Report CopyRight/DMCA Form For : Spatial Distribution Of Reaction Products In Positive Tone
Spatial distribution of reaction products in positive tone chemically ampli ed resists Gerard M Schmid and Michael D Stewart Department of Chemical Engineering The University of Texas at Austin Austin Texas 78712 Vivek K Singh Intel Corporation Hillsboro Oregon 97124 C Grant Willsona
186 Schmid et al Spatial distribution of reaction products 186. oped photoresist feature The situation is analogous to the. case of representing an image with pixels many small pixels. are better at producing a smooth image than fewer large pix . els Developing a photoresist does not produce a gray scale . rather it is a binary process in which each polymer molecule. either remains in the film or dissolves into the developer . Whether or not a given polymer chain will dissolve is depen . dent upon a number of formulation and processing variables . Prior to the PEB the polymer chains that compose a. positive tone chemically amplified resist film are all ostensi . bly insoluble This insolubility results from the presence of a. number of pendant blocking groups on the polymer chains . In a typical photoresist formulation blocking 20 30 of. the repeat units on a given chain is sufficient to render that. chain insoluble on the time scale of the development process . Illumination of the photoresist through a mask generates acid. in exposed regions At the elevated temperature of the PEB . the acid catalyzes thermolysis of the blocking groups The FIG 1 Lattice representation of a resist film where every molecule is con . end effect is that in regions of high acid concentration most sidered individually This example is two dimensional and contains tens of. of the blocking groups are removed and the polymer is very cells A lattice used in simulation is three dimensional and may contain. millions of cells , soluble Similarly in regions of low acid concentration most. of the blocking groups remain and the polymer is insoluble . In the boundary region between these two extremes lies the ume of the polymer film All of the lattice occupants were. variation in polymer solubility that determines the shape of introduced in amounts that are representative of the concen . the resist feature after development trations found in the particular photoresist that was being. Line edge roughness is the consequence of the complex simulated During the simulation steps each lattice occupant. interaction between the spatial variation in composition that behaves in a manner consistent with its chemical identity In. results from exposure and PEB and the solubility function this way it is possible to use the simulation to investigate the. that describes the action of the developer on that composi effects of the various constituents of the resist on the litho . tional distribution The first step in understanding line edge graphic performance Among the many formulation variables. roughness is accurate establishment of the spatial variation in that may be examined in this manner are polymer molecular. the molecular level composition after exposure and baking weight and polydispersity PAG loading and residual casting. In this work a mesoscale model was used to calculate the solvent concentration The main topic that will be discussed. effect of acid concentration on the compositional distribution here is the role of acid concentration in the creation of ir . produced during the PEB Lattice models of photoresist films regularities in the reaction product distribution . were exposed to produce a variety of acid distributions and. A Generation of a lattice representation of a resist. a PEB simulation was used to determine the resulting distri . bution of blocked polymer sites that is instrumental in deter . mining the topography of the developed resist feature The The lattice model of the resist film is based upon funda . final resist image is also strongly dependent on the develop mental and measurable parameters that define the composi . ment process and the reader is referred to the literature for a tion of the photoresist To specify the polymer component . detailed discussion of issues affecting photoresist the average degree of polymerization and the standard devia . dissolution 18 22 tion of the distribution of degree of polymerization must be. input These data can be calculated from the molecular. weight distribution of the photoresist resin which is measur . II MESOSCALE LITHOGRAPHY MODELING, able by techniques such as size exclusion chromatography It. Our approach to lithography modeling is based upon the is also necessary to know the average fraction of the polymer. discretization of the photoresist film into cells on a three repeat units that are blocked which can be determined by. dimensional lattice Fig 1 Each lattice cell occupies the thermogravimetric analysis These three variables effectively. same volume as a monomer repeat unit of the photoresist define the photoresist polymer as it is represented on the. polymer Lattice cells are strung together to form chains and lattice The presence of other photoresist components most. pendant groups are attached to some sites on the chains to notably the PAG loading must also be quantified Tech . represent the partial blocking of the photoresist resin Some niques for measuring such resist properties have been de . lattice cells contain photoacid generators PAGs and some scribed in the literature 23 25. contain photoproducts acid plus counterion Base additives Knowledge of the chemical composition of the photore . acid quenchers residual casting solvent and other addi sist allows one to calculate the appropriate population of. tives can also be included but in this work they were omitted each component in the lattice Each component can then be. for simplicity Vacancies in the lattice represent the free vol added sequentially until the desired concentration is present. J Vac Sci Technol B Vol 20 No 1 Jan Feb 2002, 187 Schmid et al Spatial distribution of reaction products 187. in the lattice In the first step of lattice creation straight. chains are added to the lattice by stringing cells together to. the desired chain length and chain length distribution These. chains are then equilibrated via many simulated reptations. that have the effect of introducing bends into the chains and. eventually randomizing their conformation 26 In the next. step pendant groups are added to random positions on the. chains to represent blocked sites on the polymer chains Fi . nally PAG molecules are added by randomly converting lat . tice vacancies into PAG units until the required concentration. of PAG has been created ,B Mesoscale exposure simulation. Once the lattice is configured it is exposed by selectively. converting the PAGs into acids on the basis of their location FIG 2 Equilibration of the polymer chains The mean squared end to end. within the photoresist film To this end the commercial li distance R e2 and the mean squared radius of gyration R 2g are periodi . cally calculated , thography simulation package PROLITH 7 was used This. software is capable of calculating the energy deposited into. any volume element of the resist while varying the exposure. wavelength and partial coherence the numerical aperture of randomizing the conformations of the chains The mean. the imaging system conventional annular and quadrupolar squared radius of gyration and mean squared end to end. illumination schemes and arbitrary mask structures includ distance were monitored during the equilibration process . ing phase shifts among many other processing variables The lattice was considered to be at equilibrium when lattice. The result of the exposure calculation was a deposited en properties no longer changed with further equilibration steps. ergy distribution throughout the resist film which was then Fig 2 For long chains the ratio of the mean squared end . converted into the fractional conversion of PAG as a function to end distance to the mean squared radius of gyration has. of location within the film based upon knowledge of the been calculated to have a value of 6 27 As can be seen in. local energy and the measured quantum efficiency of the Fig 2 simulation results agree well with this predicted. PAG In this way a continuous distribution of energy is con value After equilibration pendant groups were added to ran . verted into a discrete distribution of acid in the mesoscale dom locations on the polymer chains to represent the blocked. simulation fraction of polymer sites The distribution of blocking frac . tions was Gaussian with a mean of 0 30 and a standard. C Mesoscale PEB simulation deviation of approximately 0 10 Fig 3 Photoresists typi . During the PEB simulation all single cell lattice occu cally consist of random copolymers so this distribution of. pants e g PAGs and acids are allowed to move via a ran blocking fractions is an accurate representation of what. dom walk through the vacant cells of the lattice During this would be found in an actual resist The last step in creating. process some acid molecules come into contact with block the film was to add PAG molecules to the lattice by randomly. ing groups with which they may chemically react When an converting a fraction of the voids into PAGs This was done. acid unit comes into contact with a blocking group it cata . lyzes the removal of the pendant group In the actual resist . the volatile cleavage products generally leave the resist film. as gases This process was modeled in the lattice simulation. by removing the blocking group from the polymer chain to. produce a temporary void in the lattice ,III SIMULATION RESULTS AND DISCUSSION. A Generation of a lattice representation of a resist. The results presented in this work were generated from a. lattice that is 144 cells in each of the three dimensions cor . responding to a volume of approximately 100 nm 3 This. lattice was filled with monodisperse chains 35 repeat units. long and equilibrated by a simulated reptation process 26 In. the reptation process an entire chain may undergo snakelike. motions if a void is present near one end of the chain This. process is repeated many times and has the ultimate effect of FIG 3 Blocking fraction distribution of the simulated photoresist polymer . JVST B Microelectronics and Nanometer Structures, 188 Schmid et al Spatial distribution of reaction products 188. such that the number ratio of polymer sites to PAG sites was. 20 1 leaving 10 of the lattice vacant to represent the avail . able free volume in the polymer film At this stage the lattice. model represents a 100 nm 100 nm 100 nm volume of,an APEX type photoresist film . B Evaluation of catalytic effect per acid, The high sensitivity provided by chemically amplified re . sists is a result of the catalytic action of the photoproducts A. single acid molecule produced during exposure can cause. many reactions that alter the solubility of the resist film . However an acid molecule can only cause reactions if reac . tive material is present nearby Each acid must therefore un . dergo a certain degree of translational motion in order to. continue to react In this respect acid transport is desirable in. FIG 4 During the PEB volumes of acid influence quickly overlap where. that it allows a single photogenerated acid to alter the solu there are high concentrations of acid In this two dimensional cartoon the. bility of a larger portion of the resist film On the other hand volume of influence has been depicted as a circle In actuality the acid. transport of acid into unexposed regions of the resist film follows a more tortuous path as it moves through the resist film . complicates control over feature dimensions 28 29 To more. fully understand how acid molecules near the nominal fea . ture edge affect the final topography we performed several more tortuous path as the acid moves through the film The. simulations of feature edges These simulations all focus on a volumes have been depicted as spheres only for clarity of. 100 nm 100 nm 100 nm resist volume that is centered illustration . around the nominal feature edge of an individual line in an To quantify this effect simulations of line edges were. array of 100 nm equal lines and spaces performed for the exposure conditions shown in Fig 5 . The influence of an individual acid molecule in determin Three sinusoids were used to simulate exposure conditions. ing the shape of the resist feature is related to the number of with varying image slopes PAG in the lattice was converted. reactions that it catalyzes during the PEB and to the location to acid in accordance with each particular exposure condi . of those reactions The number of reactions that are catalyzed tion A PEB simulation was then performed on each lattice. during the PEB has been taken as a single fixed quantity or for a total of 5 103 simulation steps during which time the. constant in several analyses but this is an oversimplification number of reactions catalyzed per acid molecule was. of a complex mechanism As it happens the location of an tracked These data have been plotted as a function of the. acid molecule strongly determines the number of reactions final acid location in Fig 6 It is clear that the number of. that it can catalyze Consider an acid molecule that is gener reactions that each acid catalyzes is a strong function of lo . ated in the center of the exposed region of the resist film cation for the different exposure conditions In locations. Due to the high exposure in this region a large number of where there is a high concentration of acid blocking groups. PAG molecules will be converted into acids and the depro are the limiting reagent These reactive sites are quickly de . tection reaction will therefore proceed very quickly in this pleted and the resulting number of reactions per acid is low. volume All of the reactive material blocked sites in this. volume will be rapidly depleted whereupon the reactions. will cease Now consider an acid molecule that is generated. in the tail of the aerial image Very few acid molecules are. present in regions of such low exposure and so the reaction. proceeds more slowly It is therefore possible for an indi . vidual acid in a region of low acid concentration to continue. to cause deblocking reactions long after all blocked sites are. gone in the regions of high acid concentration , If the volume of resist that each acid molecule contacts is. assumed to grow spherically then it is easy to conceptualize. that these volumes of influence will overlap quickly in re . gions of high acid concentration as shown in Fig 4 In re . gions where acid molecules are sparsely distributed the vol . ume of influence must be considerably larger before another. reacted region is encountered There is less volume of resist. available per acid in regions of high acid concentration It. should be noted that the volume of acid influence does not FIG 5 Exposure conditions used for several line edge simulations The. actually grow spherically during the PEB but rather takes a three cases shown represent a range of different image contrasts IC . J Vac Sci Technol B Vol 20 No 1 Jan Feb 2002, 189 Schmid et al Spatial distribution of reaction products 189. FIG 6 Mean number of reactions per acid as a function of location for the FIG 7 Spatial irregularity in reaction product distribution standard devia . exposure conditions Acids that are present in regions of low acid concen tion of blocking fraction For equal degrees of deprotection higher acid. tration may catalyze many more reactions than acids in regions of high acid concentrations produce smoother distributions smaller standard deviation . concentration , on each lattice and the three dimensional reaction product. 10 In regions where there is a very low concentration of distribution was periodically recorded To quantify the spa . acid the acids are the limiting reagent As a result the num tial irregularity in each reaction product distribution the. ber of reactions per acid is much higher in these regions as blocking fraction of the polymer chain that occupies each. shown in Fig 6 It is therefore misleading to assign a single lattice location was calculated The mean and standard de . average value to the catalytic chain length in this sort of viation of the blocking fraction in each lattice have been. reaction analysis plotted in Fig 7 for several different acid concentrations The. It should be noted that the location of the feature edge blocking fraction of each lattice was identical prior to the. cannot be inferred from the information given in Fig 4 The PEB During the PEB each lattice progressed from higher. location of the feature edge depends on a convolution of the blocking to lower blocking right to left in Fig 7 For equal. distribution of blocked with a solubility function not on the degrees of deprotection higher acid concentrations produced. number of reactions per acid This discussion is focused only smoother distributions of blocking fraction These smooth. on quantification of the irregularity in the reaction product distributions can be expected to produce smoother features. distribution The interaction of this distribution with the de upon development if the development process is halted at the. velopment process is the subject of ongoing research studies region of higher acid concentration The effect can be visu . alized by simply plotting a two dimensional slice of the. C Calculation of spatial distribution of reaction three dimensional reaction products distribution This has. products been done for two blocking fraction distributions in Fig 8 . Complete overlap of the individual regions of acid influ where dark regions have a lower blocking fraction than light. ence during the PEB produces a volume of the resist that is regions Both of these distributions have an average blocking. fully deprotected It is the regions where the overlap is not fraction of 0 156 but the standard deviations vary by almost. complete that determines the final resist topography The fea a factor of 2 . ture edge is defined at the interface between soluble and. insoluble regions where the resin protection level has been. IV CONCLUSIONS, decreased to the degree that enables dissolution This value. is dependent upon both the chemical composition of the re Simulations have shown that there is significant variation. sist and the development conditions and is beyond the scope in the number of reactions that each acid molecule catalyzes. of this work However due to the nature of the deprotection during the postexposure bake Acids that are in regions of. reaction there are several ways to achieve the same level of high acid concentration generally participate in far fewer re . deprotection The same average level of deprotection can be actions than acids that are in regions of low acid concentra . produced by a high concentration of acid and a short PEB or tion It is therefore misleading to refer to a single catalytic. with fewer acid molecules and a long PEB The mesoscale chain length for a particular photoresist The feature edge. structure of the reaction product distribution is very different occurs in some region of intermediate acid concentration but. for the two cases the exact location of the feature edge is ultimately deter . In a series of simulations the PAG in the lattice was uni mined by both material properties and development condi . formly converted to several different levels of bulk acid con tions It is desirable to have the feature edge occur at a rela . centration that correspond to the regions of low acid concen tively high acid concentration because the simulations. tration depicted in Fig 5 A PEB simulation was performed suggest that this leads to a smoother composition gradient . JVST B Microelectronics and Nanometer Structures, 190 Schmid et al Spatial distribution of reaction products 190. Presented at the 45th International Conference on Electron Ion and Photon. Beam Technology and nanofabrication Washington DC 29 May 1 June. T Yoshimura H Shiraishi J Yamamoto and S Okazaki Jpn J Appl . Phys Part 1 32 6065 1993 , M I Sanchez W D Hinsberg F A Houle J A Hoffnagle H Ito and. C Nguyen Proc SPIE 3678 160 1999 , Q Lin R Sooriyakumaran and W Huang Proc SPIE 3999 230 2000 . T Yamaguchi H Namatsu M Nagase K Kurihara and Y Kawai Proc . SPIE 3678 617 1999 , T Ushirogouchi K Asakawa M Nakase and A Hongu Proc SPIE. 2438 609 1995 , E Shiobara D Kawamura K Matsunaga T Koike S Mimotogi T . Azuma and Y Onishi Proc SPIE 3333 313 1998 , J Nakamura K Deguchi and H Ban J Photopolym Sci Technol 11 . 571 1998 , S Masuda X Ma G Noya and G Pawlowski Proc SPIE 3999 252. 2000 , T Azuma K Chiba M Imabeppu D Kawamura and Y Onishi Proc . SPIE 3999 264 2000 , D He H Solak W Li and F Cerrina J Vac Sci Technol B 17 3379. 1999 , D He J Vac Sci Technol B 16 3748 1998 , S C Palmateer S G Cann J E Curtin S P Doran L M Eriksen A . R Forte R R Kunz T M Lyszczarz M B Stern and C Nelson Proc . SPIE 3333 634 1998 , L W Flanagin V K Singh and C G Willson J Vac Sci Technol B. 17 1371 1999 , S Yasin A Mumtaz D G Hasko F Carecenac and H Ahmed Micro . electron Eng 53 471 2000 , G Patsis N Glezos and I Raptis J Vac Sci Technol B 17 3367. 1999 , G Patsis A Tserepi I Raptis N Glezos E Gogolides and E S Vala . montes J Vac Sci Technol B 18 3292 2000 , L Ocola P A Orphanos W Y Li W Waskiewicz A E Novembre and. M Sato J Vac Sci Technol B 18 3435 2000 , P C Tsiartas L W Flanagin C L Henderson W D Hinsberg I C . Sanchez R T Bonnecaze and C G Willson Macromolecules 30 4656. 1997 , L W Flanagin C L McAdams P C Tsiartas C L Henderson W D . Hinsberg and C G Willson Proc SPIE 3333 268 1998 . L W Flanagin V K Singh and C G Willson J Polym Sci Part B . Polym Phys 37 2103 1999 , L W Flanagin C L McAdams W D Hinsberg I C Sanchez and C . G Willson Macromolecules 32 5337 1999 , A Reiser Z Yan Y K Han and M S Kim J Vac Sci Technol B 17 . FIG 8 Slices in the blocking fraction distribution of two lattices Both. 1288 2000 , lattices have the same average degree of deprotection but a was produced 23. C R Szmanda R Kavanagh J Bohland J Cameron P Trefonas and R . with a longer PEB and fewer acid molecules than b For equal degrees of. Blacksmith Proc SPIE 3678 857 1999 , deprotection higher acid concentrations produce smoother blocking fraction 24. A B Gardiner A Qin C L Henderson S Pancholi W J Koros C G . distributions The fractional conversion of PAG used to produce these block . Willson R R Dammel C A Mack and W D Hinsberg Proc SPIE. ing fraction distributions was 0 025 for a and 0 20 for b resulting in. 3049 850 1997 , standard deviations of 1 7 10 3 and 9 8 10 4 respectively 25. J F Cameron J Mori T M Zydowsky D Kang R Sinta M King J . Scaiano G Pohlers S Virdee and T Connolly Proc SPIE 3333 680. 1998 ,ACKNOWLEDGMENTS 26, G M Schmid V K Singh L W Flanagin M D Stewart S D Burns . and C G Willson Proc SPIE 3999 675 2000 , The authors would like to thank Matthew Stone and Dr 27. K Kremer and G S Crest in Monte Carlo and Molecular Dynamic. Isaac Sanchez at the University of Texas at Austin for their Simulations in Polymer Science edited by K Binder Oxford University. helpful assistance and advice This work has been supported Press New York 1995 Vol 1 p 199 . S V Postnikov M D Stewart H V Tran M A Nierode D R Me . by the SRC and DARPA G M S gratefully acknowledges deiros T Cao J Byers S E Webber and C G Willson J Vac Sci . the Eastman Kodak Corporation for support in the form of a Technol B 17 3335 1999 . graduate research fellowship 29, J Sturtevant S Holmes and P Rabidoux Proc SPIE 1672 114 1992 .