SPIE Membership Get updates from SPIE Newsroom
  • Newsroom Home
  • Astronomy
  • Biomedical Optics & Medical Imaging
  • Defense & Security
  • Electronic Imaging & Signal Processing
  • Illumination & Displays
  • Lasers & Sources
  • Micro/Nano Lithography
  • Nanotechnology
  • Optical Design & Engineering
  • Optoelectronics & Communications
  • Remote Sensing
  • Sensing & Measurement
  • Solar & Alternative Energy
  • Sign up for Newsroom E-Alerts
  • Information for:
    Advertisers
SPIE Photonics West 2018 | Call for Papers

SPIE Journals OPEN ACCESS

SPIE PRESS

SPIE PRESS




Print PageEmail Page

Lasers & Sources

X-Ray Visions

The tabletop laser system yields a plasma x-ray source.

From oemagazine March 2001
28 March 2001, SPIE Newsroom. DOI: 10.1117/2.5200103.0006

Unlike conventional optical lithography, x-ray proximity lithography generates a 1:1 image of the mask on the wafer, dispensing with reduction optics. Thus, proximity systems do not suffer from depth-of-field limitations, and clear images of the mask can be obtained over a relatively large range behind the mask. This unique feature makes x-ray proximity lithography complementary to optical lithography when a precise, flat image plane is not available. In addition, the technology is not vulnerable to mask or wafer contamination like other systems; x-rays can pass through contaminants without compromising the exposure process.

Our group has developed a tabletop laser-based x-ray source that can produce more than 20 W of x-ray output at 11.5 Å using a four-beam, 300-W, diode-pumped neodymium-based yttrium-aluminum-garnet (Nd:YAG) system capable of generating more than 1014 W/cm2 at the focal plane. Such a system may find niche applications as lithography progresses to the sub-100-nm nodes.

problems with plasma

Synchrotron x-ray sources are far too large for semiconductor fabrication facilities. A compact alternative for narrowband x-ray output involves fusing a pulsed-laser- generated plasma onto a metal target, a process that generates bursts of x-rays. In the past, investigators required high-power (20 J/pulse) lasers to achieve good x-ray conversion. Such systems suffered from beam and x-ray port contamination caused by the high level of debris coming from the metal target. In addition, the systems generated a high dose of x-ray emission per pulse, which limited dose control. For example, if a system only requires five pulses to expose the wafer, each pulse provides about 20% of the required dose; therefore, the best dose control can only be about 20%.

We addressed these issues by minimizing the laser pulse energy, which yielded a significant reduction in debris and in x-ray dose per pulse. Efficient x-ray generation requires high intensity, however. To reduce pulse energy without compromising intensity, we reduced the laser pulse duration.

Lithographic systems demand a pulse-to-pulse dose control of better than 1%, which means the x-ray dose per pulse must be lower than 1% of the total dose required per wafer exposure. High volume applications require high average power. So to meet both goals, we achieved power scaling by repetition rate and multiple beams.

laser plasma

FIGURE 1: An Nd:YAG laser system generates x-ray output.

The tabletop x-ray source consists of a laser system and a chamber for the metallic target (see figure 1). The master oscillator of the laser system provides the seed pulses that get a boost from the preamplifier and then split four ways to be further enhanced by four parallel amplifiers. The output beams of the amplifiers are combined in pairs to form two beams that feed into the target chamber via two separate ports fitted with focusing lenses.

In the master oscillator, a single lens focuses 250 W of 808 nm output from a five-bar laser diode array onto the end of the laser rod. Coated for high transmission at 808 nm and high reflectivity at 1064 nm, the back end of the laser rod forms one end of the resonator.

In order to generate short-duration pulses with near-diffraction-limited beam quality, we use a resonator design that includes a Q-switch, an acousto-optic mode-locker, and a cavity dump. As the Q-switch opens, a single mode-locked pulse is trapped inside the resonator to build up intensity. At the time of peak intensity, the cavity dump forces the entire pulse to exit the resonator. As a result, the master oscillator delivers more than 1 mJ per pulse of output with pulse duration of 700 ps, near-diffraction- limited beam quality, and a repetition rate of up to 1 kHz.

An Nd:YAG-based preamplifier boosts the energy of each pulse prior to injection into the four main amplifiers. Since the master oscillator provides 1 mJ of output per pulse, a passive four-pass amplification scheme is sufficient to generate about 100 mJ of output. The preamplifier consists of two heads transversely pumped by diode laser arrays. The dual-head configuration helps compensate for thermal lensing and stress birefringence in the laser rod. The output of the preamplifier has a pulse energy of 100 mJ and carries the same duration and repetition rate as the master oscillator. At 300 Hz the beam quality is near-diffraction-limited.

FIGURE 2: Diffraction patterns of output beams of a system show microstructures caused by hard aperturing in the laser rods.

The preamplifier beam splits four ways to feed four parallel amplifiers, which are similar to the preamplifier. Each amplifier emits a stream of 250-mJ, 700-ps pulses at a repetition rate of 300 Hz for an average power of 75 W. The beam quality is about 1.5 times diffraction-limited. The diffraction pattern displayed by the output beam of the amplifier is caused by hard aperturing in the rods (see figure 2).

generating x-rays

The complete laser system produces four linearly polarized beams with an aggregate power of 300 W. Polarizers combine the beams in pairs to yield an output of two separate beams of 150 W. The two laser beams are directed toward the metal tape target at a position 25° off of the perpendicular to the x-ray exit port. F/6 focusing optics reduce the beams to a 10-µm focal spot.

FIGURE 3: The high power of the laser system causes air breakdown, triggering a bright spark.

Focusing of the laser beam in air causes intense and very loud air breakdown (see figure 3), which stops occurring only below 70 torr of helium. At such low pressure, the fine dust-like debris from the ablated copper tape can reach and contaminate the focusing optics and x-ray port, however. To minimize contamination, we use positive helium flow from the x-ray and laser ports toward the copper target. We also flush the plasma volume.

The target material for the laser- produced plasma is a 0.5-in.-wide, 1-mil.-thick copper tape that produces 11.5-Å x-ray emission and advances after each pulse. The same technique can be used to produce extreme ultraviolet (EUV) emission at 135 Å with the use of copper or gold tape. EUV generation requires only 1012 W/cm2 rather than 1015 W/cm2 of output, opening the door for a simpler laser system to do the job.

A single laser beam from the four-beam system yields x-ray pulses of approximately 3 W. Spatial and temporal overlaps of the outputs from all four amplifiers at the target location yield higher conversion efficiency, however. In tests using film exposure and p-i-n photodiode measurements, the system generated average x-ray powers of more than 20 W into 2π sr, with an effective lithographic wavelength of 11.5 Å. We will realize higher power output in the future by scaling up the laser system.

Converting 1064-nm radiation to 532 nm via second-harmonic generation allows for tight focusing without air breakdown even at 760 torr of helium. At atmospheric pressure, the debris can be stopped from contaminating the optics and x-ray ports. At low repetition rates, the 532-nm approach demonstrated x-ray yields similar to the 1064-nm system, but we have yet to operate it at 300 Hz.

Most proximity lithographic applications require a collimated x-ray beam. Our system uses polycapillary optics to capture x-rays from the point source and direct them out of the target chamber as a collimated beam to expose a full field on the wafer. In tests we observed uniform output to within a few percent. We expect to implement a full-power demonstration in the near future.

The complete x-ray source is very compact and has the potential to be reliable and robust. The laser system requires only a 3 X 6 ft. tabletop with two 6-ft. electronic racks. The target chamber occupies a 2 X 3 X 3 ft. structure. This system can provide a unique tool for the microelectronic industry, and we believe it to be capable of exposing several wafers per hour. oe


going to extremes

X-ray proximity lithography is not the only technology for fabricating feature sizes 30 nm and lower. In September 2000, members of the industry consortium International Sematech (Austin, TX) settled on extreme ultraviolet (EUV) lithography and electron projection lithography (EPL) as the two strong contenders for the 50-nm and 35-nm nodes. "Lithographers agree that EUV and EPL are the technologies to address the unique applications of DRAM and ASIC," says Sematech's Gerhard Gross.

Extreme ultraviolet lithography is a mask-based projection lithography technique performed at wavelengths of 10 to 14 nm. Because high-energy EUV wavelengths suffer heavy absorption with refractive components, both masks and projection systems for EUV systems are all reflective, based on substrates coated with molybdenum/silicon (Mo/Si) multilayer thin films to enhance reflectivity. Development work in this area continues around the globe, with researchers focusing on increasing the robustness of condenser optics, generating defect-free mask blanks and low-defect reticles, and reducing the cost of ownership.

In EPL, which some may recognize by the moniker electron beam or e-beam lithography, a magnetically controlled electron beam exposes the photoresist on the wafer. Whether or not EPL has a role to play depends on the use. "We see no place for EPL for our applications, which are high volume," says Yan Borodovsky, director of advanced lithography at Intel (San Jose, CA). "Cost of ownership models show that the throughput of the EPL tool is too low to support our manufacturing."

—staff report


Harry Rieger

Harry Rieger is senior research scientist for JMAR Research, Inc., San Diego, CA.