- Email: [email protected]
Polymer resist materials for excimer ablation lithography
Applied Surface Science 127–129 Ž1998. 905–910
Polymer resist materials for excimer ablation lithography Kenkichi Suzuki ) , Masaaki Matsuda, Nobuaki Hayashi Electron Tube and DeÕices DiÕision, Hitachi, Hayano, Mobara, Chiba 297, Japan
Abstract Excimer ablation lithography ŽEAL. is a new lithography suitable to TFT-LCD. Among the constituent technologies, the polymer materials for the resist is most crucial, as high ablation rates are required at low fluence. To elucidate the ablation mechanism at low fluence we specified some characteristic features through observations of about 200 polymers. The low fluence features are explained by the contributions from the primary and secondary structures, and ablation dynamic process. From the results, it is derived that polyurethane is most promising material for EAL, and the design of the secondary structure is essential to improve the ablation rate. q 1998 Elsevier Science B.V. Keywords: Polymer; Resist; Excimer ablation lithography; Polyurethane
1. Introduction Excimer ablation lithography ŽEAL. is a new concept of lithography characterized by self-development and ablation removal of resist. Owing to reduction of process steps and disuse of conventional developer and remover equipments and accompanying chemicals, we can expect considerable cost reduction of the lithography processes. The feasibility for TFT-LCD has been proven with respect to the precision and size w1x. However, there remains a problem of rather longer throughput time, 4–5 times than that of the conventional technology. The throughput time itself is a kind of figure of merit, composed of various constituent technologies. Among them, the most crucial one is resist materials.
)
Corresponding author. Tel.: q81-475-25-9062; fax: q81475-24-2463; e-mail: [email protected].
The ablation rate of the resist directly determines throughput time, and the maximum rate at present is 0.05 m mrshot at 248 nm. The rate of more than two times this figure is required to catch up and cope with the conventional lithography. This low rate is due to low fluence, 100 mJrcm2 typically, which is indispensable to prevent damage to underneath TFT films. The fundamental problem is why we could not find materials with a rate higher than 0.05 m mrshot. In other words, it is very essential to clarify the mechanism of polymer ablation at low fluence again from the view point of lithography applications. The ablations mechanisms have been investigated very extensively, but they were mainly concerned with higher fluence region, as the applications have been for micromachining such as boring of PCB. Ablation phenomenon is generally understood to be a kind of explosion bombed by photons. This implies that the ablation is not so much dependent on
0169-4332r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII S 0 1 6 9 - 4 3 3 2 Ž 9 8 . 0 0 1 0 8 - 1
906
K. Suzuki et al.r Applied Surface Science 127–129 (1998) 905–910
the details of molecular structure. However, it is not clear what extent the material properties is reflected to oblation phenomenon. It is very important to estimate the contributions from molecular structures for practical applications. This paper presents the results of observations of low fluence oblations of more than 200 samples derived from eight primary polymers, and clarifies two characteristic features. These are explained by primary and secondary structure of polymers, and the general explosion process, although each contribution has not yet evaluated quantitatively.
2. Experimental The excimer laser source is Lambda Physik LPX 210i. The wave length is 248 nm, and the output is 600 mJ maximum. The beam is homogenized up to "5% over 2 = 45 mm2 slit shape area. Illumination is Kohler type. The mask consists of a dielectric ¨ multilayer reflector on quartz substrate, with five or six pairs of 1r4 wavelength HfO 2 and SiO 2 layers. The mask is fabricated by ion milling for 3Y wafer and by RIE for larger size up to 225 = 300 mm2 . The mask pattern is imaged on a polymer film through a 1 : 1 telecentric symmetrical lens with NA 0.1 and 55 mm aperture. Ablation depth is measured by a profiler, DEKTAK 16000 and the edge shapes are observed by SEM. Dynamic behaviours of ablation is measured using HSSP ŽHigh Speed Shadow Photography.. The probe beam is 444 nm dye laser with 5 ns pulse width. The minimum delay time is 25 ns. All the ablations are at 100 mJrcm2 in air. The measurements are performed using automatically controlled X–Y stage and a position synchronized laser firing output system ŽPSO.. The polymer materials investigated are phenol novolac, PMMA, polyimide, polyurethane, urea, melamine, polystyrene, and ketone. The former two materials are resist for UV and EB, and the remainings are coating materials. These are commercially available and amount to more than 120 samples. To elucidate interrelations between material structure and the ablation rate, we examined additionally newly synthesized 28 variations of polyimide and more than 50 kinds of polyurethane.
3. Results 3.1. Absorption coefficient and ablation rate The high ablation rate at low fluence means a low threshold, however, this value is inconvenient to use as a criterion to look for high rate, as it is only a result of ablation measurements. In the course of the investigation, we found a strong correlation between the absorption coefficient and the ablation rate as shown in Fig. 1. The cross point at the abscissa, 9 = 10 3 cmy1 approximately, seems to be a threshold, below which no ablation occurs. Actually, the materials with absorption coefficient lower than 1 = 10 4 cmy1 show very little or no ablation even for higher fluence. This characteristic definitely depends upon the primary structure of the polymers. Actually, among the above eight primary structures, ablation takes place for novolac, polyimide ŽPI. and polyurethane ŽPU.. The others show lower absorption coefficients, and very little ablation. To elucidate the effects of the details of primary structures, we measured samples replacing two constituent radicals in basic structure of PI. The 28 polyimide variations obtained in this way result only slight changes of the ablation rate. The rate values are scattered randomly almost within interval of 0.01 m mrshot, although the absorption coefficient varies
Fig. 1. Correlation between absorption coefficient and ablation rate of polymer materials. The absorption coefficients of PMMA, melamine, polystyrene, urea, and ketone are less than 1=10 4 cmy1 , and very little or no ablation at 100 mJrcm2 for 248 nm.
K. Suzuki et al.r Applied Surface Science 127–129 (1998) 905–910
Fig. 2. Correlation between absorption coefficient and ablation rate of polyimides. Each point corresponds to a substitution of radicals at R1 and R2 configurations.
from 13 to 23 = 10 4 cmy1 as shown in Fig. 2. Accordingly, the correlation between absorption coefficient and ablation rate is only distinguished among the primary structures, but slight among the variations of one primary structure. From the observations, it is concluded that the variations within one type polymer are due to the secondary structures, as they give more variety than the primary structure. To investigate the effects of the secondary structures, PI is inadequate due to its rigid primary structure. On the other hand, PU shows varieties of secondary structures by the combinations of the constituent polyisocyanate and polyol w2,3x. We have examined more than fifty samples of PU variations newly synthesized, and obtained the following general rules. Ž1. Among the isocyanate, TDI Žtoluene diisocyanate. is most effective for high ablation rate. MDI Ždiphenylmethane diisocyanate. is next to TDI. PU consists of HDI Žhexamethylene diisocyanate. shows very little or no ablation. Ž2. Aromatic ring is essential for 248 nm absorption, however, only a structure boning hydrocarbon through intermediary of N Žphenyl carbamate. is effective to the ablation. The details of the mechanism is not known yet, but it is very definite the
907
structure results from a photochemical dissociation of the polymer structure. Ž3. Although there are possibilities of light absorption and photodissociation at bonds in the aliphatic chains of polyol, they are far less effective compared to isocyanates. Ž4. The ring structure Žcyanurate. produces rather large sized debris. The rate of the structure is less than that of the chain structure. According to the general rule, the best structure consists of TDI with some long chains, and a sample shows the edge ablation rate Žsee Section 3.2. of 0.13 m mrshot. This seems to predict a possibility to obtain higher rate resists by modifications of PU secondary structure. Another aspect of importance of secondary structures is the thermal property. As is well known, the polymers are classified into two categories; thermoplastic and thermosetting. Thermoplastic materials
Fig. 3. Ablation depth vs. shot number. Ža. A novolac; Žb. a polyurethane.
908
K. Suzuki et al.r Applied Surface Science 127–129 (1998) 905–910
3.3. Plume dynamics The difference in the rate at the edge and the centre suggests the contributions from dynamic behaviors of the dissociated segments. The most effi-
Fig. 4. Size dependence of ablation rate. Illumination is through a dielectric multilayer mask. The slit length is 500 m m, and the width are 50, 100, 200 and 500 m m.
show melting at ablation. This is considered to be due to motions of long flexible chains around rigid segments, and somehow ‘macroscopic’ effect. It is essential to select thermosetting polymers for lithography applications. 3.2. Non-uniformity of the ablation rate oÕer the illumination area Ablation depths have been measured for 1 mm2 square illumination area, and it was found that the rate at the edge is higher than that of the centre. This behaviour is verified all the patterns as circle and rectangle. It is a characteristic of low fluence ablation, and the difference almost disappears for the fluence larger than 200 mJrcm2 . The practical ablation rate is naturally defined by completion of ablation over the whole area, and so the ablation rate is determined by the rate at the centre. This value is in part dependent upon material, and at the same time explosive behaviour of ablation as is seen in Section 3.3. On the other hand, the rate of the edge is not so dependent upon the material. These features are shown in Fig. 3, where two typical materials, high Žpolyurethane. and low Žnovolac. rate, are compared. PI is just the intermediate of both. The higher ablation rate at the edge suggests that the rate will be enhanced if the illumination size is reduced. This is proven in Fig. 4. At 50 m m slit width, the rate is completely the same to that of the edge of larger sizes.
Fig. 5. The change of the positions of shock and contact fronts vs. delay time. Ža. Definition of the shock and the contact front. The picture is for a novolac at 1 m s delay. Žb. Shock front positions of a novolac and a polyurethane. Žc. Contact front position of a novolac and a polyurethane.
K. Suzuki et al.r Applied Surface Science 127–129 (1998) 905–910
909
Fig. 6. An example of plume evolution. The material is a novolac. Each delay time is shown under the picture.
cient experiment is HSSP w4,5x. It gives a total feature of the phenomenon, although qualitative. From the observations, the dynamics of the ejecta are divided into two periods. One is a period described by general gas dynamics, where light mass ejecta act as piston to produce shock wave w6x. The shadow photographs in this time domain characterized by two discontinuities, the shock front and the contact front w7x. This period ceases when the speed of shock front becomes less than the sound velocity, nearly 10 m s after excimer incidence. The displacements of both the fronts are depicted in Fig. 5 for a PU and a novolac. Although there is nearly two times differences between ablation rates of two materials, there are only slight differences in the dynamic behaviours. Detailed observations reveal differences in the ejecta composition under the contact front. PU consists of more gaseous part than the novolac? The plume becomes distinct in the second period. The shadow photograph shows growth of mushroom shape cloud composed of heavier particles as is the case of atomic bomb as shown in Fig. 6. When the cloud heaves up, the root of the mushroom becomes thinner, as the debris are concentrating toward the centre of the illumination area. This dynamic behaviour will explain the rate difference between the edge and the centre. The plume of PU is far thinner than that of the novolac, and this is corresponding to gaseous ejecta at the first period.
4. Discussion HSSP shows lift-up of the ejecta from the surface at very early stage after illumination. Accordingly, it
is reasonable to premise a uniform dissociations over the whole illumination area in a certain depth, and to define the depth to be the rate at the edge, as it is not influenced by plume dynamics. As the rate at the edge is not so much different among the primary structures, this fact infers that the ablation is generated by a certain critical photon density inside the material. This is just a basis of thermal model of ablation. A novel fact is that there is a critical absorption coefficient to enable ablation. However, this value is so small that it is inadequate to interpret the absorbed photons maintain the critical photon density. Thus, the effect of absorption seems to act as detonator to the energized background, which is excited state of polymer as a whole, and is produced by polarizations of electron clouds due to incident photons field. The efficient way of the detonation may be photochemical dissociations of a primary structure. It is especially important in the case of low fluence, as one photon absorption is predominant. However, it is rather difficult to prove photodissociation, as the analysis by mass spectrometer and the structures of debris include some ambiguities another corroboration. The present phenyl carbamate is inferred to be the type of photochemical dissociation, but it is a kind of ‘circumstantial evidence’. We need a time resolved observations such as done for triazine polymers w8x to confirm the property. The second point is contributions from the secondary structures. PU is one of a few materials to realize varieties of the structures. Among them, the most prominent is the difference between the ring structure Žcyanurate ring. and the chain structure. The ring structure is synthesized to increase the absorption centres, that is, urethane bonds, however,
910
K. Suzuki et al.r Applied Surface Science 127–129 (1998) 905–910
the rate is less than the chain structures. This circumstances are very similar to PI, which fundamentally have ring structures and actually produce much larger sized debris. These observations should be analyzed more in detail, as it may give us some insights into the dissociation mechanisms. We have observed the effects of addition of high absorbance chromophores such as pyrene and DRI w9x. The result is that it is only effective for low ablation rate materials. Even at high fluence, photons energy are confined inside the chromophores and not transferred to the main chains as is pointed out in the investigations of the effects of chromophores w10x. In conclusion, among the eight primary structures, PU is a promising material for the high rate ablation resists at present. It is owing to the versatility for the secondary structures, and the phenyl carbamate structure in the primary structure, which enable photochemical dissociation. Together with the size effects of the ablation, we could obtain a good prospect for high throughput EAL compatible to the present technology.
Acknowledgements The authors wish to thank Prof. P.E. Dyer, Dr. P. Key, and Dr. S. Bennett of Hull University for HSSP measurements, and analyses of PU, and Mr. M. Fukuda of Dainippon Ink and Chemicals, for syntheses of PU, Dr. F. Kataoka of Production Engineering Research Laboratory, Hitachi, for syntheses of PI.
References w1x w2x w3x w4x w5x w6x w7x w8x w9x w10x
K. Suzuki et al., Proc. SPIE 2992 Ž1997. 98. S. Kuper, J. Brannon, Proc. SPIE 1598 Ž1991. 27. ¨ V. Zafropulos et al., Opt. Quantum Electron. 27 Ž1995. 1359. P. Simon, Appl. Phys. B 48 Ž1989. 253. R. Srinivasan et al., Appl. Phys. Lett. 55 Ž1989. 270. B. Sterending, J. Appl. Phys. 45 Ž1974. 3507. B. Braren et al., Nucl. Instr. Meth. B 58 Ž1991. 463. Th. Lippert et al., J. Phys. Chem. 97 Ž1993. 12296. J.C. Chon, P.B. Comita, Opt. Lett. 19 Ž1994. 1840. H. Fujiwara et al., J. Phys. Chem. 99 Ž1995. 11481.
Polymer resist materials for excimer ablation lithography Kenkichi Suzuki ) , Masaaki Matsuda, Nobuaki Hayashi Electron Tube and DeÕices DiÕision, Hitachi, Hayano, Mobara, Chiba 297, Japan
Abstract Excimer ablation lithography ŽEAL. is a new lithography suitable to TFT-LCD. Among the constituent technologies, the polymer materials for the resist is most crucial, as high ablation rates are required at low fluence. To elucidate the ablation mechanism at low fluence we specified some characteristic features through observations of about 200 polymers. The low fluence features are explained by the contributions from the primary and secondary structures, and ablation dynamic process. From the results, it is derived that polyurethane is most promising material for EAL, and the design of the secondary structure is essential to improve the ablation rate. q 1998 Elsevier Science B.V. Keywords: Polymer; Resist; Excimer ablation lithography; Polyurethane
1. Introduction Excimer ablation lithography ŽEAL. is a new concept of lithography characterized by self-development and ablation removal of resist. Owing to reduction of process steps and disuse of conventional developer and remover equipments and accompanying chemicals, we can expect considerable cost reduction of the lithography processes. The feasibility for TFT-LCD has been proven with respect to the precision and size w1x. However, there remains a problem of rather longer throughput time, 4–5 times than that of the conventional technology. The throughput time itself is a kind of figure of merit, composed of various constituent technologies. Among them, the most crucial one is resist materials.
)
Corresponding author. Tel.: q81-475-25-9062; fax: q81475-24-2463; e-mail: [email protected].
The ablation rate of the resist directly determines throughput time, and the maximum rate at present is 0.05 m mrshot at 248 nm. The rate of more than two times this figure is required to catch up and cope with the conventional lithography. This low rate is due to low fluence, 100 mJrcm2 typically, which is indispensable to prevent damage to underneath TFT films. The fundamental problem is why we could not find materials with a rate higher than 0.05 m mrshot. In other words, it is very essential to clarify the mechanism of polymer ablation at low fluence again from the view point of lithography applications. The ablations mechanisms have been investigated very extensively, but they were mainly concerned with higher fluence region, as the applications have been for micromachining such as boring of PCB. Ablation phenomenon is generally understood to be a kind of explosion bombed by photons. This implies that the ablation is not so much dependent on
0169-4332r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII S 0 1 6 9 - 4 3 3 2 Ž 9 8 . 0 0 1 0 8 - 1
906
K. Suzuki et al.r Applied Surface Science 127–129 (1998) 905–910
the details of molecular structure. However, it is not clear what extent the material properties is reflected to oblation phenomenon. It is very important to estimate the contributions from molecular structures for practical applications. This paper presents the results of observations of low fluence oblations of more than 200 samples derived from eight primary polymers, and clarifies two characteristic features. These are explained by primary and secondary structure of polymers, and the general explosion process, although each contribution has not yet evaluated quantitatively.
2. Experimental The excimer laser source is Lambda Physik LPX 210i. The wave length is 248 nm, and the output is 600 mJ maximum. The beam is homogenized up to "5% over 2 = 45 mm2 slit shape area. Illumination is Kohler type. The mask consists of a dielectric ¨ multilayer reflector on quartz substrate, with five or six pairs of 1r4 wavelength HfO 2 and SiO 2 layers. The mask is fabricated by ion milling for 3Y wafer and by RIE for larger size up to 225 = 300 mm2 . The mask pattern is imaged on a polymer film through a 1 : 1 telecentric symmetrical lens with NA 0.1 and 55 mm aperture. Ablation depth is measured by a profiler, DEKTAK 16000 and the edge shapes are observed by SEM. Dynamic behaviours of ablation is measured using HSSP ŽHigh Speed Shadow Photography.. The probe beam is 444 nm dye laser with 5 ns pulse width. The minimum delay time is 25 ns. All the ablations are at 100 mJrcm2 in air. The measurements are performed using automatically controlled X–Y stage and a position synchronized laser firing output system ŽPSO.. The polymer materials investigated are phenol novolac, PMMA, polyimide, polyurethane, urea, melamine, polystyrene, and ketone. The former two materials are resist for UV and EB, and the remainings are coating materials. These are commercially available and amount to more than 120 samples. To elucidate interrelations between material structure and the ablation rate, we examined additionally newly synthesized 28 variations of polyimide and more than 50 kinds of polyurethane.
3. Results 3.1. Absorption coefficient and ablation rate The high ablation rate at low fluence means a low threshold, however, this value is inconvenient to use as a criterion to look for high rate, as it is only a result of ablation measurements. In the course of the investigation, we found a strong correlation between the absorption coefficient and the ablation rate as shown in Fig. 1. The cross point at the abscissa, 9 = 10 3 cmy1 approximately, seems to be a threshold, below which no ablation occurs. Actually, the materials with absorption coefficient lower than 1 = 10 4 cmy1 show very little or no ablation even for higher fluence. This characteristic definitely depends upon the primary structure of the polymers. Actually, among the above eight primary structures, ablation takes place for novolac, polyimide ŽPI. and polyurethane ŽPU.. The others show lower absorption coefficients, and very little ablation. To elucidate the effects of the details of primary structures, we measured samples replacing two constituent radicals in basic structure of PI. The 28 polyimide variations obtained in this way result only slight changes of the ablation rate. The rate values are scattered randomly almost within interval of 0.01 m mrshot, although the absorption coefficient varies
Fig. 1. Correlation between absorption coefficient and ablation rate of polymer materials. The absorption coefficients of PMMA, melamine, polystyrene, urea, and ketone are less than 1=10 4 cmy1 , and very little or no ablation at 100 mJrcm2 for 248 nm.
K. Suzuki et al.r Applied Surface Science 127–129 (1998) 905–910
Fig. 2. Correlation between absorption coefficient and ablation rate of polyimides. Each point corresponds to a substitution of radicals at R1 and R2 configurations.
from 13 to 23 = 10 4 cmy1 as shown in Fig. 2. Accordingly, the correlation between absorption coefficient and ablation rate is only distinguished among the primary structures, but slight among the variations of one primary structure. From the observations, it is concluded that the variations within one type polymer are due to the secondary structures, as they give more variety than the primary structure. To investigate the effects of the secondary structures, PI is inadequate due to its rigid primary structure. On the other hand, PU shows varieties of secondary structures by the combinations of the constituent polyisocyanate and polyol w2,3x. We have examined more than fifty samples of PU variations newly synthesized, and obtained the following general rules. Ž1. Among the isocyanate, TDI Žtoluene diisocyanate. is most effective for high ablation rate. MDI Ždiphenylmethane diisocyanate. is next to TDI. PU consists of HDI Žhexamethylene diisocyanate. shows very little or no ablation. Ž2. Aromatic ring is essential for 248 nm absorption, however, only a structure boning hydrocarbon through intermediary of N Žphenyl carbamate. is effective to the ablation. The details of the mechanism is not known yet, but it is very definite the
907
structure results from a photochemical dissociation of the polymer structure. Ž3. Although there are possibilities of light absorption and photodissociation at bonds in the aliphatic chains of polyol, they are far less effective compared to isocyanates. Ž4. The ring structure Žcyanurate. produces rather large sized debris. The rate of the structure is less than that of the chain structure. According to the general rule, the best structure consists of TDI with some long chains, and a sample shows the edge ablation rate Žsee Section 3.2. of 0.13 m mrshot. This seems to predict a possibility to obtain higher rate resists by modifications of PU secondary structure. Another aspect of importance of secondary structures is the thermal property. As is well known, the polymers are classified into two categories; thermoplastic and thermosetting. Thermoplastic materials
Fig. 3. Ablation depth vs. shot number. Ža. A novolac; Žb. a polyurethane.
908
K. Suzuki et al.r Applied Surface Science 127–129 (1998) 905–910
3.3. Plume dynamics The difference in the rate at the edge and the centre suggests the contributions from dynamic behaviors of the dissociated segments. The most effi-
Fig. 4. Size dependence of ablation rate. Illumination is through a dielectric multilayer mask. The slit length is 500 m m, and the width are 50, 100, 200 and 500 m m.
show melting at ablation. This is considered to be due to motions of long flexible chains around rigid segments, and somehow ‘macroscopic’ effect. It is essential to select thermosetting polymers for lithography applications. 3.2. Non-uniformity of the ablation rate oÕer the illumination area Ablation depths have been measured for 1 mm2 square illumination area, and it was found that the rate at the edge is higher than that of the centre. This behaviour is verified all the patterns as circle and rectangle. It is a characteristic of low fluence ablation, and the difference almost disappears for the fluence larger than 200 mJrcm2 . The practical ablation rate is naturally defined by completion of ablation over the whole area, and so the ablation rate is determined by the rate at the centre. This value is in part dependent upon material, and at the same time explosive behaviour of ablation as is seen in Section 3.3. On the other hand, the rate of the edge is not so dependent upon the material. These features are shown in Fig. 3, where two typical materials, high Žpolyurethane. and low Žnovolac. rate, are compared. PI is just the intermediate of both. The higher ablation rate at the edge suggests that the rate will be enhanced if the illumination size is reduced. This is proven in Fig. 4. At 50 m m slit width, the rate is completely the same to that of the edge of larger sizes.
Fig. 5. The change of the positions of shock and contact fronts vs. delay time. Ža. Definition of the shock and the contact front. The picture is for a novolac at 1 m s delay. Žb. Shock front positions of a novolac and a polyurethane. Žc. Contact front position of a novolac and a polyurethane.
K. Suzuki et al.r Applied Surface Science 127–129 (1998) 905–910
909
Fig. 6. An example of plume evolution. The material is a novolac. Each delay time is shown under the picture.
cient experiment is HSSP w4,5x. It gives a total feature of the phenomenon, although qualitative. From the observations, the dynamics of the ejecta are divided into two periods. One is a period described by general gas dynamics, where light mass ejecta act as piston to produce shock wave w6x. The shadow photographs in this time domain characterized by two discontinuities, the shock front and the contact front w7x. This period ceases when the speed of shock front becomes less than the sound velocity, nearly 10 m s after excimer incidence. The displacements of both the fronts are depicted in Fig. 5 for a PU and a novolac. Although there is nearly two times differences between ablation rates of two materials, there are only slight differences in the dynamic behaviours. Detailed observations reveal differences in the ejecta composition under the contact front. PU consists of more gaseous part than the novolac? The plume becomes distinct in the second period. The shadow photograph shows growth of mushroom shape cloud composed of heavier particles as is the case of atomic bomb as shown in Fig. 6. When the cloud heaves up, the root of the mushroom becomes thinner, as the debris are concentrating toward the centre of the illumination area. This dynamic behaviour will explain the rate difference between the edge and the centre. The plume of PU is far thinner than that of the novolac, and this is corresponding to gaseous ejecta at the first period.
4. Discussion HSSP shows lift-up of the ejecta from the surface at very early stage after illumination. Accordingly, it
is reasonable to premise a uniform dissociations over the whole illumination area in a certain depth, and to define the depth to be the rate at the edge, as it is not influenced by plume dynamics. As the rate at the edge is not so much different among the primary structures, this fact infers that the ablation is generated by a certain critical photon density inside the material. This is just a basis of thermal model of ablation. A novel fact is that there is a critical absorption coefficient to enable ablation. However, this value is so small that it is inadequate to interpret the absorbed photons maintain the critical photon density. Thus, the effect of absorption seems to act as detonator to the energized background, which is excited state of polymer as a whole, and is produced by polarizations of electron clouds due to incident photons field. The efficient way of the detonation may be photochemical dissociations of a primary structure. It is especially important in the case of low fluence, as one photon absorption is predominant. However, it is rather difficult to prove photodissociation, as the analysis by mass spectrometer and the structures of debris include some ambiguities another corroboration. The present phenyl carbamate is inferred to be the type of photochemical dissociation, but it is a kind of ‘circumstantial evidence’. We need a time resolved observations such as done for triazine polymers w8x to confirm the property. The second point is contributions from the secondary structures. PU is one of a few materials to realize varieties of the structures. Among them, the most prominent is the difference between the ring structure Žcyanurate ring. and the chain structure. The ring structure is synthesized to increase the absorption centres, that is, urethane bonds, however,
910
K. Suzuki et al.r Applied Surface Science 127–129 (1998) 905–910
the rate is less than the chain structures. This circumstances are very similar to PI, which fundamentally have ring structures and actually produce much larger sized debris. These observations should be analyzed more in detail, as it may give us some insights into the dissociation mechanisms. We have observed the effects of addition of high absorbance chromophores such as pyrene and DRI w9x. The result is that it is only effective for low ablation rate materials. Even at high fluence, photons energy are confined inside the chromophores and not transferred to the main chains as is pointed out in the investigations of the effects of chromophores w10x. In conclusion, among the eight primary structures, PU is a promising material for the high rate ablation resists at present. It is owing to the versatility for the secondary structures, and the phenyl carbamate structure in the primary structure, which enable photochemical dissociation. Together with the size effects of the ablation, we could obtain a good prospect for high throughput EAL compatible to the present technology.
Acknowledgements The authors wish to thank Prof. P.E. Dyer, Dr. P. Key, and Dr. S. Bennett of Hull University for HSSP measurements, and analyses of PU, and Mr. M. Fukuda of Dainippon Ink and Chemicals, for syntheses of PU, Dr. F. Kataoka of Production Engineering Research Laboratory, Hitachi, for syntheses of PI.
References w1x w2x w3x w4x w5x w6x w7x w8x w9x w10x
K. Suzuki et al., Proc. SPIE 2992 Ž1997. 98. S. Kuper, J. Brannon, Proc. SPIE 1598 Ž1991. 27. ¨ V. Zafropulos et al., Opt. Quantum Electron. 27 Ž1995. 1359. P. Simon, Appl. Phys. B 48 Ž1989. 253. R. Srinivasan et al., Appl. Phys. Lett. 55 Ž1989. 270. B. Sterending, J. Appl. Phys. 45 Ž1974. 3507. B. Braren et al., Nucl. Instr. Meth. B 58 Ž1991. 463. Th. Lippert et al., J. Phys. Chem. 97 Ž1993. 12296. J.C. Chon, P.B. Comita, Opt. Lett. 19 Ž1994. 1840. H. Fujiwara et al., J. Phys. Chem. 99 Ž1995. 11481.