Phase Errors in PSMs at the 90 nm Node

Embedded attenuated phase-shift masks (EAPSMs), also referred to as half-tone masks, provide larger process windows over binary masks in a given photolithography system. Today, most wafer fabs use 193 nm EAPSM technology to pattern the most critical layers for 90 nm node logic and 110 nm node DRAM devices.^1,2 When an EAPSM is used, the associated phase angle and absorber transmission will impact the effective wafer process window. Giang and Gang performed a preliminary study on the phase-angle effect of 248 nm EAPSM on process window during etch process development.³ Because the 90 nm node is a relatively new technology, it is important to understand the effects of phase angle and transmission of a 193 nm EAPSM on process latitude, and establish practical control specifications for phase angle and transmission of EAPSMs for device production.

The phase angle and transmission of an EAPSM are determined by MoSi absorber thickness (T_m), quartz overetch depth (T_q), exposure wavelength (λ) and the absorption coefficient of the absorber (α). Figure 1 shows their relationship, where n_q and n_m are the refractive index of quartz and MoSi, respectively. In a typical EAPSM manufacturing process, the quartz overetch depth can be varied during the MoSi etch process, and the MoSi absorber thickness can be reduced by the aggressive wet clean processes used in first-time fabrication or subsequent mask repellicle and clean process.

As the exposure wavelength reduces from KrF (248 nm) to ArF (193 nm), the same amount of phase error tolerance will result in a smaller amount of allowed thickness variation. This is an important issue to keep in mind in examining the effect of phase angle and transmission of a 193 nm EAPSM to wafer process window, and its control specification. As a result, we can reduce the unnecessary EAPSM manufacturing yield loss caused by an overly tight control specification of phase angle and transmission.

1. Quartz overetch depth can be varied during the MoSi etch process, and the absorber thickness can be reduced by altering the wet cleaning process. Shown is the relationship between these thicknesses, transmission and phase angle.

To conduct this study, we first fabricated a special multiphase 193 nm EAPSM by a combination of extra quartz etch and MoSi removal steps in the mask overlay writing processes. Figure 2 shows our multiphase EAPSM design with the chromium (Cr) pattern side facing up. The mask patterns are comprised of a number of multiphase cells. Each cell pattern consists of four different areas: area A, B, C and D. The cells are 2 × 2 cm, arrayed across the mask in a 5 × 6 array. Each of the four areas (A, B, C or D) contain exactly the same patterns for lithographic process studies. While the chromium level writing processes were identical in each area, different areas in each cell received different treatment in the subsequent MoSi etch and final chromium removal and clean processes to create various phase angle and transmission combinations.

2. The specific PSM for this study is comprised of many multiphase cells, each with patterns A, B, C or D. Different areas in each cell underwent different MoSi etch and chromium removal and clean processes to create various phase angle/transmission combinations.

We used a 50 keV shaped-beam electron pattern generating system, JBX9000-MVII, to write the lithographic test patterns with typical 90 nm node design rules on the mask. The e-beam photoresist was Fuji FEP171, a positive tone chemically amplified resist. The developed resist patterns were transferred to the chromium film using a GEN-III dry etcher. Before the MoSi etch process, we measured the critical dimensions (CDs) of the chromium patterns using a KLA-8100 SEM. The measurement results indicated that the CD uniformity (of 600 nm designed contacts on the photomask) is 12 nm across the plate over 50 sampling points. The iso-to-dense CD differences are less than 8 nm on 600 nm designed contacts.

To create various phase angle and transmission samples on the photomask, we used three different overlay writing steps to create four different areas (A, B, C and D) in each cell by the combination of additional quartz etch and MoSi removing processes. During each overlay writing process, we recoated the mask with IP3600 resist. We then wrote the overlay patterns using an Alta-3700 i-line pattern generation system to expose a designed block of resist over a combination of areas.

After developing the exposed photoresist after each overlay writing process, we added an extra process to the opened areas that were not covered by the photoresist. Figure 2 lists the detailed processes of each area and its final phase angle and transmission values measured by a LaserTek MPM193 at 193 nm. Based on the measurement data, we obtained a phase angle spread of 23°, which is more than 7× larger than the current specification of 3°, and 2% of transmission spread, which is 4× larger than the current specification of 5%.

We used an ASML AT1100 193 nm scanner to conduct the wafer-level lithographic process window study on 300 mm wafers. We first applied organic BARC material to the 300 mm wafers to reduce the substrate reflections. We then applied ~4000 Å of TOK-CB01B ArF resist. The scanner exposure conditions were NA=0.65, σ=0.6. We used various focus settings and exposure dose combinations to determine the best exposure focus and process windows at different areas (phase angle and transmission values) on the fabricated multiphase EAPSM. The developed resist patterns were measured on a Hitachi S9300 CD-SEM.

Figure 3 demonstrates the wafer-level printing results of isolated contacts (nominal size of 120 nm) exposed with the multiphase EAPSM. Within top-down SEMs across different exposure focuses with different phase angle values, we highlighted the best exposure focus and exposure focus window at each phase error. One can clearly see that the best exposure focus shifts with respect to phase errors, while the exposure focus window almost stays the same across 23° phase angle regions.

3. The blue regions highlight the best exposure/focus and exposure/focus window at each phase error.

We also performed a similar study for dense contacts. We calculated ~0.003 nm best focus shift per degree phase angle change for both isolated and dense contacts.

To fully understand the impact of phase angle to best exposure focus shift, we also used Prolith to simulate this focus shift effect vs. phase error of an EAPSM at different contact sizes. Figure 4 shows the simulation results. We applied a 30 nm wafer sizing bias to reflect our wafer process bias.

4. As contact size shrinks, the focus shift becomes more sensitive to the phase error of the mask — similar to the mask error enhancement factor with CDs.

We also plotted the actual wafer printing experiment data on the graph. There is a good match between wafer printing results and Prolith simulation at 120 nm contact size. Based on the simulation data, one can see that, as the contact size shrinks, the focus shift becomes more sensitive to the phase error of an EAPSM. This phenomenon is very similar to the MEEF (mask error enhancement factor) in CD control; as the feature size shrinks, the MEEF increases as the reduction of the k₁ factor. As a result, the mask error has a larger impact on the final wafer imaging process window.

In this study, we tried to determine how the wafer focus process window deteriorates when a phase error (out of 180°) exists on an EAPSM. We used a 193-AIMS (Aerial Image Measurement System) tool to study the impact of phase error on the wafer focus process window. The commercially available tool, an MSM-193, was used to measure the across-focus aerial images of the 120 nm contacts in different areas (phase angle and transmission values) on the multiphase EAPSM.

Figure 5 shows the comparison of the aerial image focus window in four different areas. In the AIMS data analysis, we used a 10% exposure tolerance to determine the focus windows in each area of the mask. As shown, there is no noticeable focus window deterioration when the phase error is within ±10°, our available phase error range on the mask. The wafer printing study in Figure 3 shows similar results. Notice that the total exposure focus window remained >0.25 µm in all four phase angle and transmission areas.

5. To determine the rate of process window deterioration with phase error, measurements were taken across-focus using aerial images of 120 nm contacts in different areas (phase angle and transmission) on the mask. A ±10% phase error caused no notable focus latitude deterioration.

A typical EAPSM product has a mean phase angle and an associated phase angle uniformity (if multiple points are sampled across on the mask). As we discussed previously, the average phase angle does not have a noticeable impact on the wafer exposure focus window if it is within ±10° around 180°. On the other hand, since a given phase error can cause a certain shift of the best exposure focus, phase angle uniformity will definitely affect the overall wafer focus process window (called the common process window). Figure 6 demonstrates the impact of phase angle uniformity on overall process window.

6. Since a phase error can cause a shift in the best focus, phase angle uniformity will affect the overall process window. The red box indicates the overlapping process window for all focus/exposure windows across various phase angle and transmission combinations.

In Figure 6 , we used KLA PRODATA software to plot the focus/exposure window at each phase angle and transmission combination based on the results taken from wafer printing studies described previously. Each color in the figure represents the focus exposure window of 120 nm isolated contacts (with 30 nm process bias) at a specific phase angle and transmission combination.

One can easily see the best exposure focus shift when the phase angle varies from 170° to 193°. The blue line represents the overlapped process window out of all the four phase angle and transmission regions (A, B, C and D). It is obvious that the overall process window is reduced because of the best exposure/focus shifts in each region. The box in red represents the overlapping process window. By analyzing the data, we found that the depth of focus has been reduced from the previous >0.25 µm to 0.20 µm, more than a 20% reduction. Similar results have also been observed for the dense contacts.

In this wafer printing experiment and AIMS study, we did not observe noticeable wafer process window change when the absorber transmission was varied from 6.5 to 8.5%. This result indicated that the absorber transmission does not have a direct impact on wafer process window within 2% variation.

Based on the previous wafer-level printing studies and AIMS evaluation, we have the following proposals to the phase angle and transmission specifications of 193 nm EAPSM for 90 nm node wafer production:

• Phase angle uniformity is the most important parameter to control in the photomask manufacturing process. For the most critical layers, such as gate and contact, a ±3° of phase angle uniformity is suggested because of high MEEF effect (with low k₁ value). In our study of 120 nm contact wafer printing with NA=0.65 and σ=0.6, a 3° phase error resulted in <0.01 µm total focus shift. For less critical layers using 193 nm EAPSM, such as via layers, a ±5° of phase angle uniformity is sufficient since the focus shift is less sensitive to phase errors caused by low MEEF effect (with high k₁ value).
• The average phase angle is less critical to the overall process window. When the average phase error is within ±10° (out of 180°), focus latitude does not deteriorate with the phase error. In addition, one should adjust the scanner exposure focus setting to compensate the best exposure/focus shift caused by a large phase angle error. This is especially true for an EAPSM that goes through several repel-and-clean processes.
• The introduction of immersion lithography is going to improve the k₁ factor of the lithography process. This will also reduce the amount of best exposure/focus shift caused by phase errors due to increased k₁ value. As a result, there will be an opportunity to maintain or relax the phase angle uniformity slightly.
• The absorber transmission of an EAPSM does not have a noticeable impact on the wafer process window. Our exposure data demonstrated that a 2% transmission change (from 6% to 8%) does not significantly impact the wafer exposure window. A ±1% transmission tolerance will be adequate.

We conducted a thorough study of the effect of phase angle and transmission of a 193 nm EAPSM on wafer process windows for 90 nm node wafer production. Using a specially fabricated multiphase 193 nm EAPSM plate, we performed the wafer-level lithographic printing studies and AIMS measurements needed to determine the wafer process windows at various phase angle and transmission combinations. We also used Prolith lithography simulation software to study the effect of best focus shift vs. amount of phase errors at different feature sizes, in addition to proposing a practical phase angle and transmission control specification for 193 nm EAPSMs aimed at 90 nm wafer production.