Testing Optical Damage for 157 nm Lithography

As photolithographic processes use ever shorter wavelengths to produce more densely packed circuitry on silicon chips, the choice of materials suitable for use in the deep ultraviolet (DUV) spectral region becomes severely limited. This article reports preliminary life-test results for calcium fluoride (CaF₂) irradiated at 157 nm by F₂ laser beams.

The sample housing and beam delivery tubes were purged continuously with high-purity nitrogen to keep the background oxygen level as low as possible and to sweep away any potential organic gases liberated from the sample mounting hardware and overall experimental apparatus. Data were collected to evaluate induced changes in transmission, wavefront distortion and birefringence over the course of billions of shots at a nominal fluence of 0.1 mJ/cm².

Figure 1 illustrates the setup for these experiments. The four CaF₂ samples, each measuring 2 × 4 cm in cross section and 8 cm in length, are housed in an aluminum enclosure with a removable cover that can be clamped in place. An O-ring, running along the top edge of the enclosure, allows the cover to form a seal so that the interior of the enclosure can be purged with nitrogen. A purge rate of 3 L/min maintains the atmosphere around the samples at <5 ppm of O₂ and with H₂O as low as possible (generally <5 ppmv). The nitrogen is purified with a catalysis filter to remove O₂, H₂O and other contaminants as well as particles larger than 0.003 µm.

Next, 157 nm radiation from two 2 kHz F₂ lasers illuminates the samples from opposite directions, along the 8 cm dimension axis. The laser pulses are timed to fire alternately at intervals of 250 µsec, thus giving a true representation of 4 kHz laser operation and avoiding any possible issues with damage mechanisms that depend on the time between pulses. Previous experiments were conducted on optical damage at 193 nm¹ using delay lines to synthesize high pulse repetition rates, but the lasers were pulsed in bursts, where the pulse spacing was about 50 nsec and the time between bursts of four pulses was 500 µsec.

1. The setup for 157 nm sample exposure.

The two 157 nm laser beams are carefully aligned such that they are as collinear as possible. Illuminating the samples in this way simplifies the optical design, and results in a distribution of intensity through the samples that is significantly more even than that produced by unidirectional irradiation. Another feature of this arrangement is that there are no optical coatings in the beam delivery system aside from those in the lasers. Coatings for beam splitting and combining are generally more prone to optical damage than those used at normal incidence. This problem is more pronounced as the operating wavelength becomes shorter, and this experiment does not include exposure of multilayer dielectric coatings.

The laser energy is monitored using the internal energy monitors, and the pulsewidth is monitored using an external vacuum photodiode. Energy and pulsewidth are recorded for both lasers on a periodic basis along with the current O₂ and H₂O levels. This information is used to calculate the dose delivered to the samples by the integral-square definition of pulsewidth. The units are 106((mJ/cm²)²)/nsec according to the Sandstrom definition of integral square pulsewidth as it pertains to compaction dose:

After a given number of pulses, the samples are tested for changes in transmission, optical path length, and birefringence. The transmission is tested by placing a pyroelectric detector in the enclosure and measuring the pulse energy after it has passed through the exposed area of the samples, an unexposed area of the samples, and the area with the samples removed (the samples are mounted on a sliding platform to accommodate this measurement). Changes in optical path length are checked using an interferometer, and changes in birefringence are checked using a scanning birefringence measurement system.

To date, measurements have been taken at 1 billion, 2 billion and 4 billion pulses, and no measurable changes have been detected in the birefringence of the samples. The same is true of the interferograms; no refractive index gradients have been induced by irradiation at 157 nm. The approximate dose delivered to the samples at a fluence in the 0.15 mJ/cm² range after 4 billion shots is 3.07 × 106((mJ/cm²)²)/nsec.

Figure 2 shows birefringence plots for one of the samples, including results for unexposed samples and samples after 4 billion shots. The units referred to in the legend to the right of each plot are in nm/cm. The as-received average birefringence in the sample is about 0.2 nm/cm.

2. Birefringence plots for sample #100 show the initial condition (top) and the condition after 4 billion shots (bottom). The units referred to in the legend to the right of each plot are in nm/cm.

Figure 3 shows interferograms for the same sample, at 1 billion and 4 billion shots. Note that the sample actually shows a slight improvement at 4 billion shots compared with 1 billion shots. At 4 billion shots, RMS is 0.0125 µm, and at 1 billion shots is 0.0157 µm. Another sample (#102, not pictured) showed no change, at 0.0256 RMS µm wavefront distortion.

3. Interferograms show the condition of sample #100 after 1 billion (top) and 4 billion (bottom) pulses. Note that the sample actually shows a slight improvement after 4 billion shots compared with 1 billion shots.

As of the shot count reported in this document, there have been no measurable changes in birefringence or wavefront distortion in the irradiated CaF₂ samples. As of October 2001 the shot count stood at 50 billion pulses and no changes in the test materials had at that point been detected.