Reticle Purging Approaches by Nitrogen with Enhanced Efficiency

The technology roadmap toward smaller structures and thinner layers in semiconductor manufacturing directs attention more and more toward yield-affecting influences from the air quality of manufacturing environment such as water vapor, O2, CO2 etc. to absorb high-energy radiation and formation of haze form on reticle surfaces during microlithography processing. A useful method for reducing these yield-affecting influences is purging the reticle surface with nitrogen gas. The main issue is the difficulty in performing purge process in the space between the reticle and pellicle, which has a rather fragile geometry. Our study strives to find the optimized parameters by using computational fluid dynamic (CFD) simulation plus the corresponding inspection results of reticle exposed by 193 nm beam in the fab. Results show that Purging time is sensitive to both the number of purging holes and the purging flow rate. The required purging time can be reduced from 77 seconds to approximately 34 seconds by increasing the purging flow rate from 0.094 L/m to 0.376 L/m (corresponding to purging velocity of 1.0 m/s to 4.0 m/s). However, concerning the breaking of pellicle due to high velocity, the purging velocity was limited to 2.0 m/s (corresponding to its flow rate of 0.188 L/min). By doubling the number of purge/vent holes, with the same flow rate, the required purge time can be even reduced to 33 seconds.


INTRODUCTION
The technology roadmap toward smaller structures and thinner layers in semiconductor manufacturing directs attention more and more toward yield-affecting influences from the air quality of manufacturing environment such as water vapor, O 2 , H 2 O and CO 2 , to absorb high-energy radiation and formation of haze form on reticle surfaces during microlithography processing. The photomask is e x tr e me l y s e n s i t i v e l y t o A ir b o r n e Mo l e c u la r Contaminations (AMCs), especially in the 193-nm or 248-nm optical lithography process. Control on AMCs has been the key issue of yield management (Chien et al., 2007;Li et al., 2007). The most noticeable AMCs are water vapor and oxygen. A useful method for reducing AMCs is purging of the photomask box with nitrogen gas (Hu and Tsao 2006;Hu et al., 2007). In the photomask box, the aim of pellicles is to prevent AMCs from falling on the pattern. The exposure process of 193-nm technology can be accomplished in air ( Hanet et al., 2005;Kim et al., 2006;Gordon et al., 2007). Therefore, purging nitrogen gas to the pellicle space for contamination control is essential to avoid image degradation and to accomplish the exposure. The main issue is the difficulty in performing purification in the space between the reticle and pellicle, which has a rather fragile geometry (Bunner et al., 1983;Cotteet al., 2002;Abdo et al., 2003;Meixneret al., 2004). The current technology for fabricating pellicle frames is based on anodized aluminum. While there are general specifications on dimensions and surfaces, lots of attention has been directed to detailed requirements mainly because that the demands on the pellicle frame are not rigorous (Johnstone et al., 2003). However, as the optical lithography wavelength grows shorter, issues previously neglected may turn out to be major shortcomings (McClay and McIntyre et al., 1999). Exposure tool manufacturers are using a nitrogen purification system to decontaminate the optical lithography in systems (Okada et al., 1998;Liberman et al., 1999;Cottea et al., 2004). For example, the broadly adopted approach of purging the pellicle space through a pellicle film is now considered to be both difficult and problematic in leaving open the possibility of pellicle film rupture due to pressure difference across the pellicle film. Moreover, this approach can no longer accomplish sufficient purging to meet the growingly sophisticated needs. An improved approach that has been discussed in the industry is to perforate a number of small purging holes (inlets) and vent holes (outlets) in the pellicle frame, as shown in Fig. 1. While the frame dimensions vary with the manufacturers and types of pellicles. The broadly adopted dimension of the holes on the four sides is 1.0 mm in diameter. A transparent pellicle is placed on the surface of a reticle to protect its surface from contamination. During certain conditions, crystals formation of haze forming on reticle surfaces, the sulfate ion (SO 4 2-) left on the mask surface after SPM (scanning probe microscopes) step is known as the most important source resulting in the formation of haze defects. And ammonium sulfate has been known to be mostly responsible for haze defects formation on the mask surface. Recent investigation reveals that the other chemicals such as hydrocarbons, Na, F, Mg, K, Cl or Al are also the results of haze defects. A major problem in the integrated circuit (IC) industry is that the manufacturing process may experience malfunction due to haze contamination on the mask once the reticle has been exposed to ArF radiation. In addition to the previously mentioned shortcomings, the traditional way of purging works by supplying N 2 into reticle SMIF-Pod to help clean air slowly diffuse into the space between the reticle and the pellicle. However, the purging time is in need of improvement. The new approach proposed in this study seeks to avoid these issues by developing a prototype purge station to supply the nitrogen into the pellicle frame directly.
Our study further strives to find the optimized parameters by using computational fluid dynamic (CFD) simulation. The purging process should be performed right before the reticle is going to be exposed and repeated after the reticle is exposed so as to effectively reduce the generation of haze. Ultimately, this study aims to enhance the performance of purging process by identifying the optimal parameters such as number and position of purging holes and purging flow rate using computational fluid dynamic (CFD) simulation.

Model Equations
The CFD study was performed using a commercial FLUENT software (version 6.3 (2007)), which has been successfully applied to solve various types of turbulent flows. The physical dimension for a typical purge condition (with 4 vent holes) is shown in Fig. 1.
The mixing and transport of various species were modeled by solving conservation equations, which describe convection, diffusion, and reaction sources for each component species. When solving conservation equations for species. The conservation equation takes the following general form: Where Y i is solution of a convection-diffusion equation for the ith species and is fluid density. S i is the rate of creation by addition from the dispersed phase plus any user-defined sources. An equation of this form will be solved for N-1 species where N is the total number of fluid species present in the system. Since the mass fraction of the species must sum to unity, the Nth mass fraction is determined as one minus the sum of the N-1 solved mass fractions. In this study, we solve a species mixing problem without reactions. The species transport equation for the k' th species is in the following general form: Where N and S ck are the number of species and source term supplied for equations, and the species is the effective exchange coefficient for the dependent variable. The differential equation given by Eq. (2) is integrated with respect to momentum using a second-order upwind scheme. The mass diffusion was given in the following form: where Sct is the turbulent Schmidt number ( t t D , where t is the turbulent viscosity and Dt is the turbulent diffusivity) is assumed to be 0.7. Three species i.e. nitrogen, oxygen and air were considered. The purging experiment is a complex transient process, due to the tiny thickness (about ~1 m ) of the pellicle, and the relatively high diffusion coefficients of gases in the rubbery pellicle material, transient kinetics associated with establishing concentration within the pellicle film can be neglected to a first approximation.

Boundary Conditions and Grid Independence Test
Prescribed value and Neumann boundary condition were applied to the inlet and outlet, respectively. No-flux boundary condition and standard wall function were imposed in all boundary walls. These are conventional boundary conditions. The pellicle film was assumed to be a porous media that uses an addition of momentum source. The source term is composed of two parts: a viscous loss term, and an inertial loss term.
where Si is the source term for the ith (x, y, or z) momentum equation, |v| is the magnitude of the velocity, is the permeability, and C 2 is the inertial resistance factor. Table 1 shows the boundary conditions in this study. Grid testing for both space and time domain were performed. The final grid number and time step adopted were 199,474 and 1 second, respectively. Maximum and minimum cell sizes were 5.79 × 10 -12 m 3 and 1.57 × 10 -9 m 3 , respectively.

Cases Description
The reticle-purging process must be accomplished as quickly as possible without breaking the pellicle. Major factors affecting reticle purging are all related to boundary conditions, including the purging gases, the purging velocity (flow rate), the number of purging/vent holes. Table 2 tabulates the detail of the three simulation cases. The purge flow rate rages from 0.094 L/min to 0.376 L/min, corresponding to its purge velocity of 1.0 m/s to 4.0 m/s in case 1. In case 1, inlets were located in the long sides while outlets were in short sides. Conversely, inlets were located in the shorter sides while outlets were in long sides in case 2. The number of purge/vent holes were doubled i.e. four purge holes and four vent holes in case 3.
The purge process was regarded to complete when two factors were arrived, i.e. 500 ppm of oxygen content and 1% of relative humidity.

Effect of Purge Gas
Two different kind of purge gases i.e. nitrogen and compressed dry air was investigated on their influence on the removal of water vapor content. It was found that using the compressed dry air results slightly difference in purge efficiency from using nitrogen gas (see Fig. 2). This is because there is only minor difference in the diffusivity between the compressed dry air and nitrogen gas. However, it should be noted that if oxygen is the AMCs to be controlled, the compressed dry air can not be a purge gas.

Effect of Vent Hole (Outlet)
Figs. 3(a) and 3(b) show the nitrogen distribution on the reticle surface with one vent hole and with two vent holes after 30 seconds with the traditional way of purging works by supplying nitrogen into reticle SMIF-Pod to diffuse the nitrogen into the space between the reticle and the pellicle. Both figures show that the overall gas diffusion on the edge of pellicle frame is less than the one in the center and this might result in the accumulation of contaminant on the edge and cause the formation of haze. Fig. 3(b) exhibits a more uniform distribution and a higher concentration of nitrogen gas, which relates to a lower risk of defect. This was confirmed by inspection results of reticle exposed by 193 nm laser beam in the fab, as shown in Figs. 4(a) and 4(b). In Fig. 4(a), with one vent hole, significant haze formations (indicated by red triangle symbols) were detected on the edge of pellicle frame than those in Fig.  4(b). In the fab, the purging process was operating inside the chamber, and the pressure variation in the chamber was in range of 40kPa-80kPa.   Table 2. The three simulation cases (a purging hole is an inlet and a vent hole is an outlet).

Cases
Arrangement of purging holes and vent holes Purge (

Effect of the Arrangement of Purge Holes (Inlets) and Vent Holes (Outlets)
Figs. 4(a), 4(b) and 4(c) show the flow patterns on the reticle surface in case 1, case 2 and case 3, respectively. In cases 1 and 2, two large recirculation zones were noticed in the inlet region. The recirculation zones in case 1 occupy a larger area than those in case 2, causing an increase in the purging time. The other concern is the vertical velocity on the retical surface. A high vertical velocity means a high risk of breaking of the pellicle due to the inlet purge gas. In case 1, the magnitude of vertical velocities were 0.09 m/s, 0.13 m/s, 0.17 m/s and 0.23 m/s for inlet velocities of 1.0 m/s, 2.0 m/s, 3.0 m/s and 4.0 m/s, respectively. Concerning the breaking of pellicle due to high vertical velocity on the pellicle surface, the purge velocity was limited to 2.0 m/s (corresponding to its purging flow rate of 0.188 L/mim). The number of purge/vent holes were doubled i.e. four purge holes and four vent holes in case 3. Fig. 5(c) shows small recirculation areas, eliminating the residual air in the corner area of the pellicle, which is the most difficult place that nitrogen can reach. The depletion time history of oxygen concentration at the outlet in case 3 was also lower that in cases 1 and 2. Fig. 6 presents the predicted volume averaged concentration of oxygen concentration vs. elapsed time in all cases. The oxygen concentration decreases dramatically in the first 20 seconds for all cases. It is observed that in (a) t = 30 s (a) t = 30 s purge velocity from 1 m/s to 4 m/s. When purge velocity was 2 m/s, the purging efficiency of case 2 was comparable when to the case 1 with purge velocity of 3.0 m/s. As the number of purge holes increases in case 3, the resistance of the purging flow injected into the pellicle volume decreases for a given time period. Consequently, a reduction in the total purging time was achieved. For case case 1 the required purge time can be reduced from 77 seconds to approximately 34 seconds by increasing the 3, when purge velocity was 0.5 m/s (purging flow rate of 0.094 L/s), the required purge time was 68 seconds. For the cases 1 and 2 with the same purge flow rate (purging velocity of 1.0 m/s), the required purge time were 80 seconds and 92 seconds, respectively. The required purge time can be even reduced to 33 seconds with purging velocity of 1.0 m/s (purging flow rate of 0.188 L/s) in case 3.

CONCLUSION
Based on the results and discussion, the following conclusions are drawn: 1. No significant difference on purge efficiency for oxygen removal by using either nitrogen or compressed dry air. However, for removal of oxygen compressed dry air should not be used. 2. Locating the purging holes in the short sides and vent holes in the short sides of the frame (case 2) is a better arrangement in terms of purging efficiency than those locating the purging holes in the long sides and vent holes in the short sides of the frame (case 1). 3. Numerical study and photo-inspection in the fab confirm that, with the same purging flow rate, purges with two purging holes can create a more uniform nitrogen purge than that with one purging hole, especially in the edge of pellicle frame where haze formation often happens.  to its purging velocity of 1.0 m/s to 4.0 m/s). However, concerning the breaking of pellicle due to high velocity, the purging velocity was limited to 2.0 m/s (corresponding to its flow rate of 0.188 L/min). By doubling the number of purge/vent holes, with the same flow rate, the required purge time can be even reduced to 33 seconds. 5. The future works of this study include using diffuser as a purging inlet to improve the purging efficiency and taking into account the water vapor absorption/desorption property of the frame material, purging time can be reduced from 77 seconds to which has significant influence on water vapor content in the final purge state (Hu and Tsao, 2006;Hu et al., 2007).