10 Aug 2007
Researchers in Germany are using a modified excimer discharge tube to produce ultrafast pulses with energies up to 50 mJ. Peter Simon and Juergen Kolenda describe how they are using these pulses to machine nanostructures directly into the surface of common metals.
Creating and exploiting periodic nanostructures is currently one of the most active fields of photonics research. In many instances it would be advantageous to create these nanostructures using direct laser machining, but this is often difficult or impossible with current laser technology.
At the Laser-Laboratorium in Goettingen, Germany, we have built a unique laser system based on commercially available products that delivers the high-energy, short-pulsewidth, ultraviolet (UV) output ideal for producing nanostructures. We have also used the system to machine nanostructures in tougher materials, such as metals, in a single-step dry process with large-area throughput.
Suitable laser characteristics
The two laser output characteristics essential for precision micromachining are short wavelength and short pulse duration. In the case of polymers, a short (UV) wavelength means that material is removed predominantly by ablation, which minimizes or even eliminates peripheral thermal damage. This also avoids the creation of re-cast debris and allows small structures with clean edges to be fabricated.
The other advantage of using a short wavelength is spatial resolution. With both direct-write and photomask methods the limiting spatial resolution scales with wavelength because of diffraction. A short (picosecond or femtosecond) pulsewidth is also desirable because it limits the spread of local heating, which is non-zero even with deep UV lasers. In essence, a short pulsewidth maximizes peak power and processing efficiency for a given average power.
At the Laser-Laboratorium we are extending the principles and benefits of laser micromachining to nanostructuring. Specifically, we are interested in nanostructuring materials other than plastics, and in particular, metals.
A key characteristic of metals is their high thermal conductivity. Heat from a nanosecond pulse will spread quickly and cause local melting, making it difficult to create small features. Pulsewidths must be in the femtosecond regime and, for the reasons already described, the output should be in the UV. Our first challenge was to find a suitable laser system.
Excimer lasers are well-proven tools for UV materials processing. They offer very high gain and corresponding high-pulse energies, but even short-pulse versions have pulsewidths of several nanoseconds. Because of their transient gain characteristics, there is no way to modelock the output of an excimer.
The only way to reach the femtosecond regime with off-the-shelf commercial laser technology is to frequency-triple the near-IR output of a Ti:sapphire ultrafast system. The problem is that these benchtop systems have limited pulse energy and overall power. Even powerful kilohertz amplifiers, such as Coherent's Legend, will not exceed 1 mJ pulse energy at 750 nm. When frequency-tripled, this is reduced to tens of microjoules at 250 nm. While this pulse energy is adequate for scientific research, it will not support nanostructuring over extended areas at practical throughput rates. However, a custom ultrahigh power system with multiple amplification stages is not appropriate for this application in terms of its cost and complexity.
To overcome these problems, we have developed a relatively simple system that amplifies an ultrafast pulse to tens of millijoules using an excimer discharge tube as the gain medium. In the latest system, a one-box Ti:sapphire amplifier (Coherent Libra) generates approximately 0.5 mJ at 745 nm. This is frequency-tripled to produce about 50 µJ at 248 nm to match the gain peak of a KrF excimer laser. These 248 nm pulses are then passed through a modified commercial excimer laser (Coherent LPX 200), which is equipped with windows rather than cavity mirrors.
Two aspects of this approach merit discussion. Firstly, the moderate saturation fluence of excimers (2 mJ/cm2 for KrF) sets a limit for the energy-extraction efficiency. This constraint can be relaxed by applying a three-pass bowtie geometry, which increases the effective gain cross-section and offers a good balance between excimer-gain extraction and amplified spontaneous emission noise suppression (see figure 1). The second point is the spectral bandwidth and related pulsewidth. The ultrafast laser delivers a bandwidth of 0.6 nm at 248 nm that supports a pulsewidth of 150 fs. However, the KrF excimer laser has a bandwidth of only 0.4–0.5 nm, so spectral clipping occurs. In addition, the excimer plasma tube windows introduce some dispersion effects leading to further pulse broadening.
The end result is a UV pulsewidth of only 500 fs, with an amplified pulse energy of up to 20 mJ. At a repetition rate of 200 Hz, this translates into an average power of 4 W, which is more than sufficient for our nanostructuring experiments. This type of system is also power-scalable; in an earlier prototype system we reached powers as high as 10 W with pulse repetition rates up to 300 Hz. Another specially modified excimer module delivered pulse energies up to 50 mJ at low repetition rates (10 Hz).
In addition to the UV laser system, nano-structuring also requires an aberration-free optical system capable of reaching the diffraction limit and beyond. One of the main goals with this laser is to produce repetitive patterns of nanostructures. Fortunately, this is a task that is well matched to the use of interference techniques based on phase masks, which create a spatially periodic intensity profile. In some instances we split and recombine the beam with variable phase delay and even use polarization effects. For example, we have developed a scheme that we call "phase-controlled multiple-beam interference", where changing the relative phases of the interfering beams allows us to fabricate a variety of complex structures (see figure 2).
In order to support high-speed, economical fabrication of nanostructures over large sample areas, we have also developed an irradiation setup based on a homemade Schwarzschild-type reflecting objective. This device provides a field size up to 1 mm2, a numerical aperture of at least 0.25 and a demagnification of 15–30.
This entire optical setup has now been used to rapidly create repetitive nanostructures on a wide range of materials, including metals such as stainless steel, nickel, aluminium and copper. Its ability to perform large-area nanostructuring is clearly demonstrated by the stainless steel sample shown in figure 3. This hole matrix was fabricated covering a sample area of 0.5 × 0.5 mm2 with a total machining time of less than 1 s. These scanning electron microscope images show the homogeneous pattern created by the uniform illumination field and confirm structural details as small as 300 nm. Even larger area patterns could be produced by using established stepping techniques.
Nanostructures will continue to grow in importance as new applications develop. We have shown that a novel ultrafast UV laser system, based on off-the-shelf components, can create these structures in a one-step, high-speed, dry process with much lower costs than multistep alternatives, such as lithography.
• This article originally appeared in the July/August 2007 issue of Optics & Laser Europe magazine.