Access Official Technical PDF

Share Summary:LinkedInEmail

Nanoimprint lithography (NIL) is being developed for semiconductor manufacturing with a focus on practical production requirements: throughput, overlay control, defectivity and mask life. Recent equipment results show how changes to field processing, alignment and resist dispensing can address these requirements, while highlighting the process controls needed to manage wafer and mask distortion.

Unlike optical lithography, NIL forms patterns by bringing a relief-patterned mask into contact with a low-viscosity resist. The resist is deposited onto the wafer by jetting, flows into the mask features through capillary action, and is then cured with ultraviolet (UV) radiation. Removing the mask leaves the replicated pattern in the resist. This approach can reproduce nanoscale features without the wide-diameter lenses and light sources used in advanced optical equipment.

Nanoimprint lithography throughput: reducing field overhead

Throughput is a key contributor to lithography cost of ownership. The reported cluster-tool approach uses multiple imprint modules; the NZ2C configuration described uses four stations. Productivity improvements therefore depend both on processing speed at each station and on limiting the time spent on repeated operations.

A key change is multi-field dispensing. Rather than completing dispensing and imprinting sequentially for each field, the tool continuously dispenses resist across multiple fields. This strategy reduces overhead by about 0.24 seconds per field and supports throughput of up to 90 wafers per hour with the cluster system.

Increasing throughput must not come at the expense of process quality. In the reported comparison, non-fill defectivity for the multi-field approach was nominally similar to that of sequential imprinting. This is an important consideration for production evaluations: throughput changes need to be assessed alongside defectivity, not in isolation.

Alignment and overlay control in NIL

Accurate overlay—the alignment of a new patterned layer with existing wafer features—is essential in advanced device fabrication. The system uses Through-The-Mask (TTM) alignment, in which Moiré marks on the mask and wafer provide alignment signals. First-order alignment terms are measured with sensitivity on the order of 1 nm.

On device wafers, signal quality can be improved by using multiple wavelengths and controlling the intensity of each. This helps manage signal variation and blooming, which can otherwise contribute to measurement error.

High-order distortion correction also requires a different approach from that used in an optical scanner. A scanner can adjust the stage and lens during exposure; an imprint tool instead combines a magnification actuator with field-by-field heat input. An array of piezo actuators applies force to address magnification, while a digital micromirror device (DMD) array directs heat through the mask to correct higher-order distortion in individual fields.

Reported overlay results

On test wafers, cross-matched machine overlay (XMMO) against a level patterned with an ASML 1950 ArF immersion tool measured 2.9 nm in x and 3.2 nm in y for mean plus three sigma. The measurements covered 84 fields, with 12 points per field, and were averaged across 23 wafers. Single-machine overlay (SMO), measured against a level patterned with the FPA-1200 NZ2C tool, was 2.2 nm in x and 2.4 nm in y for mean plus three sigma. Those results were averaged across three wafers; reported residual distortion was 0.7 nm for mean plus three sigma.

These figures describe test-wafer results, not a universal production outcome. They provide reference points for evaluating alignment and distortion control under the conditions measured.

Drop Pattern Compensation for topography-related errors

Wafer topography and bending can create overlay errors that are difficult to address with alignment adjustments alone. Drop Pattern Compensation (DPC) adds another process control: it changes the resist-drop pattern so the resist film thickness can be adjusted to match mask and wafer shape.

Potential sources of distortion addressed by DPC include chuck-induced errors, pattern-induced distortion, grid distortion from an optical scanner and stress from previous layers. In NIL, the method can also help compensate for overlay errors associated with mask bending.

In one reported example, areas between die on a device-like wafer were recessed by about 20 nm. The topography produced overlay errors of 5.4 nm at three sigma. Applying both high-order distortion correction and DPC reduced the reported errors to 3.4 nm in x and 3.3 nm in y at three sigma. The example illustrates how resist placement can be part of an overlay-control strategy when wafer shape varies across a field.

Mask life and cost of ownership

NIL uses a master mask and replica masks, making replica-mask lifetime an important cost-of-ownership factor. The reported mask life exceeded 300 wafer lots, compared with an earlier target of 80 lots. The extension was attributed to techniques for mitigating particles in the tool, including particle source control, avoidance and collection. On-tool mask cleaning was identified as a possible next step, rather than a demonstrated capability in these results.

What the results mean for manufacturing

For engineers and manufacturing teams assessing NIL, the results point to an interconnected set of evaluation criteria. Cluster-tool throughput, resist-fill defectivity, overlay performance, wafer topography and mask lifetime all influence whether a process can meet production needs. Multi-field dispensing can reduce field overhead, while TTM alignment, distortion correction and DPC provide complementary ways to manage placement errors.

The practical implication is that NIL readiness should be judged across the full process and equipment configuration, not by resolution or throughput alone. The reported progress—including throughput up to 90 wafers per hour, nanometer-scale overlay results on test wafers and replica-mask life beyond 300 lots—shows advances in key manufacturing considerations. Continued attention to defectivity, distortion and mask management remains central to evaluating its cost and suitability for high-volume semiconductor production.

Access Official Technical PDF

About The Author