Recent advances in nonaccelerator techniques have ushered in commercially available sources of extreme UV radiation.
FAIZ RAHMAN, OHIO UNIVERSITY
The electromagnetic spectrum has several regions where emitters, detectors, or both are not readily available. Terahertz and the far-infrared are prime examples. Even within the visible spectrum, much has been said about the “green gap” — the wavelength region between 500 and 600 nm where efficient and high-brightness solid-state sources were, until recently, hard to come by (Figure 1).
Figure 1. An LDLS (laser-driven light source) module from Energetiq Technology Inc., powered by a separate solid-state laser unit (not shown). Courtesy of Energetiq Technology Inc.
Moving to shorter wavelengths, another gap exists in the extreme ultraviolet (EUV) region between the deep ultra-violet (DUV) and x-radiation. This portion of the spectrum, which covers the wavelength interval from roughly 10 to 50 nm, has traditionally had few sources, not counting large synchrotrons that have mostly driven the research scene to date.
With several pressing applications in the EUV region, much effort has been directed at developing compact sources of ultrashort-wavelength UV radiation in recent years. That work has paid off with the commercial availability of several types of EUV light sources. These include the laser-produced plasma (LPP) sources — suitable for semiconductor lithography — and discharge-produced plasma (DPP) sources that can be used for less demanding applications in mask inspection and materials analysis.
Extreme UV light sources are used in semiconductor lithography. A close-up view of a chip wafer.
The EUV gap is bounded on the long wavelength side by electric discharge-based sources that produce DUV radiation. The short wavelength side, on the other hand, abuts the soft x-ray region where radiation is generated by slamming streams of electrons into hard metal targets. Either of these approaches can, in principle, be used for generating EUV radiation. In practice, the very broad spectrum typical of x-ray bremsstrahlung radiation precludes the use of x-ray tube-like devices for efficient generation of radiation in the 10- to 50-nm region. This leaves plasmas as the only viable source of EUV radiation. Most ordinary plasma-based radiation sources, however, barely generate enough power at wavelengths shorter than 100 nm, because ordinary thermal plasmas are populated mainly by neutral atoms and low-charge-state ions.
Electronic transitions in such species don’t have the energy to generate very short wavelength UV radiation. One needs far more energetic plasmas that contain highly charged ions where gaps between electron energy states are wide enough to produce photons at EUV wavelengths. Such plasmas are hard to produce. Neutral atomic gases have to be supplied with tremendous amounts of energy to create EUV-capable plasmas. Even though this is significantly harder to do than just striking a discharge in a low-pressure gas enclosed in a glass tube, it can be accomplished with some ingenuity.
The quest for shorter wavelengths
Historically, xenon plasmas have been favored for generating broad-spectrum radiation that extends deep into the UV region. Xenon lamps are ubiquitous as general-purpose broadband light sources in optics labs. By preparing higher-temperature xenon plasmas, their emission can be extended further to shorter wavelengths. This can be done in two ways: passing a very high-pulsed-current discharge through capillary-confined xenon gas (DPP) or by heating xenon gas with intense-pulsed laser beams (LPP).
LPP EUV sources offer several advantages because they can generate significant amounts of EUV flux at reasonably high efficiency (Figure 2). DPP sources, although smaller and more manageable, are not capable of the same degree of brightness as their laser-powered counterparts. Radiative heating of tenuous plasma is well known to result in the emission of UV and x-rays from astrophysical sources. Now, the technology to do this on Earth has been perfected to a degree where it is opening doors to new applications.
Figure 2. A laser-produced plasma (LPP) extreme ultraviolet (EUV) source vessel from ASML Inc. The radiation output port is seen at the top. Courtesy of ASML Inc.
Gaseous xenon, frozen xenon, and xenon jets are all good sources of DUV and EUV radiation. Intense laser energy is focused on a small volume of xenon, which rapidly ionizes to form a multiple-charge-state plasma, capable of DUV and EUV emission. IR-to-DUV broadband sources that are based on plasma formation inside a quartz bulb and filled with gaseous xenon are available from Energetiq Technology Inc. (recently acquired by Hamamatsu Corp. of Japan). The plasma is formed by heating xenon with pulses from a solid-state, fiber-coupled laser. Essentially, energy is supplied to a bulb’s fill gas through laser-based heating. The electrode-less design results in much longer lamp life than conventional electric-discharge-based xenon lamps where electrode erosion limits the ultimate lifetime. Additionally, indirect heating through a laser beam also allows much larger amounts of energy to be coupled to the xenon plasma, thus extending the emission deeper into the UV. In principle, such a lamp can also generate EUV radiation. However, the conversion efficiency is poor (~0.5 percent) because much energy is also diverted to the production of longer wavelength radiation.
A much better approach is to use denser forms of matter, such as solid and liquid targets. Intense laser beams, usually in the IR region, are focused on metallic targets, such as gold plates and wires. Heat deposited by laser radiation causes the metal to melt, vaporize, and then form a plasma. Interaction of laser energy with the plasma raises its temperature and results in radiation emission at various wavelengths, including some in the EUV region. The plasma is extremely hot, and because of its high temperature, much of the radiation is emitted through stimulated rather than spontaneous emission.
Tin-droplet-based plasma sources
ASML, based in the Netherlands and the U.S., and Gigaphoton in Japan have developed commercial LPP EUV sources for use in semiconductor lithography. These sources utilize tin plasma, which is formed by firing CO2 laser pulses at a steady stream of tiny droplets of molten tin. The small size (~30 μm in diameter) of these droplets ensures that little contamination is generated inside the plasma vessel as each drop is hit by a laser pulse. Historically, particulate debris generated during laser ablation of solid targets had made such sources impractical because of the propensity to coat radiation collection optics inside the plasma vessel. Use of minuscule tin droplets mostly solves this problem by producing very little particulate contamination. Indeed, the development of tin-droplet-based plasma sources has been one of the most prominent inventions making modern LPP sources practical.
Other innovations include magnetic fields to divert tin ions away from the radiation collection mirror and the use of a curtain of hydrogen gas to chemically react with tin vapor, removing it as tin hydrides. Such strategies keep the insides of the ultrahigh-vacuum plasma chamber clean and minimize periodic maintenance of optics inside the chamber.
Another important technology that has been incorporated in all commercial tin plasma EUV sources is laser prepulsing. The technique involves irradiating each tin droplet with a separate lower-power laser pulse just before the main laser pulse hits it. The prepulse laser can be a smaller CO2 laser or it can be a solid-state neodymium laser. Prepulse irradiation serves to distort the molten tin droplet into an extended disk shape, which can be more effectively irradiated by the subsequent main laser pulse. This technique has resulted in a major increase in the efficiency of modern tin-plasma-based EUV sources.
Radiation generated from tin plasma emission peaks at 13.5 nm. Because of the high efficiency of tin-based plasma sources and the availability of multilayer mirrors to manipulate radiation at this wavelength, 13.5 nm has become the standard for the next generation of EUV-based photolithography. As wavelengths drop below about 300 nm, UV radiation is increasingly absorbed by ambient air. Wavelengths below 100 nm cannot be transported through air at all and thus EUV radiation needs an ultrahigh-vacuum system from the point of generation to use. Similarly, refractive optical elements such as lenses cannot be applied at these short wavelengths, making the use of reflective elements mandatory. For optical lithography, for instance, a host of mirrors — from the primary radiation collection mirror to a multitude of other curved and plane mirrors — are needed in the optical train ending at the mask reticle.
Specular elements used at EUV wavelengths have to be metallic multilayer Bragg stack mirrors, as opposed to simple metal film-on-glass mirrors. These are made by depositing hundreds of alternating silicon and molybdenum thin films on suitably shaped substrates. Although exceedingly well engineered, their EUV reflectivities do not generally exceed 60 to 70 percent. Use of a large number of such relatively high-loss elements makes for relatively little radiation actually arriving at the reticle plane. This necessitates producing a much larger quantity of EUV radiation at the point of generation than is actually needed to perform a lithographic exposure. Current state-of-the-art ASML EUV sources, for example, can produce in excess of 250 W of EUV power at the intermediate focus (Figure 2). This is sufficient to process around 125 wafers per hour — a throughput considered necessary for putting EUV lithography into commercial production. As this milestone has been reached, all major chip manufacturers are planning to transition to EUV lithography over the coming months and years. At first, lithography at EUV wavelengths will only be used for the most critical layers of integrated circuit stack but will encompass more and more layers as technology nodes shrink further in the future.
Compared to LPP, the electric discharge route for EUV generation is less technically formidable. However, it has been beset with low operational efficiency, which has prevented its use for applications that require bright sources such as mainstream semiconductor lithography (discussed above). The relative simplicity of this approach has resulted in the commercial availability of DPP EUV sources for applications that do not expressly require very high flux. These include research, as opposed to production, on EUV lithography technology (EUV optics, resist outgassing, etc.), mask inspection, and materials analysis.
The very first DPP EUV sources were based on a simple electrode-based discharge tube design. In this configuration, an extremely high-electric-current discharge was made to pass through xenon gas flowing through a narrow ceramic capillary. The high-temperature xenon plasma formed radiates with a broad spectral distribution extending from IR to soft x-rays. This scheme works and is comparatively simple to implement, but its electric power-to-EUV conversion efficiency is very low. Scaling it up to achieve higher power levels is also not straightforward. Furthermore, similar to all electrode-equipped discharge tubes, it suffers from electrode material loss, which is only exacerbated by the need to inject very high currents compared to other discharge-based light sources.
No electrode loss
An electrode-less, discharge-based approach for EUV generation, commercialized by Energetiq Technology Inc. of Woburn, Mass., completely avoids the problem of electrode loss. The company’s Z-Pinch EUV source (Figures 3 and 4) runs on pure xenon gas and comprises two copper plates with four circular through-holes. Discharge results inductively, forming three plasma loops that snake around the holes in the copper plates. The high current discharge is fed from a capacitor bank with appropriate control electronics. This current flows through magnetic coupling cores located above and below the two copper plates, which then inductively induce a current in the xenon gas, ultimately forming a plasma. Current in the plasma loops generates a strong magnetic field of its own, which pinches the plasma and confines it in such a way that it is kept away from any solid surfaces.
Figure 3. This CAD rendering shows the central part of a Z-Pinch source from Energetiq Technology Inc. Courtesy of Energetiq Technology Inc.
The highly concentrated plasma in the center bore is the location where EUV radiation is generated. The radiation can be coupled to a highly evacuated external beam line — no radiation collection mirror is needed. Although there are no electrodes to get eroded over time, sputtering of copper in the central bore can rapidly corrode its inner wall surface. This problem is greatly mitigated by the use of a silicon carbide liner that protects the bore surface. Silicon carbide also gets sputtered by bombardment from energetic ions in the xenon plasma, but at rates that are much lower than that of copper. The bore liner is thus the only consumable part that needs replacement, typically after one billion shots.
Figure 4. An Energetiq Technology EQ-10HP Z-Pinch EUV system. Courtesy of Energetiq Technology Inc.
EUV light sources from ASML, Gigaphoton, and Energetiq Technology now provide access to radiation in the interval between DUV and x-ray regions, filling a conspicuous gap in the spectrum at the short-wavelength end. The commercial product offerings from these companies are varied enough to serve in a number of application areas. Ultimately, both larger and smaller sources, compared to what is currently available, will come to the market. This will benefit both existing needs, such as semiconductor lithography, and open up new applications for EUV radiation in materials research, biomedical science, and photochemistry, among other areas. EUV technology thus seems poised for a bright future.
Meet the author
Faiz Rahman is a faculty member at Ohio University. He is a senior member of The Optical Society of America and of IEEE; email: [email protected].