoe, J. G. Mathews, J. R. Morrow, H. E. McQuire and S. J. Holmes, Proc. SPIE, 1991, 2, 1446. 3 T. Hirai, Y. Hatano and S. Nonogaki, J. Electrochem. Soc., 1971, 118, 669. 4 Y. Taniguchi, Y. Hatano, H. Shiraishi, S. Horigome, S. Nonagaki and K. Noraoka, Jpn. J. Appl. Phys., 1979, 18, 1143. 5 L. F. Thompson, J. P. Ballantyne and E.D. Feit, J. Vac. Sci. T echnol., 1975, 12, 1280. 6 S. Imamura, J. Electrochem. Soc., 1979, 126, 1628. 7 R. G. Tarascon, M. A. Hartney and M. J. Bowden, inMaterials for Microlithography, ed. L. F. Thompson, C. G. Wilson and J. M. J. Fre� chet, ACS Symp. Ser. 266, Am. Chem. Soc.,Washington, DC, 1984. 8 M. A. Hartney, R. G. Tarascon and A. E. Novembre, J. Vac. Sci.T echnol. B, 1985, 3, 360. 9 R. G. Jones, Y. Matsubayashi, P. Miller Tate and D. R. Brambley, J. Electrochem. Soc., 1990, 137, 2820. Fig. 10 Data taken from the contrast curve of Fig. 5 plotted in 10 K. J. Stewart, M. Hatzakis, J. M. Shaw and D. E. Seeger, J. Vac. Sci. T echnol. B., 1989, 7, 1734. accordance with the Charlesby model for a single-stage crosslinking reaction (—) compared with a similar representation for a system of 11 M. Hatzakis, K. J. Stewart, J. M. Shaw and S. A. Rishton, J. Electrochem. Soc., 1991, 138, 1076. the same sensitivity which undergoes a chain crosslinking process (,) 426 J. Mater. Chem., 1997, 7(3), 421–42712 P. Argitis, I. Raptis, C. J. Aidinis, N. Glezos, M. Baciocchi, 17 Ll. G. Griffiths, R. G. Jones and D. R. Brambley, Polym. Commun., 1988, 29, 173. J. Everett and M. Hatzakis, J. Vac. Sci. T echnol. B., 1995, 13, 3030. 13 R. Truxa and M. Suchopa�rek,Makromol. Chemie, 1990, 191, 1931. 18 D. R. Brambley, R. G. Jones, Y. Matsubayashi and P. Miller Tate, J. Vac. Sci. T echnol. B, 1990, 8, 1412. 14 R. G. Jones, P. C. Miller Tate and D. R. Brambley, J.Mater. Chem., 1991, 1, 401. 19 A. Charlesby, Atomic Radiation and Polymers, Pergamon Press, Oxford, 1960. 15 A. Novembre and T. N. Bowmer, in Materials for Microlithography, ed. L. F. Thompson, C. G. Wilson and 20 Y. I. Estrin and S. G. Entelis, Polym. Sci. USSR, 1968, 10, 3006; 1969, 11, 1286. J. M. J. Fre� chet, ACS Symp. Ser. 266, Am. Chem. Soc.,Washington, DC, 1984. 16 P. C. Miller Tate, R. G. Jones, J. Murphy and J. Everett, Paper 6/06322E; Received 13th September, 1996 Microelectronic Eng., 1995, 27, 409. J. Mater. Chem., 1997, 7(3), 421–4