Mastering the resist-leaching and aqueous-contact-angle challenges
Second in a series.
A unique aspect of 193nm immersion (193i) lithography is the water between the front lens and the wafer, which forms a meniscus. This moves with the exposure head across the wafer. The resist stack on the wafer is exposed under the water in a dynamic way with the step and scan process. As 193i lithography moves into manufacturing, two new technical challenges must be mastered to fully realize its benefits without degrading performance: the resist stack used in this immersion process must offer both low leaching behavior and a hydrophobic surface.
The first of these requirements is caused by the fact that the photo-acid generator (PAG), quencher, or other small molecular components in the resist leach into the water, they contaminate it and may also degrade the resist's performance. The contaminated water can further contaminate the lens and the wafer stage of the scanner. Thus, a resist stack with minimal leaching is required for the immersion exposure.
The second requirement is related to the fact that the water meniscus moves fast: it follows the exposure head as it moves across the wafer at a peak speed of ∼500mm/s. As a result, water droplets may be left behind forming circular defects. A hydrophobic surface reduces the number of water droplets and allows the meniscus to move easily. Consequently, static and dynamic contact-angle measurements are being introduced to evaluate the surface hydrophobicity of resist stacks.
Leaching test
Leaching tests measure the resist components that leach into the water. In general, de-ionized (DI) water is dispensed onto a resist stack, forming a puddle on the surface. After some time, the water sample is extracted and sent to analysis. Various methods of immersing the resist film and extracting the water sample have been developed,1 such as the water-extraction-and-analysis (WEXA) apparatus and technique described by William Hinsberg and coworkers.2
The water sample is analyzed for resist components. Because different chemical analytical methods have different sensitivities towards certain components, the the best analytic method must be chosen for each compound. For example, sulfonates of PAG can be detected separately using liquid chromatography mass spectroscopy with a detection limit of 0.2ng/mL or parts per billion (ppb). If one includes the water contact area and immersion retention time, the dynamic leaching rate (presented in units of ng/cm2/s) can be calculated from the detected concentration value.
Leaching is a very non-linear process.3 The resist components, mainly PAG components, leach quickly at the beginning of water contact. Within a few seconds, however, the leaching process becomes saturated. The concentration of PAG (C) in water versus the water contact time (t) can be approximately described by an exponential relation:4
where C∞ is the saturated PAG concentration and β is the time constant. At initial time, t=0, the leaching rate is dC/dt|t=0 = C∞·β. This is called the ‘dynamic leaching rate’ and describes how fast the component leaches from the solid on contact with water. If the 193i scanner is designed so that the immersion water flow is continuous, then some of the leached components are flushed away by the water flow. The dynamic leaching rate therefore describes the concentration of resist components in immersion water near the lens better than the saturate leaching value. The amount of leaching an immersion tool can tolerate depends on its exposure head design and the water flow. Vendors of 193i scanners have suggested the leaching specifications listed in Table 1.
The leaching test discussed above is performed without any exposure. Exposed to 193nm beam, it is reasonable to expect that the leaching level would increase. Several resist and topcoat samples with different leaching performances were tested with exposure. Topcoats provide a temporary solution for blocking leaching. To test this, the resists or resist topcoat stacks were immersed in DI water and exposed to different exposure doses from E0 to 2E0 (where E0 is the dose required to remove the resist in an open-frame exposure). After exposure, the water samples were collected and analyzed for the PAG components of sulfonate C1 and sulfonate C4. Figure 1 shows the test results. Topcoat TC1 is such an effective leaching barrier that no leaching was detected even at the exposure dose of 2E0. Without the topcoat as a barrier, more C1 and C4 leach into water as the dose increases from 0 to 2E0. However, the increased value is less than 15%. The irradiation at 193nm does not dramatically increase leaching. We were therefore able to use the leaching test results obtained without exposure to judge whether the resist stack could be exposed.
Resist process with pre-rinse
The development of manufacturing-worthy 193i resists is still in progress. The requirement that the material offer both low leaching and high-resolution behavior makes resist development very difficult. For example, the low-activation-energy resist is baked at low post-exposure bake (PEB) temperatures and has a small acid-diffusion length, which is promising for high resolution. However, the material tends to have a high leaching level.
Rinsing the resist film with DI water before exposure has been suggested as a way to wash away a significant proportion of the leaching components. Compared to non-rinsed wafers, those that have been pre-rinsed have a leaching level of only about 12% and a time constant (β) that about a factor of about 2 lower.4 Until low-leaching and high-performance 193i resists become available, ‘high-leaching’ 193nm resist plus a DI water rinse prior to exposure could be an option. Figure 2 shows the process flow.
One major concern with the pre-rinse process, however, is the extent to which it changes the resist sensitivity. Open frame exposures with a dose meander (where the dose increases in fixed steps) were carried out on wafers with different pre-rinse times: no rinse, pre-rinse for 10s, and pre-rinse for 30s. The maximal pre-rinse time setting of 30s was decided according to the assumption that leaching happens within the first 30s of contact with water.3 After development, the thickness of resist residual in the dose meander area was measured. The resist thickness versus dose—the contrast curve—is obtained, as shown in Figure 3. There is almost no difference between the contrast curves measured at no rinse and the 10s and 30s pre-rinse cases. This situation was observed even for a ‘dry’ resist sample that has a leaching level of about 24ppb total sulfonates.
For the high-leaching-resist samples, we know—from the leaching test results—that the PAG leaches out as soon as it contacts water. Why does the contrast curve not show any change when introducing a pre-rinse? We believe this has to do with the exposure head and water flow design. The water is confined between the lens and wafer: the water flow continues through the exposure head both during and between exposures. When the head moves to the next die, the die is flushed by the water before the exposure starts. Therefore, there is an ‘intrinsic flush’ going on right before exposure. According to simulation results by Ivan Pollentier et al.,7 the pre-intrinsic-flush time is about 1–2s. Either way, the wafer surface receives the intrinsic flush right before exposure, which can smear the footprint of the pre-rinse that we inserted into the process flow. Therefore, the resist sensitivity change caused by the pre-rinse is not observed in 193i lithography.
Water contact angle of resist stacks
The angle at which the water contacts the resist surface is another important parameter for understanding the resist or topcoat performances in 193i exposure. A high contact angle means that the surface is hydrophobic and water moves easily on it, as seen in Figure 4(a). The opposite situation happens with a surface with a small contact angle. In the 193i scanner, the water meniscus moves with the exposure head, as sketched in Figure 4(b). The static contact angle is insufficient to describe both the shape of the water meniscus and the water movement on resist surface: dynamic contact angles must therefore be introduced. We considered both the advancing contact angle, θa, and the receding contact angle, θr, in Figure 4(b). This describes the water contact angles in the front and rear of the meniscus at the wafer scan speed of v.
The dynamics of the water meniscus have been investigated in a detail by a joint work of University of Wisconsin and Sematech.8,9 The dynamic contact angles are related to the height of the water gap (h), velocity of the wafer stage (v), viscosity and surface tension of the water, and hydrophobicity of the resist surface. Figure 5 shows sketched results of the detailed calculations from the group.8,9 As the scanning speed increases, the receding contact angle decreases and the advancing contact angle increases. There is a hysteresis region at v=0, which is caused by surface tension and viscosity of the water. The static advancing and receding contact angles (θa|v=0, θr|v=0) are the dynamic contact angles at the critical points, as labeled in Figure 5.
The receding-contact-angle value has a strong correlation to the number of water droplets left behind. As the receding contact angle gets smaller, the head sheds water droplets more easily. When the receding contact angle (θr) approaches 0°, the exposure head leaves a thin film of water behind. (While this is an important technique for painting, it isn't the desired effect here.) To reduce or eliminate these water droplets, we need a large receding contact angle.
The tilting wafer method was developed to evaluate hydrophobicity of resist surfaces as well as measuring the dynamic contact angles. Figure 6 shows a diagram of the setup. A drop of water, in a volume of ∼50μl, is dispensed on a wafer that is coated with resist or topcoat. The wafer is tilted slowly. Immediately prior to the moment when the droplet slides, the contact angles in the front and rear of the droplet are measured. These angles correspond to static advancing and static receding contact angles: θa|v=0 and θr|v=0 . The wafer tilting angle is called the sliding angle.
We measured the static receding contact angles from various resists or topcoats using the tilting wafer method. The defect counts of water marks have also been measured from these resists or topcoats exposed by full-field scanners. A large amount of data about defect counts versus surface hydrophobicity data has been gathered.10,11 The experimental results suggest that the receding contact angle of 70° is needed to avoid trailing water droplets at full scanning speed. However, increasing a resist's hydrophobicity tends to cause side-effects in its performance. For example, hydrophobic resist tends to have a small dissolution rate in the standard aqueous tetramethylammonium hydroxide (TMAH) developer, which is over ∼99% water. Resist formulation has to be optimized to give the best overall lithographic performance.
Resist leaching and resist surface hydrophobicity have been recognized as two new technical challenges to 193i lithography. The leaching mechanism is understood to some extent, various leaching evaluation methods have been developed, and the performance of numerous resists has been quantified. Development of resists with low leaching characteristics and superior lithographic performance is in progress by several resist suppliers. Until low leaching resists are developed, topcoats provide a temporary solution to block leaching. Rinsing the resist film with DI water before exposure has been suggested as an alternative way of addressing this issue.
The amount of water left on the wafer after the exposure clearly depends on the degree of hydrophobicity of the resist surface. Static and dynamic contact angles are measured to describe the surface hydrophobicity and the shape of water meniscus. The receding contact angle has been shown to have a strong correlation to the quantity of water droplets left behind. The experimental results suggest that the receding contact angle >70° is needed to avoid trailing water droplets at the current full scanning speed. Although both of these technical issues are new to photolithography, substantial progress has been made in measurement techniques and process optimization.
Yayi Wei is a staff engineer of Qimonda North America Corp (formerly the Memory Product Division of Infineon Technologies North America Corp.) and a member of SPIE working on advanced lithographic process development and photoresist evaluation.
