- Email: [email protected]
Physical and chemical aspects of PMMA vapour development
364
Nuclear
Instruments
and Methods
in Physics Research
B46 (1990) 364-368 North-Holland
PHYSICAL
AND CHEMICAL
E.M. LEHOCKEY, Department
of Materids
ASPECTS
I. REID
and I. HILL
Engineering,
The Univemty
OF PMMA VAPOUR
of Wesrern Ontario,
London,
DEVELOPMENT
Ontario
N6A
SB9, Canudu
The physical and chemical response of Poly(methy1 lne~ha~~lat~) PMMA films and their spontaneous “ vapourization” during X-ray bombardment is examined. Three micrometer PMMA films were spun onto SO, substrates, and irradiated using Al X-rays. The film thickness vapourized is shown to increase as the initial molecular weight decreases; and the rate of vapour development is shown to diminish with increasing dose. Raman spectroscopy studies of irradiated films indicate that vapour development arises from depol~merization. resulting from main chain scission induced by ester side group (C-O-CH,) abstraction. At large X-ray doses the Raman spectra show the evolution of primary C=C bonds between chain molecules at end group sites activated during the scission process. This increases the molecular weight and acts to oppose the depolymerization process, resulting in a reduction in the observed erosion rate with dose
1. Introduction
PMMA films were subsequently annealed at 200 o C for 60 min to remove residual solvents and stresses developed during spinning. Samples were partially masked with a 3 pm thick lead strip and irradiated with X-rays in a vacuum. X-rays were generated using an 11 kV Veeco VeB-6 electron gun evaporator unit with a target carouse1 containing Cu, Al and carbon slugs. The X-ray flux was monitored by the photoelectric current emitted from a gold surface placed near the samples. Dose rates are therefore reported in units of PA/cm’. The extent of film erosion was established by measuring the height differential between the exposed and masked regions of the film using a Sloan Dektak II profilometer. The thickness vapourized as a function of dose was compared for films having different initial molecular weights and wavelength distribution exposures. Raman spectra were obtained using a Dilor OMARS X9 multichannel instrument. A series of spectra were generated for unirradiated films, and for samples having been irradiated using Al X-rays to doses of 0.5, 6.0 and 16.0 ( ~AA/cm2)min. Variations in the peak intensities corresponding to species contained within the PMMA structure were monitored as a function of dose. Spectra obtained at all doses were accumulated for 300 s in the range between 550 and 1750 cm--‘. pm
The response of polymer systems to ionizing radiation is governed by two competing chemical reactions: Cross-linking and chain s&ion. Cross-linking increases the polymer’s average molecular weight by generating primary bonds between chain molecules. On the other hand. chain scission, arising from random chain fracture, reduces the average molecular weight of the polymer, The net effect on the polymer is determined by the dominance of one reaction over the other [l]. If scission dominates it is postulated that material loss occurs vis B vis a “reduction” of the polymer chains into simple gaseous free radicals [2]. Subsequent expulsion of gaseous scission products by diffusion results in a physical erosion of polymers known as vapour development (31. Polymers which exhibit large erosion rates may be incorporated into the lithographic process to eliminate wet chemical etching of the polymer after exposure. Unfortunately. PMMA has traditionally not been considered as a good vapour developing polymer because of its low scission yield. Hatzakis reports only a few hundred Bngstriims of erosion during e-beam bombardment of PMMA 141. In this work we show that substantial vapourized thicknesses can be achieved with PMMA under X-ray exposure. Methods for enhancing the erosion yield are discussed along with the chemistry responsible for vapour development.
3. Experimental results 2. Experimental procedure
3. I. The physical response
Solutions of 20 + 2 wt.% PMMA were prepared using a chlorobenzene solvent and spun onto a layer of SiO,; thermally grown atop pure Si wafers. The resulting 3
Fig. 1 shows the thickness vapourized as a function of dose
016~-58~X/90/$03.50 (North-Holland)
8 Efsevier Science Publishers
B.V.
lar
weights
(M,.
= 443
000
of
the
PMMA
for two initial
in (a) and
M,
film
molecu-
= 30000
in
E. M. Lehockey et (11./ PMMA uapour derlelopment 0.7 0.6 2 -? 0.5 z 2 0.4 3 0 E 0.3 T1 0 0.2 .N L zp
0.1
9 0
5
X-ray
10
25
30
Dose { uA*min/cm2
j
15
20
35
Fig. 1. Vapourized thickness as a function of dose for variations in initial molecular weight, and incident X-ray wavelength. (a) M, = 443000 Al K; (b) M, = 30000 and Al K; (c) M, = 443000 and Cu-K
(b)). Both vapour development profiles show an initially rapid rate of increase in vapour development yield with dose {slope = 0.04 f 0.005 ).LA cm-’ mm)}. However, the rate at which vapour development proceeds decreases with dose; approaching zero at a dose of 30 ( pA/cm*)min, as is evidenced by the slope of curves (a) and (b). The rate of vapour development declines more
365
rapidly for larger initial molecular weights. As a consequence, for the PMMA of 30000 initial molecular weight, a vapourized thickness of 0.65 pm was achieved for a dose of 30 (pA/cm2)min, while only 0.4 km of the 443 000 molecular weight PMMA film was removed at this dose. The results were checked for a dose rate dependence using fluxes of 1.0 + 0.1 (PA/cm’) and 4.0 + 0.3 ( pA/cm2), but none was detected. Fig. 1 also indicates that the vapour development profile is dependent on the wavelength distribution of the incident radiation used. Reducing the wavelength of the incident X-rays from Al-K (curve a) to the Cu-K (curve c) curtailed the thickness vapourized by a factor of 2 at a dose of 30 ().~A/cm~)min. The initial rate of erosion was likewise reduced with the change in target material. 3.2. The chemical response Fig. 2(b) details the variation in the Raman spectra for bands in the range from 550-1050 cm-’ as dose is increased from 0 to 16 ( t.tA/cm2)min. The intensity of the band at 800 cm-‘. corresponding to the stretching mode of the singly bonded ester group oxygen (fig. 2a) [5], was observed to decrease 86% f 8% through a dose of 16 (pA/cm*)min. An identical percentage decrease was exhibited by the peak representing the ester group backbone bonds situated at 600 cm-’ [5].
(a 14
13 100
cb)
12
11
cm-’
DOSE (IA/cm*
n c=o
min
6
16 17
16
15
14
Fig. 2. (a) The Poly(methylmethacrylate) monomer unit. (b) The variation in the Raman spectrum as a function of dose for the spectral range of 600 to 1050 cm-‘. The irradiation dose from 0 to 16 (pA/cm*)min is indicated. (c) The variation in the Raman spectrum as a function of dose for the range of 1050 to 1400 cm-’ obtained upon irradiating to doses of 0 and 16 ( pA/cm2)min. (d) The variation in Raman band intensities as a function of dose for bands in the range 1400 to 1750 cm-‘. VI. POLYMERS
,’ ORGANICS
366
E. M. Lehockqv er cd. / PMMA
Fig. 2(c) compares the Raman spectrum in the range between 1050-1450 cm-’ obtained for unirradiated PMMA with that recorded upon irradiating to a dose of 16 (pAA/cm’)min. The broad, low intensity peak at 1125 cm -’ arises from a C-C vibration in the PMMA main backbone chain [5]. The decline experienced by this band amounts to 45% of its original intensity. Fig. 2(d) delineates the change in the Raman spectrum with dose monitored in the spectral range between 1400-1750 cm -‘, The peak at 1450 cm-’ reveals a 50 + 4% decline in C-H species intensity from its original value over the dose range between 0 and 16 (~AA/cm’)min. Half of the signal contributing to this peak emanates from the CH, within the ester group, while the remaining half originates from bonds within the methyl side group (fig. 2a) [5]. Fig. 2(d) also reveals that, accompanying irradiation. is the emergence of a band at 1640 cm ‘. While this peak does not correspond to any structure contained within unirradiated PMMA it is indicative of the stretching mode associated with the C=C species [5].
4. Discussion 4.1. Chemical response The identical 86 f 8% reduction in intensity exhibited by both the main ester species (600 cm-‘) and the singly bonded ester oxygen (800 cm-‘) suggests that the ester side group is abstracted during irradiation. This is in agreement with our previously reported XPS data for the PMMA 0,, peak which displays a 75 + 5% reduction in the singly bonded oxygen concentration upon irradiating to a dose of 16 (+A/cm’ )min [6]. Ester group removal and/or dissociation is further substantiated by the 50% decline in the CH, concentration (1450 cm-‘). Given the dual origin of the CH, signal its decline in magnitude. relative to the C-0-C concentration, suggests the CH, loss may be attributed almost entirely to the loss of this species contained in the ester group. This implies that the other half of the CH, species, comprising the methyl side group, is not liberated during irradiation. Preferential ester abstraction is consistent with gas chromatography results obtained by Todd [3] and IR data reported by Taylor [7]. The relative “instability” of the ester group may be partially attributed to its greater interaction cross-section for scission resulting from its larger size and “radius of conformation”, as compared with the methyl side group [1.2]. The 45% decrease in intensity associated with the CC skeletal bond (fig 2c: 1125 cm-‘) amounts to half of the 86% attenuation in the ester species’ intensity over the same dose range. Since the PMMA structure accommodates two skeletal C-C bonds for every ester
cqour
dewlopmenr
group, this suggests that the remocd ~jone ester group I.Y uccompunied by one backbone chain .sci.won. Destabilization and subsequent fracture of a main chain molecule resulting from side group scission is commensurate with a Norrish Type I reaction [S]. The Norrish reaction is dominant in polymers containing C=O carboxyl species. Evidence connecting this Norrish reaction to physical erosion can be obtained by correlating figs. 1 and 2(b). According to fig. 1, the greatest erosion rate occurs at doses below 8 (~A/cm’)min. The Raman data of fig. 2(c) illustrates that the greatest ester group loss also occurs in this dose range. Previously reported dissolution studies on PMMA indicate that irradiation through this dose range renders the greatest increase in polymer solubility corresponding to a reduction in fractured molecular weight (9,101. This substantiates the connection between depolymerization and vapour development. The results in fig. 2(d) show that C=C sites are generated during irradiation. Given its shape and spectral position, this peak arises from a bond contained within a long skeletal chain of high molecular weight. We have previously shown that a decrease in PMMA dissolution rate and a corresponding increase in the molecular weight occurs at high doses where the C=C population density is largest [9]. Evolution of the C=C species is thought to be initiated in the following manner: As illustrated in figs. 3(i) and (ii). main chain scission proceeds by ester group abstraction. resulting
c=o
c=o
k=O
;-CH3
d-CH3
;-CH3
CH
CH
I 3
(iii)
-CH?$.
6-CH;
_t
.CH$ y=o 0-CH,
CH,
(iv)
-h-CH&
CH,
CH, +
CH,
.&C”&-
A=0
t=o
&-CH3
0-CH,
CH
CH 13 -C-CH,-c=C-CH&
I 3
CH,
CH
I
Lo t=O 0-CH,
A-CH, \
Fig. 3. Illustrates and recombination incident radiation.
end-
link
the mechanism for chain scission (i) to (iii) (iv) activated in PMMA in response to Shows recombination is scission dependent.
E. M. Lehockey et al. / PMMA vapour development
from secondary electron yield produced by primary photon interactions. Scission sites activated near the surface generate free radicals from the dissociation of ester groups and skeletal chains. Close proximity to the surface permits the diffusion of free radicals from the polymer without substantial recombination along the short diffusion length. This results in material loss. The decrease in molecular weight accompanying scission increases the number density of dangling end group bonds (see fig. 3(iii)), thus increasing the probability that two end groups will be in close enough proximity to bond (fig. 3(iv)). Valency restrictions dictate that a double bond carbon would form the end-link [l]. Recombination of this type opposes the reduction of larger chains into mobile gaseous species required to further vapour development. A significant feature of the Raman results is that C=C end-links appear to “repair” scission activated end group sites. Hence, the population of potential recombination/cross-linking sites, and therefore the rate of cross-linking is dependent on the number of scission sites activated. This would account for the rapid initial rate of vapourization resulting from dominant scission in the low dose regime were the number of potential cross-linking sites is few. The dominance of recombination at larger doses would account for the decline in the vapour development rate observed in fig. 1 at doses in excess of 8 ()rA/cm’)min. 4.2. The physical response The decline in rate of vapourization with dose shown in Fig. 1 is consistent with data reported by Bowden for PBS polymers [ll]. Using e-beam radiation, Bowden achieved a vapour developed thickness limited of 3325 A, as the rate of increase approached zero, as compared with 4523 A obtained here for PMMA. Under electron bombardment, PBS is known to have a scission yield (G-value) of ten times that of PMMA. X-ray exposure thus appears to enhance vapour development sensitivity, when compared to electron bombardment. Fig. 1 illustrates two approaches for maximizing the vapour development yield. A comparison of curves (a) and (c) reveals that, by increasing the wavelength of the incident X-rays, increases in the vapourization sensitivity of PMMA may be affected. It is thought that this increase in vapourization yield reflects the increase in the linear absorption with increasing wavelength. Enhanced absorption increases the energy deposited within the vapour development depth, resulting in a greater number of activated scission sites; thus augmenting the free radical formation per incident photon. In the case of the Cu-K X-rays a larger portion of the incident energy is absorbed in the bulk were contribution to vapour development is reduced. This accounts for the higher initial vapourization rate, obtained with Al-K X-rays as compared to Cu-K incident radiation
361
indicated by slopes S(A1) and S(Cu) in fig. 1. This change in yield with wavelength distribution qualitatively agrees with results by Henderson for the exposure of nitrocellulose films in the visible wavelength regime
WI. Profiles (a) and (b) of fig. 1 illustrate that a significant enhancement in the vapour development yield sensitivity can also be achieved by reducing the initial molecular weight of the PMMA. The increase in vapour development with decreasing initial molecular weight differs from vapourization profiles reported by Bowden for PBS polymers [ll]. On the basis of a reduction in the vapourization yield of PBS with decreasing initial weight, Bowden proposed an unzipping depolymerization mechanism involving the spontaneous scission of the backbone chain along its length following end group abstraction. Fig. 1 suggests that “molecular unzipping” does not occur in PMMA. but rather that depolymerization occurs in a discrete fashion. If PMMA depolymerization did result from unzipping, the yield would increase with initial molecular weight, due to a higher yield per activated end group. This concurs with the Raman results showing discrete main chain scission following ester removal. Finally, attempts were made to maximize the erosion yield by combining the above techniques. Exposing low initial molecular weight PMMA (M, = 30 000) to long wavelength X-rays obtained from a carbon target produced erosion thicknesses of 1.0 pm in relatively short exposure times of 5 min. From the point of view of feasibility in lithography, this represents a substantial improvement over the few hundred angstroms of vapourization achieved by electron beam irradiation [4].
5. Conclusions (1) The interaction of X-rays with PMMA initiates concurrent chain scission and recombination. (2) Scission proceeds by preferential ester group removal leading to discrete main chain scission. Both reactions generate free radicals which can diffuse toward the surface to affect material loss. (3) During irradiation. repair of fractured chains occurs with the evolution of C=C bonds at scission end group sites. Thus the degree of recombination depends on the scission yield. (4) Substantial amounts of material loss are possible with X-ray irradiation. With low initial molecular weights and longer wavelength exposures enhanced sensitivity yielding erosion depths of 1.0 pm can be attained! (5) The vapour development rate is initially rapid but decreases with dose as a result of the polymerization effect accrued to C=C recombination. VI. POLYMERS / ORGANICS
368
E.M. Lehockqv et al. / PMMA
The authors wish to express their appreciation to the members of Surface Science Western for the use of their equipment, and our many fruitful discussions.
oapour dec)elopment
[6] [7] [8]
References [I] M. Dole, Radiation Chemistry of Macromolecules. 2 (Acedemic Press. New York. 1973). [2] M.J. Bowden, J. Polym. Sci.: Polym. Symp. 49 (1975) 221. (31 A. Todd, J. Polym. Sci. 42 (1960) 223. [4] M. Hatzakis. J. Electrochem. Sot. 116 (1969) 1033. [5] I. Hill and M. Flieshman, in: Comprehensive Treatise of
[9] [lo] [ll] [12]
Electra-chemistry, vol. 8, ed. R. White (Pergamon. London. 1981) p. 373. E.M. Lehockey and 1. Reid. Surf. Interf. Anal. 11 (1988) 302. J. Taylor and W. Bohn, Anal. Chem. 53 (1981) 1082. J. Guillet, presented at 5th Int. Conf. Radiation Effects in Insulators. Hamilton. Ontario, Canada, June 1989. E.M. Lehockey and 1. Reid, Can. J. Phys. 65 (1986) 975. J. Greenich, J. Electrochem. Sot. 121 (1977) 121. M. Bowden and L. Thompson, Polym. Eng. Sci. 14 (1974) 525. D. Henderson and J. White. Appl. Phys. Lett. 46 (1985) 901.
Nuclear
Instruments
and Methods
in Physics Research
B46 (1990) 364-368 North-Holland
PHYSICAL
AND CHEMICAL
E.M. LEHOCKEY, Department
of Materids
ASPECTS
I. REID
and I. HILL
Engineering,
The Univemty
OF PMMA VAPOUR
of Wesrern Ontario,
London,
DEVELOPMENT
Ontario
N6A
SB9, Canudu
The physical and chemical response of Poly(methy1 lne~ha~~lat~) PMMA films and their spontaneous “ vapourization” during X-ray bombardment is examined. Three micrometer PMMA films were spun onto SO, substrates, and irradiated using Al X-rays. The film thickness vapourized is shown to increase as the initial molecular weight decreases; and the rate of vapour development is shown to diminish with increasing dose. Raman spectroscopy studies of irradiated films indicate that vapour development arises from depol~merization. resulting from main chain scission induced by ester side group (C-O-CH,) abstraction. At large X-ray doses the Raman spectra show the evolution of primary C=C bonds between chain molecules at end group sites activated during the scission process. This increases the molecular weight and acts to oppose the depolymerization process, resulting in a reduction in the observed erosion rate with dose
1. Introduction
PMMA films were subsequently annealed at 200 o C for 60 min to remove residual solvents and stresses developed during spinning. Samples were partially masked with a 3 pm thick lead strip and irradiated with X-rays in a vacuum. X-rays were generated using an 11 kV Veeco VeB-6 electron gun evaporator unit with a target carouse1 containing Cu, Al and carbon slugs. The X-ray flux was monitored by the photoelectric current emitted from a gold surface placed near the samples. Dose rates are therefore reported in units of PA/cm’. The extent of film erosion was established by measuring the height differential between the exposed and masked regions of the film using a Sloan Dektak II profilometer. The thickness vapourized as a function of dose was compared for films having different initial molecular weights and wavelength distribution exposures. Raman spectra were obtained using a Dilor OMARS X9 multichannel instrument. A series of spectra were generated for unirradiated films, and for samples having been irradiated using Al X-rays to doses of 0.5, 6.0 and 16.0 ( ~AA/cm2)min. Variations in the peak intensities corresponding to species contained within the PMMA structure were monitored as a function of dose. Spectra obtained at all doses were accumulated for 300 s in the range between 550 and 1750 cm--‘. pm
The response of polymer systems to ionizing radiation is governed by two competing chemical reactions: Cross-linking and chain s&ion. Cross-linking increases the polymer’s average molecular weight by generating primary bonds between chain molecules. On the other hand. chain scission, arising from random chain fracture, reduces the average molecular weight of the polymer, The net effect on the polymer is determined by the dominance of one reaction over the other [l]. If scission dominates it is postulated that material loss occurs vis B vis a “reduction” of the polymer chains into simple gaseous free radicals [2]. Subsequent expulsion of gaseous scission products by diffusion results in a physical erosion of polymers known as vapour development (31. Polymers which exhibit large erosion rates may be incorporated into the lithographic process to eliminate wet chemical etching of the polymer after exposure. Unfortunately. PMMA has traditionally not been considered as a good vapour developing polymer because of its low scission yield. Hatzakis reports only a few hundred Bngstriims of erosion during e-beam bombardment of PMMA 141. In this work we show that substantial vapourized thicknesses can be achieved with PMMA under X-ray exposure. Methods for enhancing the erosion yield are discussed along with the chemistry responsible for vapour development.
3. Experimental results 2. Experimental procedure
3. I. The physical response
Solutions of 20 + 2 wt.% PMMA were prepared using a chlorobenzene solvent and spun onto a layer of SiO,; thermally grown atop pure Si wafers. The resulting 3
Fig. 1 shows the thickness vapourized as a function of dose
016~-58~X/90/$03.50 (North-Holland)
8 Efsevier Science Publishers
B.V.
lar
weights
(M,.
= 443
000
of
the
PMMA
for two initial
in (a) and
M,
film
molecu-
= 30000
in
E. M. Lehockey et (11./ PMMA uapour derlelopment 0.7 0.6 2 -? 0.5 z 2 0.4 3 0 E 0.3 T1 0 0.2 .N L zp
0.1
9 0
5
X-ray
10
25
30
Dose { uA*min/cm2
j
15
20
35
Fig. 1. Vapourized thickness as a function of dose for variations in initial molecular weight, and incident X-ray wavelength. (a) M, = 443000 Al K; (b) M, = 30000 and Al K; (c) M, = 443000 and Cu-K
(b)). Both vapour development profiles show an initially rapid rate of increase in vapour development yield with dose {slope = 0.04 f 0.005 ).LA cm-’ mm)}. However, the rate at which vapour development proceeds decreases with dose; approaching zero at a dose of 30 ( pA/cm*)min, as is evidenced by the slope of curves (a) and (b). The rate of vapour development declines more
365
rapidly for larger initial molecular weights. As a consequence, for the PMMA of 30000 initial molecular weight, a vapourized thickness of 0.65 pm was achieved for a dose of 30 (pA/cm2)min, while only 0.4 km of the 443 000 molecular weight PMMA film was removed at this dose. The results were checked for a dose rate dependence using fluxes of 1.0 + 0.1 (PA/cm’) and 4.0 + 0.3 ( pA/cm2), but none was detected. Fig. 1 also indicates that the vapour development profile is dependent on the wavelength distribution of the incident radiation used. Reducing the wavelength of the incident X-rays from Al-K (curve a) to the Cu-K (curve c) curtailed the thickness vapourized by a factor of 2 at a dose of 30 ().~A/cm~)min. The initial rate of erosion was likewise reduced with the change in target material. 3.2. The chemical response Fig. 2(b) details the variation in the Raman spectra for bands in the range from 550-1050 cm-’ as dose is increased from 0 to 16 ( t.tA/cm2)min. The intensity of the band at 800 cm-‘. corresponding to the stretching mode of the singly bonded ester group oxygen (fig. 2a) [5], was observed to decrease 86% f 8% through a dose of 16 (pA/cm*)min. An identical percentage decrease was exhibited by the peak representing the ester group backbone bonds situated at 600 cm-’ [5].
(a 14
13 100
cb)
12
11
cm-’
DOSE (IA/cm*
n c=o
min
6
16 17
16
15
14
Fig. 2. (a) The Poly(methylmethacrylate) monomer unit. (b) The variation in the Raman spectrum as a function of dose for the spectral range of 600 to 1050 cm-‘. The irradiation dose from 0 to 16 (pA/cm*)min is indicated. (c) The variation in the Raman spectrum as a function of dose for the range of 1050 to 1400 cm-’ obtained upon irradiating to doses of 0 and 16 ( pA/cm2)min. (d) The variation in Raman band intensities as a function of dose for bands in the range 1400 to 1750 cm-‘. VI. POLYMERS
,’ ORGANICS
366
E. M. Lehockqv er cd. / PMMA
Fig. 2(c) compares the Raman spectrum in the range between 1050-1450 cm-’ obtained for unirradiated PMMA with that recorded upon irradiating to a dose of 16 (pAA/cm’)min. The broad, low intensity peak at 1125 cm -’ arises from a C-C vibration in the PMMA main backbone chain [5]. The decline experienced by this band amounts to 45% of its original intensity. Fig. 2(d) delineates the change in the Raman spectrum with dose monitored in the spectral range between 1400-1750 cm -‘, The peak at 1450 cm-’ reveals a 50 + 4% decline in C-H species intensity from its original value over the dose range between 0 and 16 (~AA/cm’)min. Half of the signal contributing to this peak emanates from the CH, within the ester group, while the remaining half originates from bonds within the methyl side group (fig. 2a) [5]. Fig. 2(d) also reveals that, accompanying irradiation. is the emergence of a band at 1640 cm ‘. While this peak does not correspond to any structure contained within unirradiated PMMA it is indicative of the stretching mode associated with the C=C species [5].
4. Discussion 4.1. Chemical response The identical 86 f 8% reduction in intensity exhibited by both the main ester species (600 cm-‘) and the singly bonded ester oxygen (800 cm-‘) suggests that the ester side group is abstracted during irradiation. This is in agreement with our previously reported XPS data for the PMMA 0,, peak which displays a 75 + 5% reduction in the singly bonded oxygen concentration upon irradiating to a dose of 16 (+A/cm’ )min [6]. Ester group removal and/or dissociation is further substantiated by the 50% decline in the CH, concentration (1450 cm-‘). Given the dual origin of the CH, signal its decline in magnitude. relative to the C-0-C concentration, suggests the CH, loss may be attributed almost entirely to the loss of this species contained in the ester group. This implies that the other half of the CH, species, comprising the methyl side group, is not liberated during irradiation. Preferential ester abstraction is consistent with gas chromatography results obtained by Todd [3] and IR data reported by Taylor [7]. The relative “instability” of the ester group may be partially attributed to its greater interaction cross-section for scission resulting from its larger size and “radius of conformation”, as compared with the methyl side group [1.2]. The 45% decrease in intensity associated with the CC skeletal bond (fig 2c: 1125 cm-‘) amounts to half of the 86% attenuation in the ester species’ intensity over the same dose range. Since the PMMA structure accommodates two skeletal C-C bonds for every ester
cqour
dewlopmenr
group, this suggests that the remocd ~jone ester group I.Y uccompunied by one backbone chain .sci.won. Destabilization and subsequent fracture of a main chain molecule resulting from side group scission is commensurate with a Norrish Type I reaction [S]. The Norrish reaction is dominant in polymers containing C=O carboxyl species. Evidence connecting this Norrish reaction to physical erosion can be obtained by correlating figs. 1 and 2(b). According to fig. 1, the greatest erosion rate occurs at doses below 8 (~A/cm’)min. The Raman data of fig. 2(c) illustrates that the greatest ester group loss also occurs in this dose range. Previously reported dissolution studies on PMMA indicate that irradiation through this dose range renders the greatest increase in polymer solubility corresponding to a reduction in fractured molecular weight (9,101. This substantiates the connection between depolymerization and vapour development. The results in fig. 2(d) show that C=C sites are generated during irradiation. Given its shape and spectral position, this peak arises from a bond contained within a long skeletal chain of high molecular weight. We have previously shown that a decrease in PMMA dissolution rate and a corresponding increase in the molecular weight occurs at high doses where the C=C population density is largest [9]. Evolution of the C=C species is thought to be initiated in the following manner: As illustrated in figs. 3(i) and (ii). main chain scission proceeds by ester group abstraction. resulting
c=o
c=o
k=O
;-CH3
d-CH3
;-CH3
CH
CH
I 3
(iii)
-CH?$.
6-CH;
_t
.CH$ y=o 0-CH,
CH,
(iv)
-h-CH&
CH,
CH, +
CH,
.&C”&-
A=0
t=o
&-CH3
0-CH,
CH
CH 13 -C-CH,-c=C-CH&
I 3
CH,
CH
I
Lo t=O 0-CH,
A-CH, \
Fig. 3. Illustrates and recombination incident radiation.
end-
link
the mechanism for chain scission (i) to (iii) (iv) activated in PMMA in response to Shows recombination is scission dependent.
E. M. Lehockey et al. / PMMA vapour development
from secondary electron yield produced by primary photon interactions. Scission sites activated near the surface generate free radicals from the dissociation of ester groups and skeletal chains. Close proximity to the surface permits the diffusion of free radicals from the polymer without substantial recombination along the short diffusion length. This results in material loss. The decrease in molecular weight accompanying scission increases the number density of dangling end group bonds (see fig. 3(iii)), thus increasing the probability that two end groups will be in close enough proximity to bond (fig. 3(iv)). Valency restrictions dictate that a double bond carbon would form the end-link [l]. Recombination of this type opposes the reduction of larger chains into mobile gaseous species required to further vapour development. A significant feature of the Raman results is that C=C end-links appear to “repair” scission activated end group sites. Hence, the population of potential recombination/cross-linking sites, and therefore the rate of cross-linking is dependent on the number of scission sites activated. This would account for the rapid initial rate of vapourization resulting from dominant scission in the low dose regime were the number of potential cross-linking sites is few. The dominance of recombination at larger doses would account for the decline in the vapour development rate observed in fig. 1 at doses in excess of 8 ()rA/cm’)min. 4.2. The physical response The decline in rate of vapourization with dose shown in Fig. 1 is consistent with data reported by Bowden for PBS polymers [ll]. Using e-beam radiation, Bowden achieved a vapour developed thickness limited of 3325 A, as the rate of increase approached zero, as compared with 4523 A obtained here for PMMA. Under electron bombardment, PBS is known to have a scission yield (G-value) of ten times that of PMMA. X-ray exposure thus appears to enhance vapour development sensitivity, when compared to electron bombardment. Fig. 1 illustrates two approaches for maximizing the vapour development yield. A comparison of curves (a) and (c) reveals that, by increasing the wavelength of the incident X-rays, increases in the vapourization sensitivity of PMMA may be affected. It is thought that this increase in vapourization yield reflects the increase in the linear absorption with increasing wavelength. Enhanced absorption increases the energy deposited within the vapour development depth, resulting in a greater number of activated scission sites; thus augmenting the free radical formation per incident photon. In the case of the Cu-K X-rays a larger portion of the incident energy is absorbed in the bulk were contribution to vapour development is reduced. This accounts for the higher initial vapourization rate, obtained with Al-K X-rays as compared to Cu-K incident radiation
361
indicated by slopes S(A1) and S(Cu) in fig. 1. This change in yield with wavelength distribution qualitatively agrees with results by Henderson for the exposure of nitrocellulose films in the visible wavelength regime
WI. Profiles (a) and (b) of fig. 1 illustrate that a significant enhancement in the vapour development yield sensitivity can also be achieved by reducing the initial molecular weight of the PMMA. The increase in vapour development with decreasing initial molecular weight differs from vapourization profiles reported by Bowden for PBS polymers [ll]. On the basis of a reduction in the vapourization yield of PBS with decreasing initial weight, Bowden proposed an unzipping depolymerization mechanism involving the spontaneous scission of the backbone chain along its length following end group abstraction. Fig. 1 suggests that “molecular unzipping” does not occur in PMMA. but rather that depolymerization occurs in a discrete fashion. If PMMA depolymerization did result from unzipping, the yield would increase with initial molecular weight, due to a higher yield per activated end group. This concurs with the Raman results showing discrete main chain scission following ester removal. Finally, attempts were made to maximize the erosion yield by combining the above techniques. Exposing low initial molecular weight PMMA (M, = 30 000) to long wavelength X-rays obtained from a carbon target produced erosion thicknesses of 1.0 pm in relatively short exposure times of 5 min. From the point of view of feasibility in lithography, this represents a substantial improvement over the few hundred angstroms of vapourization achieved by electron beam irradiation [4].
5. Conclusions (1) The interaction of X-rays with PMMA initiates concurrent chain scission and recombination. (2) Scission proceeds by preferential ester group removal leading to discrete main chain scission. Both reactions generate free radicals which can diffuse toward the surface to affect material loss. (3) During irradiation. repair of fractured chains occurs with the evolution of C=C bonds at scission end group sites. Thus the degree of recombination depends on the scission yield. (4) Substantial amounts of material loss are possible with X-ray irradiation. With low initial molecular weights and longer wavelength exposures enhanced sensitivity yielding erosion depths of 1.0 pm can be attained! (5) The vapour development rate is initially rapid but decreases with dose as a result of the polymerization effect accrued to C=C recombination. VI. POLYMERS / ORGANICS
368
E.M. Lehockqv et al. / PMMA
The authors wish to express their appreciation to the members of Surface Science Western for the use of their equipment, and our many fruitful discussions.
oapour dec)elopment
[6] [7] [8]
References [I] M. Dole, Radiation Chemistry of Macromolecules. 2 (Acedemic Press. New York. 1973). [2] M.J. Bowden, J. Polym. Sci.: Polym. Symp. 49 (1975) 221. (31 A. Todd, J. Polym. Sci. 42 (1960) 223. [4] M. Hatzakis. J. Electrochem. Sot. 116 (1969) 1033. [5] I. Hill and M. Flieshman, in: Comprehensive Treatise of
[9] [lo] [ll] [12]
Electra-chemistry, vol. 8, ed. R. White (Pergamon. London. 1981) p. 373. E.M. Lehockey and 1. Reid. Surf. Interf. Anal. 11 (1988) 302. J. Taylor and W. Bohn, Anal. Chem. 53 (1981) 1082. J. Guillet, presented at 5th Int. Conf. Radiation Effects in Insulators. Hamilton. Ontario, Canada, June 1989. E.M. Lehockey and 1. Reid, Can. J. Phys. 65 (1986) 975. J. Greenich, J. Electrochem. Sot. 121 (1977) 121. M. Bowden and L. Thompson, Polym. Eng. Sci. 14 (1974) 525. D. Henderson and J. White. Appl. Phys. Lett. 46 (1985) 901.