J O U R N A L O F C H E M I S T R Y Materials Influence of composition and glass transition temperature on the diVusion and solubility behaviour of methyl ethyl ketone–isopropyl alcohol mixtures in poly(methyl methacrylate) Richard A. Pethrick* and Kathleen E. Rankin Department of Pure and Applied Chemistry, University of Strathclyde, 295 Cathedral Street, Glasgow, UK G1 1XL Received 31st March 1998, Accepted 14th October 1998 The process of removal by a solvent mixture of low molar mass polymer generated as a consequence of chain scission is a critical step in electron beam lithography.The development of the image depends on a number of factors, including the composition of the solvent and the nature of the polymer. In this paper the eVects of change in the composition of mixtures of methyl ethyl ketone and isopropyl alcohol, a common development solvent used in lithography, and the glass transition temperature of poly(methyl methacrylate) films on the mutual diVusion coeYcient are reported.The mutual diVusion coeYcient decreases and becomes asymmetric towards low volume fraction of solvent as the proportion of isopropyl alcohol in the mixture is increased. The higher glass transition temperature films are prone to exhibiting crazing on exposure to solvent.The diVusion of the mixture into the polymer film is selective and preferential for methyl ethyl ketone. DiVusion becomes complex as the content of the mixture moves towards a higher isopropyl alcohol composition. Also, there is evidence for both lowering of the glass transition temperature and re-precipitation of the polymer by the non-solvent (isopropyl alcohol ).Change in the initial Tp of the films leads to small changes in the swelling rate. In the development process of electron beam resist films used in semiconductor lithography, crazing probably plays as important a role in the overall development process as simple solvent driven dissolution.polymer undergoes progressive scission producing a lower Introduction molar mass product. Changes in the solvent mixture and The ability to achieve very large-scale integrated (VLSI) circuit temperature result in the mixture changing from being a good fabrication depends critically on the precision of the litho- to a poor solvent for the polymer. Since the solubility is a graphic processes used. While photolithography remains the function of molar mass, it is possible to select a mixture which primary tool for large-scale semiconductor fabrication, mask will selectively dissolve the low molar mass material without generation and specialist circuit fabrication is dominated by significantly swelling the unexposed, higher molar mass compoelectron beam technology.1 The lithography process is based nents.Despite the importance of electron beam lithography in on radiation induced changes in the dissolution rate of thin VLSI fabrication, little research appears to have been carried resist films. The type of development process is controlled by out on the mechanism of polymer dissolution in these mixed the nature of the interaction of the radiation with the resist, solvent systems.PMMA can exist in three diVerent stereochemwhich may be to either degrade or crosslink the polymer. In ical forms with diVerent values of the Tg. Isotactic PMMA general, there are several diVerent regimes for polymer dissolu- has a Tg of approximately 40 °C, the atactic form has a Tg tion, each of which requires a separate model to describe the of 117 °C and syndiotactic PMMA has a Tg of 125 °C.The process. The mechanism for poly(styrene) dissolution in a diVerences in the Tg values will also influence the solubility good solvent is clearly very diVerent from that for poly(methyl rates of polymers with the same molar mass.9 DiVerences in methacrylate) and both are diVerent from the dissolution of stereochemistry also influence the sensitivity of the polymer the inhibited phenolic polymers used in photoresists.1 Of these to electron beam irradiation.9–11 A combination of a more processes, only the first, dissolution of glassy amorphous marked dependence of the solubility on molar mass for the polymers like poly(styrene), is reasonably well understood.2,3 isotactic polymer and changes in the distribution of the molar Peppas and co-workers have applied scaling concepts4,5 to mass for the degraded polymer have a significant eVect on the the description of the dissolution of poly(styrene).Their theory electron beam sensitivity. Ignoring slight diVerences in the assumes the initial formation of a gel layer at the polymer– degradation mechanism, the solubility rate of the exposed solvent interface as the solvent diVuses into the film.Once this PMMA correlates well with the segmental mobility of the gel layer is formed, it propagates at constant thickness through polymer. Changes in the size of the solvent used in the the polymer film as dissolution occurs. The dissolution rate development process indicate that the solubility rates are and thickness of the gel layer are dependent on the molecular strongly aVected by the relative size of the interstitial distance weight of the polymer and are described accurately by reptation (free volume) between the highly interpenetrating chain of the theory.5 Experiments on high molar mass poly(styrene) indi- polymer and size of the solvent molecule.Plots of the solubility cate that methyl ethyl ketone (MEK) dissolves at a rate against molar mass of the solvent exhibit a sharp break in predicted by theory, but these predictions failed when applied slope between propyl acetate and butyl acetate.The power to poly(methyl methacrylate)6–8 (PMMA), where dissolution dependence of the solubility on the molar mass between methyl occurred without the formation of a significant gel layer.acetate and propyl acetate is relatively weak, but between Many electron beam resists are based on PMMA and butyl acetate and the higher acetate homologues is very strong. development of the lithographic pattern is achieved by the use For lower molar mass solvents there is relatively little correof mixtures of isopropyl alcohol (IPA) and MEK.Isopropyl lation between the motion of the polymer chain and that of alcohol is a non-solvent for PMMA, whereas MEK is a good the solvent whereas, above propyl acetate, the motion is hindered by the polymer and a high degree of correlation solvent. During the exposure process, the high molar mass J. Mater. Chem., 1998, 8, 2591–2598 2591between the chain and solvent is required for diVusion.A Fabry Perot interferometer experiment for assessment of solvent diVusion typical development mixture, used in practice, consists of 153 MEK to IPA, at room temperature. Selection of the composi- The basic interferometer used for assessment of the solvent tion of the solvent mixture is usually based in measurement diVusion process was constructed using two 6 mm thick glass of dissolution rates.It has been shown that the dissolution plates which measured 25×25 mm. These were coated with rate (S) is related to the molar mass through eqn. (1), chromium to achieve a transmission level of approximately 20%. The coating was carried out using an Edwards Coating S=KMa (1) System E306A and the transmission measured with a Perkin- Elmer 257 IR spectrometer.The polymer films were either where K and a are solvent dependent parameters that are spun coated onto the glass or, alternatively for thicker films, specific to the particular polymer system under consideration.14 cast by slow evaporation from an 8 wt% solution in MEK. A number of instruments have been developed for the measure- The film sandwiched between the plates was clamped using a ment of dissolution rates.The technique usually involves the normal IR plate holder. This construction was placed in a measurement of changes in thickness of the resist layer, loss water jacket connected to a thermostatted bath and held at of weight or time to complete dissolution measured by end 30±0.1 °C. A more detailed description of the experimental point analysis.15 However, these studies do not give a molecular apparatus used has been published elsewhere.16 The fringe interpretation of the dissolution process.A novel technique pattern was recorded using an Olympus CHC binocular micro- has been proposed16 which allows examination of the dissoluscope with an attached Olympus OM2 camera. The interfer- tion process by optical examination of the change that occurs ometer was illuminated with a sodium vapour lamp which has in the optical interference, as a function of time, for micron a strong band at 589 nm and the images recorded using a thick films sandwiched into a Fabry Perot interferometer Kodak black and white Tri X Pan film which had a speed of configuration.16 In this paper, the results of a study of the 400 ASA and was sensitive to yellow light.eVects of solvent variation on the dissolution process for atactic PMMA are reported. The polymer selected is typical of the type of material commonly used in many electron Operation of the Fabry Perot interferometer and data analysis beam resists. Mixtures of MEK–IPA, in the range of 352 w/w MEK–IPA, Solvent diVuses into the polymer, and the sharply defined edge are usually used for the development of electron beam resists.of the film and the associated interference pattern becomes Compositions of 151, 352 and 3159 w/w MEK–IPA were distorted, taking up a sigmoidal form which reflects the way investigated in an earlier paper.16 It was observed that, as the in which the fringe pattern changes as the solvent diVuses into amount of IPA is increased, the mutual diVusion coeYcient the polymer.Change in the number of fringes per unit dimenwas reduced. Also, it was noted this as the point at which the sion is a direct measure of the concentration–distance profile. volume fraction of solvent coincides with the maximum mutual This approach was initially proposed by Crank18 and by diVusion coeYcient moves towards lower volume fractions of Crank and Park.19 Each interference fringe represents a plot solvent in the films.In certain instances, the films exhibited of refractive index versus distance, over the concentration environmental stress cracking, consistent with the good solvent range from pure solvent to pure polymer. Assuming that the (MEK) shocking the films’ surface.This eVect could be refractive index is linearly proportional to concentration and overcome simply by not baking the film, thus allowing the that there is negligible volume change on mixing of polymer and solvent, then the refractive index plot represents the residual casting solvent to remain and to plasticise the film. A concentration profile. The profiles were recorded at 3 min two stage diVusion process was also observed for certain intervals and the fringes traced from the photograph.Where compositions. In this paper, an attempt will be made to explore the fringe is horizontal at the film edge, the concentration of further the nature of the two stage diVusion process and to diVusing solvent is zero (i.e. ws=0) and where it is horizontal quantify the eVect of baking on the properties of the films.on the solvent side, the polymer concentration is zero (i.e. ws=1.0). The two measurable features are the change in the concentration profile and movement of the polymer film edge, Experimental which is a direct measure of the rate of swelling of the polymer film. When there is a discontinuity in the fringe pattern, which Materials and thin film formation is a direct indication that the solvent is a poor solvent for the Poly(methyl methacrylate) was obtained from Merck (Poole, polymer (Fig. 1), then the maximum number of fringes, nT, UK), and had a nominal molar mass of 100 000 g mol-1. The that would be observed between pure solvent and pure polymer molar mass and its distribution were determined by gel per- is calculated using eqn.(2),18,19 meation chromatography and a value of M9 n of 96 000 g mol-1 and a heterogeneity index of 1.66 was obtained. Methyl ethyl ketone and isopropyl alcohol were used as solvents and were nT= 2l l (np-ns) (2) obtained from Merck (Poole, UK) as AnalaR grade reagents. The refractive indices of the solvent mixtures were measured where l is the film thickness, l is the wavelength of the using an Abbe refractometer. Films used in this study were monochromatic light source and np and ns are respectively the spun onto chromium-coated glass substrates from a 3 wt% refractive indices for polymer and solvent.The mutual solution of polymer in MEK, using a Headway Research diVusion coeYcient can be calculated using eqn. (3),18,19 Incorporated spinner operating with speeds between 1000 and 500 revolutions per minute (rpm). 50 ml of polymer solution were applied directly to the substrate (2×2 cm). All solutions Dm= -A1 2tB Adx dwsBws=ws¾ Pws¾ 0 xdws (3) used were filtered through a 0.22 mm filter prior to use. The best films were produced when 15 s was left between application of the droplet and spinning at a pre-set speed for 60 s.7 where ws is the volume fraction of solvent at some time t, at The films were baked at 120 °C for 1 h to remove excess a plane distance x away from the original boundary between solvent and to aid development of a smooth profile.These solvent and polymer. The original position of this boundary films are of similar dimensions to those used in electron beam is always set in the same place.The Boltzmann transformation for any fixed value, ws, versus x/Ót, for the data collected at lithography. 2592 J. Mater. Chem., 1998, 8, 2591–2598various times should all fall on a single average curve if the process is Fickian.3,18,19 In this case, eqn. (3) gives eqn. (4). Dm=- A1 2B Adxt 1 2 dws Bws=ws¾ Pws¾ 0 xt 1 2dws (4) Gradient of Area under tangent to the curve the curve at ws¾ from 0 to ws¾ This equation was used to calculate the concentration–distance profile for the system investigated. Results and discussion Secondary boundary phenomenon Investigation of the diVusion behaviour with a 352 w/w IPA–MEK mixture showed that the curves up to about 16 min conformed to a simple one stage process.However, after 25 min a second boundary appeared at the edge close to the solvent and rapidly diVused into the film which already contained solvent.This was very marked at about 36 min, where it was observed that the second boundary had now reached a point about half way between the film surface and edge of the solvent diVusion front into the polymer. It is not possible to calculate a diVusion coeYcient for the separate processes because it is not possible to determine the composition of the solvent mixture at the line between the two diVusion regions.In order to obtain an apparent mutual diVusion coeYcient, an average value over the whole curve was calculated in the manner presented in the previous paper.16 Two hypotheses can be proposed to explain the observation of two diVusion regions.Gel–glass boundary. It is possible that the feature is similar to that observed in poly(styrene),4 where a two stage diVusion process is associated with the solvent lowering the glass transition temperature. In the initial stages, the mixture diVuses into a glassy polymer and is controlled by the osmotic pressure –solubility of the solvent in the matrix. After suYcient solvent has diVused into the polymer matrix, the polymer is able to swell and changes from a glass into a gel state.In this initial stage, the solvent is diVusing into a mobile polymer state and diVusion is influenced by segmental motion and reptation of the polymer chains. Crank and Robinson20 have termed this process middle boundary behaviour. Preferential solvent absorption. When the solvent contains both good solvents and non-solvents, preferential absorption of the good solvent lowers the Tg.The expanded matrix allows access of the non-solvent which will reprecipitate the polymer and lead to the mixed solvent diVusion having a diVerent rate to that of the initially absorbed good solvent. The second boundary is then associated with precipitation of polymer rather than the occurrence of the glass to gel transition.In assessing the possible eVects of the solvent on depression of the Tg, measurements using a penetration technique were made, as described previously.21 The apparatus consisted of a 6 mm diameter glass rod with a round tip at one end, which was allowed to rest on the sample which had been previously plasticised with a known composition of solvent.A chromel– alumel thermocouple was attached to the tip of the rod and Fig. 1 Interference photographs for PMMA/MEK: (a) before solvent the temperature measured using a Digitron 3750-K digital added, and after (b) 3, (c) 5 and (d) 10 min; (e) concentration– thermometer. A Shlumberger linear variable diVerential transdistance curve for PMMA/MEK. ducer (LVDT) was placed on the top of the probe and its movement recorded against temperature, using a two pen recorder.The sample was cooled with a methanol–cardice mixture or heated with a paraYn oil bath. The temperature J. Mater. Chem., 1998, 8, 2591–2598 2593was changed at a rate of 3 K min-1. The samples, in sealed in samples 90 to 30%, with the excess solvent becoming cloudy in the 90 to 70% samples.phials, had been previously equilibrated for two days, in a temperature controlled oven at 40 °C, to allow the mixture The penetration technique was used to measure the solids which contained various levels of absorbed solvent. The results to penetrate into the solid completely, and were cooled to room temperature, before the phials were opened, to avoid are presented in Fig. 2. It is evident that the Tp, which is closely associated with the Tg, has essentially the value expected evaporation. The variations of the Tg, determined by the penetration for depressions produced by MEK for low values of polymer to solvent, and it is only at higher values of solvent absorp- method, are shown in Fig. 2. When MEK is used, the observed Tg deviates significantly from ideal mixing behaviour.Similar tion that deviations from the ‘ideal MEK’ curve are observed. The values of Tp are very close to those of the Tg measured experiments were performed, starting with solvent having compositions of 151, 352 and 753 w/w IPA–MEK. The previously on bulk samples.21 The liquid layer was analysed using 1H NMR signal intensities to determine its compositions polymer was equilibrated with the solvent for two days at 40 °C and then the samples were measured.The following (Table 1). Analysis of the composition of the excess liquid layer shows that, for solutions with low solids content, the solution observations were made. composition is identical to that of the solution. However, as the volume of solvent in the polymer mixture is reduced, the 151 w/w IPA–MEK/PMMAmixtures.At room temperature, for 60 to 90% solvent to polymer mixtures, not all of the good solvent is preferentially absorbed leading to changes in the measured composition compared with that originally used added solvent was absorbed. Below 60%, all the solvent was absorbed and the solid was clear. Below 0 °C, the excess solvent to produce the mixture.This implies that, at any point during the diVusion process, the composition can deviate from that is cloudy in the 90 to 80% samples, but is clear in the 70 and 60% samples. For 90 to 40% samples, there is a double layer of the contacting mixture as a consequence of preferential absorption of MEK. The average mutual diVusion coeYcient visible in the polymer layer. The lower part of the solid is clear.However, the upper layer is opaque. On cooling, the 30 must therefore be considered as reflecting the overall behaviour of the system and incorporates the eVects of the precipitation to 20% samples remain unchanged. as well as solubility–diVusion. 352 w/w IPA–MEK/PMMA mixtures. Two phase behaviour is visible at room temperature for the 90 to 40% solvent to EVect of glass transition temperature on the diVusion behaviour polymer mixtures.The excess solvent is clear at room temperature, but cloudy below 0 °C. The spun polymer resist is usually baked at a temperature above the Tp before being exposed to electron beam irradiation. The process is carried out to increase the mechanical rigidity, 753 w/w IPA–MEK/PMMA mixtures.At 40 °C, the polymer mixtures are clear in most cases. The exception is that in the flatness of the film and to improve the development characteristics, after electron beam exposure. For PMMA, baking is 90 to 60% solvent to polymer mixtures, there is a surface layer of opaque polymer. Excess solvent is evident in the 90 to 40% usually performed at between 130 and 160 °C for 1 h.Not all of the residual casting solvent is removed during the samples. At room temperature, two phase behaviour is evident Table 1 Variation in the composition with the volume of solvent added for various starting compositions of MEK and IPA Solvent Solvent IPA in Solvent Solvent IPA in added (%) absorbed (%) solvent residue (%) added (%) absorbed (%) solvent residue (%) 1:1 w/w IPA–MEK 90 56 55 80 53 55 70 55 57 60 55 60 50 50 61 40 40 no residual 30 30 no residual 20 20 no residual 3:2 w/w IPA–MEK 90 55 62 80 53 64 70 51 64 60 49 71 50 49 no residual 40 39 no residual 30 30 no residual 20 20 no residual 7:3 w/w IPA–MEK 90 54 71 80 45 76 70 46 77 60 46 80 50 42 78 40 40 77 30 30 no residual 20 20 no residual Table 2 Description of the PMMA films used in the study of the eVects of Tg on the mutual diVusion coeYcients Initial Final Film Tg/°C Thermal–time treatment Tg/°C A 66 7 days @ ambient T and P 73 B 59 12 days @ ambient T and P and 24 h in a 78 vacuum oven at ambient T C 61 vacuum oven; 40 °C/48 h, 60 °C/24 h, 98 80 °C/36 h D 60 ambient P, 130 °C/1 h 106 E 64 ambient P, 160 °C/1 h, 112 cooled straight from oven F 61 ambient P, 160 °C/1 h, 110 cooled very slowly over 24 h 2594 J.Mater. Chem., 1998, 8, 2591–2598Fig. 2 Variation of the penetration temperature (Tp) for various IPA–MEK mixtures with PMMA: (%) pure and (2) 151, (&) 352 and (1) 753 MEK, w/w IPA–MEK. The errors in the experimental points are ±3 K; the lines are guides to the data. spinning process and it is appropriate to examine the eVects of Tp and, hence, the residual solvent content on the development/ diVusion behaviour.Fig. 3 Boltzmann transformation curves. (a) Volume fraction of sol- A series of films was produced by casting. Their initial vent versus distance/Ótime for film A. (b) Mutual diVusion coeYcients values of Tp are presented in Table 2. These low Tp films which for diVerent solvent mixtures for film A.(%) 151, (2) 352 and (&) 753 w/w IPA–MEK. The errors are estimated to be ±0.05 in the contain residual casting solvent will slowly lose solvent and volume fraction in (a) and ±0.05×10-11 m2 s-1 for each data point increase their Tp at a rate of approximately 1 °C per day. in Dm in (b). Measurements were performed within 2 h of the Tp measurements. The values quoted in Table 2 are, therefore, the values of the films used in the diVusion study.Since the diVusion cell is a sandwich of heavy glass plates, further significant loss of 753 w/w IPA–MEK mixtures, the curves are crescent shaped and truncated at ws=0.7 and ws=0.5 respectively. The curves solvent is unlikely once the cell has been constructed. are almost indistinguishable up to ws=0.5 and, as a consequence, the mutual diVusion curves [Fig. 4(b)] are essentially Film A (Tp=73 °C). The concentration–distance profiles for the exposure of the films to three diVerent solvent compositions identical. are shown in Fig. 3(a). The plot for the 151 w/w IPA–MEK mixture is sigmoidal in shape and ws=0.5 when x/Ót=0, Film C (Tp=98 °C). The composition–distance curves obtained with the three solvent mixtures are shown in Fig. 5(a). indicating that the rate of solvent penetration into the film is equal to the rate of polymer dissolution into the solvent and In the case of the 151 w/w IPA–MEK mixtures, the film edge disappears after 25 min and recedes as dissolution takes place. that simple Fickian diVusion is observed. Consistent with this assumption is the disappearance of the film edge after about The 352 w/w IPA–MEK mixture relationship is again sigmoidal in behaviour and extends up to ws=1.0.The edge is 8 min of exposure to solvent. For the 352 w/w IPA–MEK mixture, there is a discontinuity at ws=0.8, indicating that the not well defined and abrupt kinks in the fringe pattern are observed rather than a distinct discontinuity. In the case of solvent mixture is no longer able to dissolve the polymer completely. The behaviour of the 753 w/w IPA–MEK solutions the 753 w/w IPA–MEK mixture, a secondary boundary was evident even after 120 min.The diVusion coeYcients for the is similar to that of the 352 w/w IPA–MEK mixture, except that it shows less penetration of solvent into the polymer and lower concentrations are shown in Fig. 5(b). less swelling. The calculated mutual diVusion coeYcient–concentration curves [Fig. 3(b)] show almost symmetrical behav- Film D (Tp=106 °C). The concentration–distance curves for two of the mixtures are shown in Fig. 6(a). As with film C, iour, the curves for the 352 and 753 w/w IPA–MEK mixtures being truncated, reflecting swelling without dissolution of the the shapes of the curves are similar to those observed for other films.A distinct edge receding with time was observed after polymer in the solvent. 16 min. This edge quickly disappears as dissolution of the polymer takes place. In the case of the 753 w/w IPA–MEK Film B (Tp=78 °C). From studies of 151 w/w IPA–MEK mixtures, the concentration–distance curves indicate, once mixture, it was not possible to calculate curves as the films exhibited both environmental stress cracking and also the more, approximately sigmoidal behaviour, with the disappearance of the film edge after 9 min [Fig. 4(a)]. For the 352 and presence of a secondary boundary. J. Mater. Chem., 1998, 8, 2591–2598 2595Fig. 4 Boltzmann transformation curves. (a) Volume fraction of sol- Fig. 5 Boltzmann transformation curves.(a) Volume fraction of solvent versus distance/Ótime for film B. (b) Mutual diVusion coeYcients vent versus distance/Ótime for film C. (b) Mutual diVusion coeYcients for diVerent solvent mixtures for film B. (%) 151, (2) 352 and (&) for diVerent solvent mixtures for film C. (%) 151, (2) 352 and (&) 753 w/w IPA–MEK. The errors are estimated to be ±0.08 in the 753 w/w IPA–MEK. The errors are estimated to be ±0.05 in the volume fraction in (a) and ±0.1×10-11 m2 s-1 for each data point volume fraction in (a) and ±0.15×10-11 m2 s-1 for each data point in Dm in (b).in Dm in (b). Film E (Tp=112 °C) and film F (Tp=110 °C). Film F would solid state. The non-equilibrium chain structure will attempt be expected to have a lower residual solvent content as it was to gain its equilibrium state as the film is swollen and, hence, kept at a higher temperature for a longer period of time.In will generate stresses that may lead to stress crazing. The both cases, environmental stress cracking made analysis of the removal of solvent will be accompanied by generation of diVusion behaviour very diYcult. In both cases, the film edge denser, more compact structures and a concomitant reduction receded with time as dissolution of the polymer molecules into in the diVusion coeYcient would be observed.The apparently the solvent occurred. The film edge disappeared after a period anomalous behaviour of the 106 °C film can be explained by of about 27 min for film E and after 100 min for film F. Stress the fact that, in this case, contact with poor solvent allows recracking is a consequence of the osmotic pressure increasing dissolution of the polymer molecules in the surface without quickly in the film edge and the polymer behind being unable significant re-swelling of the molecules that form the bulk of to release the stress which is generated.the material. Hence, the rate of dissolution of the polymer Comparison of the data from the various films exposed to becomes comparable to the rate of solvent penetration into the 151 w/w IPA–MEK indicates that all the curves are the polymer and crazing is not observed.virtually superimposable, with only slight diVerences, in the region ws=0.7–1.0 and in the range ws=0–0.3, being observed Swelling behaviour (Table 3).There is a marked decrease in Dm as the solvent mixture is changed from 151 to 753 w/w IPA–MEK, this eVect The development process is a combination of dissolution and swelling of the polymer matrix. The swelling rate can be being particularly marked for the low Tp film. There is slight reduction in the diVusion coeYcient with increase in the Tp. measured directly from the movement of the solvent polymer interface, obtained from the interferograms. Change in the Tp However, this eVect is not as marked as the solvent eVect.There are no data presented for films C and D for 753 w/w of the films leads to small changes in the swelling rate with the 352 w/w IPA–MEK (Fig. 7). There are two eVects which IPA–MEK as these exhibited marked crazing. As the composition of the solvent mixture is changed, the plots of Dm against need to be considered in interpretation of these data.Firstly, the polymer films are obtained from a good solvent. Also, it ws become asymmetric towards low values of ws. The polymer chains will be extended in the good solvent is probable that the polymer molecules may have retained their expanded conformations with the possible eVect of micro- used for spin casting the films and this extended structure, as a consequence of chain entanglement, will be retained in the crazing at the surface in the higher Tp films influencing the 2596 J.Mater. Chem., 1998, 8, 2591–2598Fig. 7 Swelling rate curves for (a) 352 and (b) 753 w/w IPA–MEK. Fig. 6 Boltzmann transformation curves. (a) Volume fraction of sol- (%) Film A, (2) film B, (&) film C and (1) film D.The errors in vent versus distance/Ótime for film D. (b)Mutual diVusion coeYcients the distances are of the order of 0.1×10-4. for diVerent solvent mixtures for film D. (%) 151 and (2) 352 w/w IPA–MEK. The errors are estimated to be ±0.05 in the volume fraction in (a) and ±0.1×10-11 m2 s-1 for each data point in Dm in (b).The errors at the extremes of the curves are greater than the and precipitation of polymer in the swollen layer by the average values. poorer solvent. Table 3 Mutual diVusion coeYcients for films with diVerent values of Acknowledgement Tg One of the authors (K.E.R.) wishes to acknowledge the Dm (max)/10-10 m2 s-1 support of the EPSRC in provision of support in the form of a studentship for the period of this study.IPA % Film A Film B Film C Film D 50 0.26 0.23 0.20 0.22 References 60 0.14 0.13 0.05 0.12 70 0.08 0.11 — — 1 P. C. Tsiartas, L. W. C. L. Henderson, W. D. Hinsberg, I. C. Sanchez, R. T. Bonnecaze and C. GrantWillson, Macromolecules, 1997, 30, 4656. 2 J. M. D. McElroy, DiVusion in Polymers, ed. P. Neogi, Marcel initial diVusion behaviour.Secondly, there are marked diVer- Dekker, New York, Basel, Hong Kong, 1996, p. 1. 3 J. Crank, The Mathematics of DiVusion, Oxford, 1985, p. 256. ences in the swelling rate with change in solvent. 4 N. A. Peppas, J. C.Wu and E. D. von Meerwell, Macromolecules, 1994, 27, 5626. Conclusions 5 P. G. de Gennes, Scaling Concepts in Polymer Physics, Cornell University Press, Ithaca, NY, 1979.Change in the composition of the solvent mixture used in the 6 A. C. Ouano and J. A. Carothers, Science and Technology of measurement of the mutual diVusion coeYcient for PMMA Polymer Processing, ed. N. P. Sung and N. H. Sung, MIT Press, Cambridge, MA, 1979, p. 755. films has significant eVects on the Tp of the film. The higher 7 A. C. Ouano and J. A. Carothers, Structure–Solubility the Tp of the films formed after baking, the more susceptible Relationships in Polymers, ed.F. W. Harris and R. B. Seymour, are the films to crazing. In the resist development process, Academic Press, New York, 1977, p. 11. crazing probably plays as important a role as dissolution in 8 A. C. Ouano, Y. O. Tu and J. A. Carothers, Polym. Prepr., 1976, the overall process. The diVusion of the solvent into the 17, 329. polymer film is a complex process which involves selective 9 A. C. Ouano, Polymers in Electronics, ed. T. Davidson, ACS Symp. Ser., 1984, 242, 84. diVusion of the better solvent, lowering the Tp of the films J. Mater. Chem., 1998, 8, 2591–2598 259710 R. A. Pethrick, Irradiation EVects on Polymers, ed. D. W. Clegg 16 K. E. Rankin and R. A. Pethrick, Microelectron. Eng., 1995, 26, and A. A. Collyer, Elsevier Applied Science, London, 1991, p. 403. 141. 11 V. K. Sharma, R. A. Pethrick and S. AVrossman, Polymer, 1982, 17 K. R. Dunham, Solid State Technol., 1971, 14(6), 41. 23, 1732. 18 J. Crank, The Mathematics of DiVusion, Oxford Science 12 T. Kato, Microelectronic Polymers, ed. M. S. Htoo, Marcel Publications, Oxford, 1985, p. 230. Dekker, New York, 1989, p. 1199. 19 J. Crank and G. S. Park, DiVusion in Polymers, ed. J. Crank and 13 M. M. O’Toole, Microelectronic Polymers, ed. M. S. Htoo, Marcel G. S. Park, Academic Press, New York, 1968, ch. 1, p. 4. Dekker, New York, 1989, p. 315. 20 J. Crank and C. Robinson, Proc. R. Soc. London, Ser. A, 1951, 14 M. J. Bowden, Materials for Microlithography, ACS Symp. Ser., 204, 549. 1984, 266, 71. 21 R. J. Elwell, D. Hayward and R. A. Pethrick, Polym. Int., 1993, 15 L. F. Thompson, Introduction to Microlithography, 2nd edn., ed. 30, 55. L. F. Thompson, C. Grant Willson and M. J. Bowden, ACS Professional Reference Books, ACS, Washington DC, 1994, p. 280. Paper 8/02466I 2598 J. Mater. Chem., 1998, 8, 2591–2598