Evaluation of halomethylated poly(methylphenylsilane)s as electron-beam resists Simon J. Holder, Richard G. Jones* and Julian J. Murphy Centre for Materials Research, Department of Chemistry, University of Kent at Canterbury, Kent, UK CT 2 7NH Polysilane analogues of halomethylated poly(styrene)s, chloromethylated and bromomethylated poly(methylphenylsilane), have been prepared from the parent polymer by reaction with the appropriate halomethyl methyl ether.The polymers undergo a singlestage crosslinking reaction when irradiated with 20 kV electrons. As electron beam resists they operate in negative-working mode but their performance is poor in comparison to the corresponding poly(styrene) derivatives. The low lithographic sensitivities and attainable contrasts are shown to arise as a consequence of a competitive chain scission reaction which in the case of the bromomethylated system increases with increasing bromomethyl content.The radiation chemistries of the systems are rationalised in terms of modifications of the crosslinking and scission mechanisms that are thought to operate in the corresponding resists based on poly(chloromethylstyrene-stat-styrene). Since the discovery of tractable poly(diorganosilanes) in the of so-called CMS negative-working resists.10–18 Halogenated 1970s,1,2 a number of potential applications have been pro- resists derived from the homopolymer, PMPS, can be conposed.Interest in these unique polymers stems not only from sidered no less appropriate for comparison with the CMS the catenated silicon backbone which has led to one of their structures.Accordingly, in the present study, chloromethylated most important commercial applications, namely as ceramic and bromomethylated poly(methylphenylsilanes) over a range precursors,3 but also from the unusual conjugation of the s- of compositions were synthesised with the intention that their bonds of the backbone which gives rise to electron delocalis- application within e-beam lithography might be assessed, and ation. This has resulted in the polymers being assessed for a an understanding gained of the underlying radiation chemical number of electrical and optical applications.4,5 processes that occur within electron beam resists that might Arising from the direct photosensitivity of the silicon back- be based on PMPS and its derivatives.bone of polysilanes, one of their most promising potential uses is as positive-working resists in microlithography. An important advantage that they possess over carbon-based polymers for such applications is a high silicon content, imparting the O2 plasma and reactive-ion etch resistance required in multilayer microlithographic processing.Accordingly, a large number of organopolysilanes have been tested as photoresists although none have found wide application.6 However, future generations of very large scale integrated (VLSI) circuitry will require microlithographic resolutions that are beyond those Experimental that can be attained using conventional resists and photoprocessing. Typically, resolutions of less than 0.25 mm will have Apparatus and procedures to be routinely achieved.To this end electron-beam (e-beam) The polymer structures were characterised by 1H, 13C and 29Si lithography finds increasing application in the production of NMR, FTIR and UV spectroscopy. FTIR and NMR spectra application specific integrated circuitry (ASIC) and photomwere obtained using ATI Mattson Genesis Series FTIR and asks with target resolutions down to 0.1 mm.7 Thus, it is as JEOL JNM-GX270 NMR spectrometers, respectively. NMR electron beam resists that polysilanes might find a useful role samples were prepared as solutions in CDCl3 and chemical if they can be tailored to suit the requirements of this imaging shifts are quoted in relation to SiMe4.Cr(acac)3 was used as technique, which must be performed with structures which an internal relaxation agent to record the 29Si NMR spectra.would not release silicon-containing fragments to the gas UV spectra were obtained from ca. 10-4 mol dm-3 CH2Cl2 phase, that might contaminate the writing tool, during the solutions using a PU 8740 UV–VIS scanning spectrophoto- lithographic process. meter. Thermal analyses of the polymer samples were per- Poly(methylphenylsilane) (PMPS) is the cheapest tractable formed under a nitrogen atmosphere on a Perkin-Elmer DSC7 homopolysilane available in relatively high yields from the dierential scanning calorimeter operating at scan rates of Wurtz-type condensation polymerisation of dichloroorganosil- 10 K min-1.anes. However, positive-working behaviour (i.e.polymer chain Molecular weights of the polymers were obtained relative degradation) with only a very low sensitivity has been preto polystyrene standards using HPLC equipment (Polymer viously observed for this polymer upon e-beam exposure, Laboratories) with a 30 cm×10 mm PLgel mixed-D column. disqualifying its use as a possible resist in an unmodified form.8 The eluent was THF and determinations were carried out at The introduction of chloromethyl groups onto the pendant a flow rate of 2 ml min-1 at ambient temperature using a UV phenyl of poly(methylphenylsilane)-co-(dimethylsilane) has detector. It has been reported4 that the molecular weights of been shown to result in a marked increase in the sensitivity of polysilanes as determined by size exclusion chromatography this polymer towards negative-working behaviour.9 Arising relative to polystyrene standards are likely to be too low by a from its structural similarity to polystyrene, the trivial name factor of ca. 2.3. However, recent studies19 have indicated that poly(silastyrene) has been coined for poly(methylphenylsilane)- the dierence in the hydrodynamic volumes of polystyrene and co-(dimethylsilane).It follows that its chloromethylated deriv- PMPS is very small. It is also unlikely that relatively low atives might reasonably be compared to chloromethylated polystyrenes which are examples of the well-characterised class loadings of halomethyl groups would substantially aect the J. Mater. Chem., 1997, 7(9), 1701–1707 1701hydrodynamic volume of the PMPS. Accordingly, it is assumed dierent from the remaining halomethylated polymers. Both series of copolymers were isolated from samples of halomethyl- that the molecular weights as determined by size exclusion chromatography are suciently accurate for the purposes of ation reaction mixtures taken at various times during extended reactions.The polymer structural and thermal parameters are the present study.Resist solutions were formulated by dissolving the copoly- given in Table 1. mers in chlorobenzene to produce 20% w/v solutions. The BPMPS solutions were filtered through 0.5 mm Millipore filters and spun directly onto 3 inch silicon wafers using a Headway ECdH( CDCl3) -1.1 to 0.6 (vbr m, CH3Si), 3.0 (br m, CH3O), 4.3 101 spinner and prebaked at 120 °C for 30 min to produce (br s, ClCH2), 6.0–7.7 (vbr m, aromatic).dC(CDCl3) -8.0, good quality uniform films with a thickness of approximately -7.5, -6.9, -6.0 (br m, CH3Si), 33.5 (s, CH2Br), 50.5 (s, 1 mm. Exposure was performed using a Cambridge Instruments CH3O), 126.8, 126.5, 134.2, 135.7 (m, aromatic). dSi(CDCl3) EBMF-10.5 electron beam lithography tool operating at 20 kV -41.1, -39.9, -39.2 (br t, SiCH3Ph), 8.7 (s, SiOCH3), 14.6 accelerating potential.Development was accomplished by (d, SiCl). nmax(thin film)/cm-1 3065, 3046, 3014 (C–H stretch, immersing a wafer fragment in methyl isobutyl ketone (MIBK) aromatic), 2958, 2893, 2868 (C–H stretch, aliphatic), 1426 (C–C for 60 s, followed by a 30 s rinse in isopropyl alcohol (IPA). ring stretch, Si–phenyl), 1394 (CH2Cl bend), 1245 (C–H bend, This development procedure was not optimised but was chosen CH3Si), 1095 (C–C ring stretch+Si–C stretch, Si-phenyl), 782, to follow established practice for the corresponding halomethy- 754, 730, 697, 664 (various C–C ring stretch+Si–C stretch), lated poly(styrene) resists.15,16 Developed wafers were dried in 605, 619 (Si–Br stretch), 465 (Si–Si stretch).lmax/nm (e) 239 a filtered stream of nitrogen. Film thickness measurements (5700), 279 (3800), 341 (7900). before and after exposure were taken using a Nanospec/AFT 210 film thickness system. Sensitivities were estimated as the CPMPS dose corresponding to 50% thickness remaining after development (Dn0.5). All resist thicknesses were normalised to the dH(CDCl3) -1.2 to 0.6 (vbr m, CH3Si), 3.0 (br m, CH3O), 4.4 original spun thickness.Lithographic contrasts (c) were calcu- (br s, ClCH2), 6.0–7.6 (br m, aromatic). dC(CDCl3) -8.0, lated from Dn0.5 and the gel dose (Dn0) using eqn. (1). -7.3, -6.8, -5.9 (br m, CH3Si), 46.0 (s, CH2Cl), 50.6 (s, CH3O), 126.7, 134.3, 135.8 (m, aromatic). dSi(CDCl3) -41.0, c=1/[2 log (Dn0.5/Dn0)] (1) -39.9, -39.1 (br t, SiCH3Ph), 8.7 (s, SiOCH3).nmax(thin film)/cm-1 3047, 3012, 2991 (C–H stretch, aromatic), 2958, 2893, 2868 (C–H stretch, aliphatic), 1427 (C–C ring stretch, Materials Si–phenyl), 1394 (CH2Cl bend), 1259, 1246 (C–H bend, CH3Si), 1097 (C–C ring stretch+Si–C stretch, Si–phenyl), 781, 754, Polymer syntheses and characterisation. The PMPS samples 730, 698, 664 (various C–C ring stretch+Si–C stretch), 465 used for halomethylation were prepared from the Wurtz-type (Si–Si stretch).lmax/nm (e) 239 (6300), 279 (4100), 341 (8600). reaction of distilled dichloromethylphenylsilane (Lancaster) using freshly prepared sodium sand in refluxing diethyl ether Results in the presence of 15-crown-5.20 Fractionation of PMPS was accomplished by washing the broad distribution polymer that Polymer structure and thermal characterisation is initially prepared with n-hexane.Halomethylations of unfractionated PMPS samples were carried out in accordance with We have previously reported that the halomethylation of PMPS by halomethyl ether prepared in situ leads to a substan- published procedures.21 Based on the composition range identified as being optimal for the lithographic performance of tial degradation of the polysilane backbone as evidenced by size exclusion chromatography.21,22 This probably occurs via corresponding copolymers of the CMS series of resists such as poly(styrene-stat-chloromethylstyrene), the extents of halome- the mechanisms outlined in Scheme 1, as both the SnCl4 and the halomethyl ethers have been observed to degrade polymers, thylation were targeted within the range 15 to 40%.The homopolymer PMPS 1 was prepared by the direct reaction of although it must be emphasised that no conclusive evidence for the formation of terminal SiCH2X groups (X=Br or Cl) dichloromethylphenylsilane with molten sodium in the absence of a solvent. Fractionation of this sample by repeated Soxhlet has yet been obtained.However, when bromomethyl octyl ether is used in the bromomethylation reaction in place of extractions of the sample with hexane gave the fractionated homopolymer PMPS 2. bromomethyl ether, after several purifications of the product by precipitation, washing and drying under vacuum, a number Representative spectroscopic data are given below for both bromomethylated PMPS and chloromethylated PMPS of alkyl peaks in the region d 2.0 to 0.5 can be observed in the 1H NMR spectrum.These are presumed to arise from end- samples, BPMPS and CPMPS respectively. BPMPS 1 and CPMPS 1 were prepared from PMPS samples that were capping of the polysilane chains by octyloxy groups, so provid- Table 1 Characterisation and lithographic parameters of the homo- and co-polysilanes studied DSC polymer Mw [PD] CH2X (%) T peak/°C Qpeak/J g-1 D0.5/mC cm-2 c Gx Gs Gs/Gx PMPS1 51500 7.5 0 a a >300 — — — — PMPS2 3300 1.4 0 a a %300 — — — — BPMPS1 15400 2.1 16 — — >300 — 0.32±0.03 0.66±0.09 2.4 BPMPS2 14500 2.3 24 284 -240 135 1.2 0.45±0.03 0.64±0.06 1.4 BPMPS3 12600 2.2 33 295 -290 110 0.9 0.71±0.02 0.93±0.03 1.3 BPMPS4 9900 1.8 38 299 -350 86 0.8 1.36±0.04 0.98±0.05 0.72 CPMPS1 8000 2.4 23 269 - 86 137 1.5 0.85±0.02 0.55±0.04 0.65 CPMPS2 14300 2.1 32 303 -260 60 1.7 0.62±0.02 0.38±0.04 0.61 CPMPS3 13900 2.1 33 299 -305 55 1.9 0.64±0.02 0.30±0.03 0.47 CPMPS4 14000 1.9 39 310 -367 46 1.8 0.83±0.03 0.45±0.04 0.54 a No distinct peak attributable to melting or decomposition observed 1702 J.Mater. Chem., 1997, 7(9), 1701–1707cessing.To this end, they should have glass transition temperatures in excess of 80 °C. Weak glass transition temperatures were observed at 113 °C for PMPS 1 and at 96 °C for PMPS 2. The diculty of observing a glass transition temperature for PMPS has previously been noted, however our observations are in accordance with previous observations.24,25 Even weaker thermal events were observed in the region 90–100 °C for some of the halomethylated polymers.It is therefore on this basis that the copolymers are considered to have thermal properties that are acceptable for application as resists. However, it must be stated that for the solutions in the casting solvent of the BPMPS series of polymers in particular, significant alterations to their molecular weight distributions were observed to occur over a period of days.This would be a most unattractive feature in the lithographic context. Lithographic assessments Lithographic contrast curves for PMPS are shown in Fig. 1, and a representative pair of contrast curves for the copolymer systems are shown in Fig. 2. The lithographic parameters are listed in Table 1. Unfractionated PMPS 1 displays positive-working behaviour with a very low sensitivity when exposed to the electron beam.Over the same dose range, the fractionated PMPS 2, which is of comparable molecular weight to the BPMPS and CPMPS samples, shows considerably less sensitivity to radiation- induced processes. In a recent study of the radiation chemistry of poly(cyclohexylmethylsilane) it was shown that two values for the radiation chemical yield for chain scission, Gs, could be determined.26 For high and low molecular weight polymers the values were 17.4 and 1.8 respectively.This Scheme 1 Polymer degradation accompanying the halomethylation of PMPS ing indirect evidence for the degradation mechanism given in Scheme 1(a). Peaks are always observed in the 1H, 13C and 29Si NMR spectra corresponding to terminal methoxy groups and also, occasionally, for terminal SiCl groups, the latter most probably remaining from their incomplete conversion to methoxy groups during the isolation and purification of the precursor PMPS.Contrary to an earlier explanation that these scission reactions occurred at randomly placed siloxy linkages in the polymer chain, it is now believed that they occur at points of conformational disorder that separate the otherwise extended s-conjugated sequences23 in PMPS which are known to be on average approximately 40 repeat units long.19 The extent of halomethylation was calculated from the Fig. 1 Representative contrast curves for PMPS: (&) PMPS 1, 1H NMR spectra. For both the bromomethylated and chloro- unfractionated; ($) PMPS 2, fractionated methylated samples, a general increase in halomethyl content from XPMPS 1–4 (X=Br or Cl) is indicated by NMR spectroscopy.Two structural features that must be emphasised as being of relevance to the subsequent assessment of the copolymers for lithographic applications are as follows: (i) the NMR, FTIR and UV analyses confirm the copolymers to have the expected statistical structures; (ii) the main position of substitution by the halomethyl groups is the para position but previous studies21 indicate that a small percentage are in the meta position.Data from the thermal analysis of the copolymers are also given in Table 1. A broad exotherm was observed for all of the samples at temperatures in excess of 250 °C.The peak temperatures and the energies of these transitions increase with increasing extent of halomethylation and they are accordingly attributed to thermally induced crosslinking processes. No similar exotherms were observed for the homopolymer samples PMPS 1 and PMPS 2. Polymers for application as resists Fig. 2 Representative contrast curves for halomethylated PMPS: ($) BPMPS 3; (&) CPMPS 3 should be robust under the conditions of lithographic pro- J.Mater. Chem., 1997, 7(9), 1701–1707 1703observation accords with the observations of the present study of PMPS and the phenomenon is perhaps more general amongst polysilanes than has previously been recognised. The structures of the high and low molecular weight polysilanes, and in the present context the two PMPS samples, probably dier in one crucial respect.It is likely that the fractionated sample of low molecular weight consists of polymer molecules in which the silicon atoms tend to catenate in all-trans sequences, whilst the unfractionated polymer, which has a higher average molecular weight and a bimodal distribution, contains molecules in which the all-trans sequences are separated by some linkages that are gauche.27 Such linkages continuously translate themselves along the high molecular weight polymer chain when they are in solution and they are considered to be points of weakness in the chain.It is proposed that in the solid state in which these gauche linkages are immobilised, they are then the positions that are most vulnerable to scission following electron transfer to the polymer chain such as might occur during irradiation.In clear contrast to the positive-working PMPS samples, all the resists of the two copolymer series are negative-working. Although the lithographic sensitivities within each series increase with increasing halomethyl content, the contrasts are low, and are particularly poor for the BPMPS series.The sensitivity variations are to be expected given that the crosslinking sites in these systems are presumed to arise from the halomethyl groups (see Scheme 2) and that scission is expected to remain at a constant level irrespective of halomethyl content, therefore the greater the number of active sites the greater the extent of crosslinking for the same radiation doses.However, it is notable that the BPMPS systems respond more sluggishly than the CPMPS systems and that for both systems the curves tend to normalised thicknesses remaining of only 0.7 to 0.8. This is indicative of a radiation-induced chain scission competing with crosslinking. Furthermore, the contrast curves of Fig. 2 are for bromomethylated and chloromethylated PMPS systems of comparable loadings so it is clear that the replacement of the chloromethyl by bromomethyl groups results in a significantly decreased sensitivity.This is surprising, since the lower bond dissociation energy and the higher reactivity of the C–Br bond when compared with the C–Cl bond (average bond enthalpies at 25 °C of 285 and 339 kJ mol-1, respectively) would be expected to lead to higher lithographic sensitivities for the BPMPS systems.This observation would seem to indicate that the radiation chemistries of the two systems are not directly comparable. A measure of the relative extents of crosslinking and chain scission can be obtained by plotting the lithographic data for the halomethylated, negative-working systems in accordance with the Charlesby–Pinner equation [eqn.(2)] for radiationinduced crosslinking of polymers with a most-probable (normal) molecular weight distribution.28 From the polydispersities shown in Table 1, the application of the equation is Scheme 2 Simple crosslinking mechanism for a halomethylated PMPS appropriate to the present systems (X=Br or Cl) s+Ós=Gs/2Gx+9.65×105/MwGxr (2) where s (=1-g) is the sol fraction in the exposed region, g values in Table 1 even approach this figure but neither are being the gel fraction and taken to be equal to the normalised they suciently low as to characterise a satisfactory negativethickness remaining after development, r is the absorbed working performance.Taken with the very low sensitivities radiation dose in Mrad and Gs and Gx are the radiation and poor contrast values, this signifies that on a number of chemical yields for chain scission and crosslinking, respectively.accounts these systems fail as potentially useful electron beam In Fig. 3 the data points of the representative contrast curves resists. It is nonetheless worthwhile to rationalise the variations of Fig. 2 are plotted in accordance with eqn.(2). A dose of Gs and Gx values with possible radiation chemical mechanconversion factor of 2 Mrad cm2 mC-1 has been applied.29 The isms in order to establish the reasons for the failure of the values of Gs and Gx, estimated from the slopes and intercepts halomethylated systems in the lithographic context, and to of these and similar plots for the remaining halomethylated serve that end the variations of the G values with composition polymers, are listed in Table 1 together with values of the ratio are depicted in Figs. 4 and 5. A value of Gx=0 is represented Gs/Gx. Values of Gs/Gx that are at least 4 are usually taken to for PMPS as it is appears from Fig. 1 that it does not undergo radiation-induced crosslinking. characterise a potential positive-working resist.30 None of the 1704 J.Mater. Chem., 1997, 7(9), 1701–1707further shown to correlate with a concerted crosslinking reaction of either chain-centred or substituent-centred benzylic radicals which originated from a radiation-induced excited state charge transfer interaction, and evidence for the requisite intermediates was provided from pulse radiolysis studies. This mechanism was assumed to operate over and above the crosslinking mechanism that results from the combination of benzylic radicals formed directly through the radiation-induced scission of carbon–chlorine bonds.Whether or not the details of such processes are accepted, in all the systems studied the optimal values of Gx were found to correlate with copolymer compositions in which there are about 33% halogenated (electron accepting) substituents, the remaining substituents being non-halogenated and therefore electron-donating.The Gx values of the CPMPS systems display similar variations with composition to those described above but they never attain values that compare favourably with those of the CMS systems. Although only a narrow band of composition has been investigated, it is considered that the composition at Fig. 3 Charlesby–Pinner plots of the data taken from the contrast maximum Gx, in the region of a 30 mol% chloromethyl curves for BPMPS 3 ($) and CPMPS 3 (&) shown in Fig. 2 loading, is suciently close to that found for the CMS systems as to allow the comparison. Within the composition range investigated, the variation of the Gs values with composition appears to follow the Gx values, a feature which might be taken as an indication that scission and crosslinking arise from similar intermediates.This is also the case for those CMS systems in which the two processes, when they occur, can be identified as stemming from chain centred benzylic radicals. It is not possible to identify analogous structures in the CPMPS systems and in the absence of data that are representative of a wider range of compositions it would be unwise to take either these comparisons or the analysis any further, other than to observe that chain scission in the chloromethylated systems is significantly more ecient than in the CMS systems.There are three reasons why the radiation chemistry of the bromomethylated polymers would dier from that of the chloromethylated polymers: (i) the lesser electronegativity of Fig. 4 Variation of the radiation chemical yields for ($) scission (Gs) bromine would not be as favourable as chlorine for excited and (&) crosslinking (Gx) with composition for CPMPS state charge transfer interactions, (ii) with bromine being a heavier atom than chlorine and the C–Br bond being weaker than the C–Cl bond, there is an increased likelihood of direct scission of the carbon–halogen bond, and (iii ) bromine atoms are significantly less reactive than chlorine atoms towards hydrogen atom abstraction.† The variation of G values with composition for the BPMPS system are indeed quite dierent from those of the CPMPS system, though the changes that occur within the narrow composition range investigated are again notable. Whereas both parameters apparently increase linearly up to about 30% bromomethyl content, thereafter Gx undergoes an extremely sharp increase.A log10Gx versus log10[CH2Br] plot is shown in Fig. 6 from which it is evident that, whereas Gx varies linearly with bromomethyl content at low values of the latter, at greater values the variation is of a higher order and appears, at the very least, to be in accordance with the square of the bromomethyl content.Any mechanism Fig. 5 Variation of the radiation chemical yields for ($) scission (Gs) of radiation-induced crosslinking that is consistent with these and (&) crosslinking (Gx) with composition for BPMPS observations requires a concerted reaction of increasing probability as the bromomethyl loading is increased.It would be in addition to the crosslinking reaction that follows from the Discussion direct formation of benzylic radicals through the radiationinduced scission of carbon–bromine bonds, and it would have It has previously been shown for the CMS series of resists prepared by statistical copolymerisation of chlorostyrene or to become the dominant crosslinking reaction at high bromomethyl loadings. A possible mechanism is shown in Scheme 3 chloromethylstyrene with either styrene or methylstyrene, that it is not uncommon for the variation of Gx with composition in which PBr represents a bromomethyl group in the polymer, D represents some kind of proximate association of bromo- to display both maximum and minimum values.31 The maximum values of Gx in the lithographically useful systems methyl groups such as a dimer, P is any repeat unit of the (copolymers of vinyl benzyl chloride and a methylstyrene) are least 2 and they occur in copolymers with chlorine-containing † For the reaction XV+CH3–H�HX+CH3V, the values of DH° are monomer loadings of about 30–40 mol%, at which Gs is estimated to be -4 and -69 kJ mol-1 for bromine and chlorine respectively (ref. 28). eectively zero. Such variations of Gx with composition were J. Mater. Chem., 1997, 7(9), 1701–1707 1705intercept of Fig. 5 the Gs value for PMPS can be estimated to be about 0.3. Some of the reactions depicted in Scheme 3 are as likely to apply to the CPMPS series as to the BPMPS series of copolymers.The most notable of these is the halogen atom induced chain scission reaction. The apparent greater eciency with which chain scissions occur in the BPMPS series can be attributed to the bromine atoms being ineectual at abstracting hydrogen atoms from the methyl substituents of the polymer chain. Without this alternative reaction pathway the bromine atoms are bound to induce chain scissions, i.e.k1>k2. As indicated above, hydrogen abstraction by chlorine atoms is significantly more exothermic so if chlorine atoms are formed within the CPMPS resists, they have the choice of two reaction pathways. It is reasonable to assume that in this case k1<k2. Fig. 6 A log–log plot for the Gx data of Fig. 5 Conclusion It is evident that the halomethylated PMPS series of resists are more susceptible to chain scission than their polystyrene counterparts and this can be attributed to halogen atom attack on the polymer backbone in a reaction that is particular to the polysilanes.Chain scission in the CMS series of resists arises from a b-scission reaction following the abstraction of a main chain hydrogen atom from the substituted carbon atom.The polysilanes do not have main chain hydrogen atoms, but the Si–Si bond is weaker than the C–H bond‡ and is readily susceptible to radical attack. Furthermore, the halomethylated PMPS resists do not undergo crosslinking with comparable eciency to the chloromethylated polystyrenes or poly(methylstyrene) s. For the CPMPS series, in which the eects of chain scission appear to be confined, Gx values of 0.8 would at best make them comparable to the chlorostyrene–methylstyrene copolymer resists.Though greater values of Gx may well be Scheme 3 A representation of a possible mechanism for the radiation chemistry of BPMPS accessible for the BPMPS series of resists at higher bromomethyl contents, not only would the enhanced chain scission reactions preclude their application in e-beam lithography, polymer chain, RV is a substituent-centred benzylic radical or but so also would their shelf life.Notwithstanding the high a radical formed by abstraction of a hydrogen atom from a silicon content which lends them a high resistance under the substituent methyl group of the polymer chain, and RE is a conditions of oxygen reactive-ion etching, simple derivatives chain end-centred radical resulting from chain scissions.The of PMPS are most unlikely to ever make useful negativenatures of the other species represented are self evident, and working electron beam resists. G1, G2 and G3 are radiation chemical yields for the primary processes represented. We thank the EPSRC for the award of a Research Studentship Assuming stationary state conditions for all radical species, (J.J.M.) and of a Postdoctoral Research Fellowship (S.J.H.).it follows that: We also gratefully acknowledge the assistance of Professor Ron Lawes and the sta of the Central Microstructure Facility [BrV]= G1PBr k1+k2 (3) of the Rutherford Appleton Laboratory, and in particular Mr Ejaz Huq, for facilitating the microlithographic evaluations.so References Gx=k3[RV]+G2D=G1PBr+k2 [BrV] P+G2D (4) 1 K. S. Mazdyasni, R. West and L. D. David, J. Am. Chem. Soc., =G1{1+k1/(k1+k2)}PBr+G2KPBr2 (5) 1978, 61, 504. and 2 R. E. Trujillo, J. Organomet. Chem., 1980, 198, C27. 3 S. Yajima, J. Hayashi and M. Omori, Chem. L ett., 1975, 931. 4 R. D. Miller and J. Michl, Chem. Rev., 1989, 89, 1359. GsP=G3P+k1 [BrV] P=G3P+ k1G1PBrP k1+k2 (6) 5 Inorganic Polymers, ed.J. E. Mark, H. R. Allcock and R. West, Prentice-Hall, New Jersey, 1992, ch. 5, p. 186. so 6 R. D. Miller and G. M. Wallraf, Adv. Mater. Opt. Electron., 1994, 4, 95. 7 D. R. Brambley, B. Martin and P. D. Prewett, Adv. Mater. Opt. Gs=G3+ k1G1PBr k1+k2 (7) Electron., 1994, 4, 55. 8 S. J.Webb, Ph.D. T hesis, University of Kent, 1994.The requisite variations of Gx and Gs are represented in eqns. 9 T. Tada and T. Ushirogouchi, Solid State T echnol., 1989, 91. 5 and 7. At low bromomethyl loadings when the first term of 10 S. Imamura, T. Tamamura, K. Harada and S. Sugawara, J. Appl. Polym. Sci., 1982, 27, 937. eqn. 5 dominates, Gx is directly proportional to [PBr], but as 11 S. Imamura, T. Tamamura, K. Sukegawa, O.Kogura and the loading increases and the probability of bromomethyl S. Sugawara, J. Electrochem. Soc., 1984, 131, 1122. groups being in close proximity increases, in accordance with observation the second term assumes a greater importance until it eventually dominates. Also in accordance with obser- ‡ The Si–Si bond dissociation energy in Me3Si–SiMe3 is 337 kJ mol-1 (ref. 32) whilst that of the C–H bond in Et–H is 410 kJ mol-1 (ref. 33). vation, eqn. 7 shows Gs varying linearly with PBr and from the 1706 J. Mater. Chem., 1997, 7(9), 1701–170712 R. G. Tarascon, M. A. Hartney and M. J. Bowden, inMaterials for 23 K. A. Klingensmith, J. W. Downing, R. D. Miller and J. Michl, J. Am. Chem. Soc., 1986, 108, 1046. Microlithography, ed. L. F. Thompson, C. G. Wilson and J.M. J. Frechet, ACS Symp. Ser. No. 266, American Chemical 24 S. Demoustier-Champagne, S. Cordier and J. Devaux, Polymer, 1995, 36, 1003. Society, Washington DC, 1984, p. 361. 13 A. Ledwith, M. Mills, P. Hendy, A. Brown, S. Clements and 25 J. M. Ziegler,Mol. Cryst. L iq. Cryst., 1990, 190, 265. 26 J. Kumagain, K. Oyama, H. Yoshida and T. Ishikawa, Radiat. R. Moody, J. Vac.Sci. T echnol. B., 1985, 3, 339. 14 Ll. G. Griths, R. G. Jones, D. R. Brambley and P. C. Miller Tate, Phys. Chem., 1996, 47, 631. 27 R. G. Jones, U. Budnik, S. J. Holder and W. K. C. Wong, Makromol. Chem., Macromol. Symp., 1989, 24, 201. 15 D. R. Brambley, R. G. Jones, Y. Matsubayashi and P. C. Miller Macromolecules, 1996, 25, 8036. 28 A. Charlesby and S. H. Pinner, Proc. R. Soc. L ond., Ser. A., 1959, Tate, J. Vac. Sci. T echl, B., 1990, 8, 1412. 249, 367. 16 R. G. Jones, Y. Matsubayashi, P. C. Miller Tate and D. R. 29 A. Novembre and T. N. Bowmer, in Materials for Brambley, J. Electrochem. Soc., 1990, 137, 2820. Microlithography, ed. L. F. Thompson, C. G. Wilson and 17 M. A. Hartney, J. Appl. Polym. Sci., 1989, 37, 695. J. M. J. Frechet, ACS Symp. Ser. No. 266, American Chemical 18 R. G. Jones, P. C. Miller Tate and D. R. Brambley, J.Mater. Chem., Society, Washington DC, 1984, p. 241. 1991, 1, 401. 30 Introduction to Microlithography, ed. L. F. Thompson, 19 C. Strazielle, A.-F. de Mahieu, D. Daoust and J. Devaux, Polymer, M. J. Bowden and C. G. Wilson, ACS Symp. Ser. No. 219, 1992, 33, 4174. American Chemical Society, Washington DC, 1983. 20 R. H. Cragg, R. G. Jones, A. C. Swain and S. J. Webb, J. Chem. 31 R. G. Jones, P. C. Miller Tate and D. R. Brambley, Polymer, 1993, Soc., Chem. Commun., 1990, 1147. 34, 1768. 21 A. C. Swain, S. J. Holder, R. G. Jones, A. J. Wiseman, M. J. Went 32 R. Walsh, Acc. Chem. Res., 1981, 14, 246. and R. E. Benfield, in Metal-containing Polymers, ed. 33 S. W. Benson, T he Foundations of Chemical Kinetics, McGraw- C. U. Pittman, Jr., C. E. Carraher, Jr., M. Zeldin, J. E. Sheats and Hill, New York, 1960. B. M. Culbertson, Plenum, New York, 1996, p. 161. 22 R. G. Jones, R. E. Benfield, A. C. Swain, S. J.Webb and M. J.Went, Polymer, 1995, 36, 393. Paper 7/00413C; Received 17th January, 1997 J. Mater. Chem., 1997, 7(9), 1701–1707 1707