New Technologies—Analysis and Metrology

Using mid-IR lasers in semiconductor manufacturing

Patrick J. McCann, Khosrow Namjou, and I-Na Chao, School of Electrical and Computer Engineering, University of Oklahoma

Advances in mid-infrared (IR) laser technology are possible by using new heteroepitaxial semiconductor laser fabrication methods. These advances should lead to the commercialization of mid-IR lasers that can be operated with thermoelectric cooling modules. These new lasers will be much more economical than current devices, which must be cryogenically cooled. Consequently, they should find cost-effective uses in a variety of semiconductor manufacturing processes in which molecular measurements can help increase yields.

Instruments based on mid-IR laser technology are compatible with most semiconductor processing equipment. They can simultaneously measure multiple molecular compounds and perform reliable molecular measurements in corrosive environments.

This article discusses possible future advances in mid-IR laser fabrication technology that will further the development of compact spectrometers for the measurement of many molecular compounds. The semiconductor industry could use such measurements to monitor chemical processes. Applications include plasma etching endpoint detection, process gas purity monitoring, short-lived radical density measurements in plasma reaction chambers, and emissions monitoring. The measurements can be performed by shining a laser directly through processing chambers or gas lines equipped with IR-transparent windows composed of such materials as sapphire, calcium fluoride, or zinc selenide. The signal from a detector facing the laser provides information on the identity and concentration of the molecules present as the laser is tuned across the relevant absorption bands.

Mid-IR Laser Materials

Semiconductor lasers fabricated using epitaxially grown heterostructures such as those used in fiber-optic communications systems and CD-ROM applications offer important advantages. Nevertheless, semiconductor materials for mid-IR lasers have much lower electron transition energies (band gaps) than those used to make visible and near-IR lasers, which results in relatively low device operating temperatures. Three types of compound semiconductor materials are used to make mid-IR lasers: IV-VI materials (lead salts), III-V materials containing antimony, and III-V quantum cascade (QC) structures.

Mid-IR lasers made from IV-VI materials and QC structures operate in pulsed mode near room temperature, while diode lasers made from III-V materials operate only below 225 K even in pulsed mode. IV-VI lasers exhibit the highest continuous wave (CW) operating temperature, 225 K, which is much higher than that of QC lasers, 160 K.^1,2 Enhanced active-region heating due to Umklapp phonon scattering in the superlattice structure could be the cause of low CW operating temperatures in QC lasers, and may place a fundamental performance limit on devices made from this material. Mid-IR lasers made from IV-VI compound semiconductors thus appear to be in the best position to reach CW operation at temperatures above 250 K, at which thermoelectric cooling modules operate.

IV-VI Semiconductor Fabrication of Mid-IR Lasers

Thermal analysis of IV-VI semiconductor mid-IR lasers shows that a substantial improvement in active-region heat dissipation can be achieved by replacing the thermally resistive IV-VI substrate material with a more thermally conductive material such as copper.³ Such device packaging is made possible by using a novel substrate removal and cleaving procedure that is presently under development at the University of Oklahoma (Norman).⁴ It involves bonding an epitaxial laser structure to an assembly of plates, removing the growth substrate by dissolving a water-soluble BaF₂ buffer layer that was grown between the laser structure and the substrate, and then cleaving the epitaxial laser structure by releasing the plates. This procedure yields Fabry-Perot resonant cavities whose lengths are equal to the thicknesses of the plates. More important, however, is the fact that it allows the laser's active region and only its thin cladding layers to be sandwiched between thermally conductive materials. Since the amount of thermally resistive IV-VI material surrounding the active device region is reduced to a minimum, it should be possible to obtain CW operation well above 250 K.³

Initial experiments to develop the steps for this new laser fabrication procedure were performed using lead selenide (PbSe) layers grown by liquid-phase epitaxy (LPE) on (100)-oriented silicon substrates. These substrates contained buffer layers composed of PbSe, barium fluoride (BaF₂), and calcium fluoride (CaF₂) grown by molecular beam epitaxy (MBE).⁵ The use of gold indium metallurgy has proved to be very successful in forming a thin and uniform bonding layer between the PbSe and a copper plate assembly, and the use of ~1-µm-thick BaF₂ buffer layers allows silicon substrate removal after the substrate has been soaked in water for about three days. All of the steps in this new laser fabrication procedure are sufficiently refined to apply the procedure to the fabrication of lasers using structures grown entirely by MBE on BaF₂-coated PbSe substrates. Unlike structures grown on silicon substrates, these structures have a good thermal expansion match to the growth substrate. Furthermore, because IV-VI semiconductor heterostructures that are lattice-matched with BaF₂ can be grown, this system should be able to offer excellent prospects for obtaining high-performance mid-IR lasers.⁶ In addition to enabling a dramatic enhancement in active region heat dissipation, substrate removal laser fabrication procedures allow expensive IV-VI semiconductor substrate material to be reused. This method can therefore reduce mid-IR laser production costs below those of competing technologies.

Real-Time Mid-IR Laser Measurement of Water and Carbonyl Fluoride

A potentially useful semiconductor industry application for mid-IR lasers is plasma etching endpoint detection. For example, a mid-IR laser emitting in the 5.12-µm spectral region can be used to monitor concentrations of COF₂, which is produced by the reaction of oxide layers exposed to fluorocarbon plasmas. Mid-IR lasers have been used to perform COF₂ measurements in a plasma etching chamber.⁷ In this work the v₄ absorption lines in the 8.14-µm spectral region were probed. Probing COF₂ absorption lines in the shorter-wavelength v₁ band, however, offers the advantage of higher signal-to-noise ratios, since mid-IR detectors have higher sensitivities at shorter wavelengths. Figure 1 shows the absorption lines from the Hitran database for COF₂ in the 5.11—5.17-µm spectral range and the measured emission spectra of a commercially available cryogenically cooled mid-IR laser at various injection currents. This laser exhibits continuous single-mode tuning ranges of approximately 3 cm^—1 in each of the three main branches of the v₁ band.

Figure 1: Mid-IR laser emission spectra at 108 K for injection currents of 550—1010 mA in 20-mA increments as measured by a Fourier transform infrared spectrometer with a 0.5-cm^—1 resolution (top) and absorption lines for COF₂ molecules from the Hitran database (bottom).

A useful mid-IR laser spectroscopy measurement technique is second-harmonic detection. Figure 2 shows three different laser absorption spectra in which the detector signal is plotted as a function of laser frequency, which is tuned by varying the above-threshold injection current with a 20-Hz sawtooth ramp. The top spectrum is a direct absorption measurement showing two dips associated with two water absorption lines. The middle spectrum is a first-harmonic measurement in which the laser wavelength is modulated by superimposing a 20-kHz ac current on the above-threshold injection current and the detector signal is sampled at the same frequency by a lock-in amplifier. This produces a spectrum that is analogous to the first derivative of the direct absorption spectrum. The bottom spectrum is a second-harmonic measurement in which the detector signal is sampled at twice the laser modulation frequency. This spectrum is analogous to the second derivative of the direct absorption spectrum. Second-harmonic detection, which sensitively detects weak absorption features without an optical chopper, provides a flat baseline serving as a common reference for determining molecular concentrations.

Figure 2: Direct, first-harmonic, and second-harmonic mid-IR laser absorption spectra (top) and the strong features of two different water absorption lines (bottom). The second-harmonic spectrum has a flat baseline and peaks corresponding to the water absorption frequencies.

Figure 3: Second-harmonic mid-IR laser absorption spectrum of a 200-Torr gas sample obtained from burning Teflon (top) and absorption lines for COF₂ and water from the Hitran database (bottom).

Figure 3 illustrates a second-harmonic laser absorption spectrum between 1953.5 and 1956.0 cm^—1 for a 200-torr gas sample along with absorbtion lines for COF₂ and H₂O from the Hitran database. This sample was obtained by burning Teflon, and laser absorption lines associated with COF₂ and water are clearly observed. The high spectral resolution of this technique, better than 0.005 cm^—1, enables the simultaneous measurement of COF₂ and water molecules. Measurements were performed by shining the mid-IR laser through a 10-cm-long gas cell equipped with infrared-transparent CaF₂ windows. This cell was able to detect a minimum of ~50 ppm of COF₂. Better sensitivities can be obtained with longer- or multiple-pass gas cells. This spectrum was obtained with a thermoelectrically cooled mercury-cadmium-zinc-tellurium mid-IR detector and a commercially available lead-europium-selenide mid-IR laser mounted in a closed-cycle cryogenic refrigerator.

Figure 4: COF₂ concentration values obtained by scanning mid-IR laser radiation across COF₂ absorption lines in the 5.1-µm spectral region with a time resolution of better than 100 ms.

This study provides the first known conclusive proof that COF₂, a highly toxic molecule with a minimum threshold limit value of 2 ppm, is a reaction product in air/Teflon chemistry. Figure 4 shows COF₂ concentration values obtained from second-harmonic peak heights every 100 millisecond over a 20-minute period following the collection of the COF₂ gas sample. An exponential decrease in the concentration is observed, and a fit to the data shows that these COF₂ molecules have a lifetime of about 12 minutes. Water molecules also showed an exponential decrease in concentration and had the same lifetime. These data suggest that COF₂ reacts with water to produce hydrogen fluoride and carbon dioxide. It should be a relatively straightforward exercise to perform similar concentration-versus-time measurements for COF₂ molecules in a plasma etching chamber. This would provide an accurate method for detecting the etching endpoint of oxide materials. In this case, a constant COF₂ concentration should be measured during oxide etching and a rapid decrease should be observed just before etching is finished.

Conclusion

This article discussed the results of measurements of carbonyl fluoride (COF₂) molecules that were obtained with an instrument that cost less than $40,000 to assemble. Compact and, ultimately, affordable, the mid-IR laser spectrometer is user-friendly because it does not use liquid nitrogen cryostats. It has proved to be a very reliable tool for measuring both COF₂ and water molecules, and with minimal modification it can be used commercially in the semiconductor manufacturing environment.

Acknowledgments

The authors would like to thank the National Science Foundation, the Oklahoma Center for the Advancement of Science and Technology, and Ekips Technologies (Norman, OK) for their financial support.

References

1. Z Feit et al., "Low Threshold PbEuSeTe/PbTe Separate Confinement Buried Heterostructure Diode Lasers," Applied Physics Letters 68 (1996): 738—740.

2. A Tredicucci et al., "High Performance Interminiband Quantum Cascade Lasers with Graded Superlattices," Applied Physics Letters 73, (1998): 101—103.

3. KR Lewelling and PJ McCann, "Finite Element Modeling Predicts Possibility of Thermoelectrically Cooled Lead-Salt Diode Lasers," IEEE Photonics Technology Letters 9 (1997): 297—299.

4. PJ McCann, "Method for Fabricating Semiconductor Laser," U.S. Patent Number 5,776,794, July 7, 1998.

5. BN Strecker et al., "LPE Growth of Crack-Free PbSe Layers on (100)-Oriented Silicon Using MBE-Grown PbSe/BaF₂/CaF₂ Buffer Layers," Journal of Electronic Materials 26 (1997): 444—448.

6. I Chao et al., "Growth and Characterization of IV-VI Semiconductor Heterostructures on (100) BaF₂,"Thin Solid Films 323 (1998): 126—135.

7. DB Oh et al., "In Situ Diode Laser Absorption Measurements of Plasma Species in a Gaseous Electronics Conference Reference Cell," Journal of Vacuum Science Technology B 13 (1995): 954—961.

Patrick J. McCann, PhD, has been a faculty member of the school of electrical and computer engineering at the University of Oklahoma (Norman) since 1990. He supervises an active research group concerned with the application of IV-VI semiconductors and group II-A fluoride insulators. His research objectives include the fabrication of high-operating-temperature mid-IR lasers for molecular spectroscopy applications, the development of light-emitting devices based on rare-earth-doped fluoride layers grown by MBE, and the integration of optoelectronic devices with silicon circuitry. He received his BS in engineering physics from the University of California, Berkeley, in 1981 and his PhD in electronic materials from MIT in 1990. (McCann can be reached at 405/325-4288 or pmccann@ou.edu.)

Khosrow Namjou, PhD, joined the school of electrical and computer engineering at the University of Oklahoma in 1998 as a postdoctoral research associate. He is researching the development of room-temperature IV-VI diode lasers for IR spectroscopy applications. Previously he focused on developing the techniques of frequency and wavelength modulation for semiconductor diode lasers and infrared molecular spectroscopy with IV-VI and quantum-cascade lasers. He was the first to apply quantum cascade lasers to infrared absorption spectroscopy. Namjou is a member of the American Physical Society and the Optical Society of America. He received his MS in 1991 and his PhD in 1998 in engineering physics from Stevens Institute of Technology in Hoboken, NJ. (Namjou can be reached at 405/325-4748 or knamjou@ hotmail.com.)

I-Na Chao is a PhD candidate in the school of electrical and computer engineering at the University of Oklahoma. Her research includes Fourier transform infrared characterization of IV-VI tunable diode lasers and epilayers and growth of IV-VI semiconductor heterostructures on barium fluoride. She received her BS in physics from National Cheng-Kung University (Taiwan), an MS in audiology from Phillips University (Enid, OK), and an MS in electrical engineering from the University of Oklahoma. (Chao can be reached at 405/325-5419 or ichao@ou.edu.)

MicroHome | Search | Current Issue | MicroArchives
Buyers Guide | Media Kit