Laser, dry and plasmaless, photoresist removal
07/01/1997
Laser, dry and plasmaless, photoresist removal
Boris Livshits, Ofer Tehar-Zahav, Eli Iskevitch, Menachem Genut, Oramir Semiconductor Equipment Ltd., Haifa, Israel
The decrease in device dimensions has brought new challenges in microlithography and, consequently, in photoresist stripping. Current fabrication technology is capable of dealing with device feature sizes of 0.35 ?m, but the critical dimensions of ICs are expected to approach 0.18 ?m by the year 2000. The anticipated reduction in dimensions requires considerable changes in manufacturing technologies, including the stripping process.
Two main photoresist stripping methods are currently used in the semiconductor fabrication industry [1-4]:
1. Dry stripping uses mostly plasma, and to a lesser extent, O3 -, O3/N2O -, or UV/O3 - based stripping; and
2. Wet stripping uses acids such as H2SO4/H2O2 (Piranha) or organic solvents.
The current dry method, reactive plasma ashing, suffers from major drawbacks due to incomplete removal of photoresist and resist popping. Plasma ashing is also associated with several types of damage mechanisms introduced or aggravated by the plasma. These damage mechanisms are due to charges, currents, electric-field-induced UV radiation, contamination (such as alkali ions, heavy metals, and particulates), and elevated temperatures. Resist removal is often incomplete, especially after "tough" resist processing such as HDI and RIE, where the photoresist undergoes chemical and physical changes and forms hard sidewalls [5, 6].
Since plasma ashing often leaves residues, a wet strip must follow to complete the stripping process. In many cases, to avoid alkali and heavy metals contamination, the plasma ashing is stopped before the endpoint, and the wafer is transferred to a wet bath [2].
Drawbacks of the wet-stripping method include solution concentrations that change with the number of wafers being stripped, thus affecting stripping quality and throughput; accumulation of contaminants in the baths, which drastically affects yield; and severely corrosive and toxic solutions that impose high handling and disposal costs and create serious safety considerations. Other wet-stripping problems are due to mass transport and surface tension associated with the solutions. For deep submicron technologies, the solutions cannot circulate and tend to accumulate within the patterned structure. This situation is intolerable, as it contaminates the wafer with foreign materials that can lead to drastic yield losses. All of these wet-stripping problems will become even more critical for 300-mm wafers.
 |
Figure 1. Block diagram of the laser-stripping technology. C.D. refers to the catalytic destructor.
The need to perform stripping in a completely dry, single-step process is well established [7-8]. The new laser-stripping method described here combines laser ablation with a plasmaless reactive chemistry. Laser ablation of polymers [9-11], by itself, was unable to completely remove the debris generated during resist stripping. In the new laser-stripping system, reactive chemistry instantly volatilizes the ablation products. A combined etching process removes materials that are not directly ablated, such as residues embedded in the photoresist, sidewalls previously formed, and adhesion promotion and contrast enhancement films (silylation layers).
The laser-stripping process
The laser-stripping process combines chemical-assisted laser ablation/etching, laser-induced chemical etching of sidewall and foreign materials, and enhanced combustion in the gas volume. Reactive species are initially generated by atmospheric pressure process gases, oxygen and nitrous oxide, flowing through a so-called silent electric discharge (gas excitor). More reactive species result from laser-induced photochemical and thermal reactions inside the process chamber. A combination of chemical-assisted laser ablation/etching with laser-induced chemical etching provides efficient stripping at high rates and relatively low temperatures. The exact mechanisms occurring inside the stripping chamber were suggested by Livshits [12-13], and were further studied by the present authors [14].
Laser stripping is a plasmaless process without the damage mechanisms associated with plasma stripping. The removal process also eliminates alkaline and heavy-metal contamination. An earlier plasmaless stripping process based on O3/N2O could be assisted with a UV lamp, but required high substrate temperatures and was unable to strip HDI and contaminants introduced by RIE of polysilicon and aluminum [1-3].
The laser-stripping technology. Figure 1 shows a block diagram of the process. UV photons generated by a KrF excimer laser (248 nm) are introduced to the process chamber via a quartz window. Oxygen and nitrous oxide first pass through silent electric discharges. Molecular O2 is excited in the first gas excitor to produce a mixture of O2 and ozone, O3. A separate gas excitor excites nitrous oxide (N2O) to produce a mixture of N2O, NO, and NO2 gases. Both gas mixtures are then introduced to the process chamber. There, the UV excimer laser irradiates the silicon wafer while the laser-induced reactive gas mixture sweeps over it.
 |
Figure 2. The L-stripper (front panel).
The volatilized products and effluents are evacuated by the fast laminar process gas flow. The catalytic destructor destroys the ozone as well as the additional gases (NxOy), and the nontoxic/nonhazardous gases (CO2, H2O, N2, O2) are vented out. The laser-stripping method is therefore an environmentally friendly process.
The stripping mechanisms. The photoresist strongly absorbs the 248-nm (5-eV) UV photons of the KrF excimer laser. Each laser pulse delivers a peak power of 16 MW corresponding to 5 ? 1017 photons/25 nsec pulse duration. A large number of photoresist polymer bonds are broken during this very short period. This reaction instantly heats the upper layers of the photoresist. Ablation occurs when laser fluence exceeds a threshold value Fth. For novolak/DNQ positive photoresist, Fth is approximately 50 mJ/cm2.
The process gas photolysis and ablation induce significant photothermal decomposition of the process gases, and an extremely high, nonequilibrium concentration of atomic radicals forms in the irradiated gas volume. The oxygen radicals enhance combustion of the ablation products and isotropically etch the foreign materials that defy laser ablation. The resultant stripping/etching process can be described as laser-induced atomic oxygen chemical etching in addition to the main process of laser ablation/etch of polymeric sidewalls and other tough residues.
For the removal of sidewall residues after RIE processes, it is necessary to add halogen (Cl, F) to the process gas. Unlike plasma stripping, where the plasma electrons lead to dissociation of halogen process gas, the dissociation of these molecules in the laser-stripping method is mainly thermal. Therefore, the optimal gas mixture should include halogen-containing molecules with a low-energy dissociation. Nitrogen trifluoride gas (NF3) intensively dissociates at the low temperatures corresponding to laser-induced process gases, and is reactive only during the chemical reactions between laser pulses.
The system. The L-Stripper (Fig. 2) was used to study the laser-stripping process and develop a robust process window that is optimized for high throughput and perfect stripping. Analysis on wafers processed with the tool showed no damage or stripping residues. The tool contains the following subsystems:
a KrF Excimer laser;
the process chambers, the core of the system, in which the stripping process takes place - to increase the machine`s throughput, there are two process chambers that operate alternately;
the optical system, which conveys the beam up from the laser, then shapes and homogenizes the laser pulse to give a uniform beam profile, switches the laser beam to one of the chambers, and scans the entire wafer;
the process gas generator, which supplies the process gases to the process chamber;
the front panel, installed on the wall separating the cleanroom bay and the equipment chase in which the processing unit is installed, and which includes the loading door and an operator interface unit;
the wafer-handling system that includes the cassette holders and the handling robot, which aligns, loads, and unloads the wafers in and out of the process chamber;
vacuum and gas systems, which control the flow and the pressure of the process gases; and
the main vacuum pump.
One operational and maintenance advantage of the system is that laser stripping is a self-cleaning process since the laser radiation and reactive chemistry combination constantly clean the process chamber.
Laser-stripping results
The excimer laser operates at a pulse repetition rate of 250 Hz. A 100 W excimer laser can strip a 1-cm2 area with 1-?m thick photoresist in 0.15 sec . This corresponds to 30 sec/150-mm wafers and 45 sec/200-mm wafers. (Similar stripping rates can be achieved by plasma ashing; however, a wet follow-up strip is necessary to complete the stripping and remove the ashing residues.) Since the throughput is directly proportional to the laser`s average power, one can achieve higher throughput rates by using a more powerful laser system.
We carried out evaluation tests, including microscopic inspection, microanalysis, and electrical tests to verify complete resist removal and to prove that neither metallurgical nor electrical damage had occurred. These results were obtained without subsequent wet processing.
Results for stripping photoresist after HDI. Figure 3 shows a typical Auger Electron Spectroscopy (AES) spectrum following laser stripping of a HDI wafer (BF2+, 4 ? 1015 atoms/cm2, 80 keV). The data corresponds to two areas of a patterned wafer. One area was previously covered with resist and the other area was not covered with resist, but was adjacent to the first area. (Since the two AES spectra were identical, only one spectrum is shown here). Only Si and O are present on the surface as can be expected from the SiO2 on the wafer. If photoresist residues were present, one would expect to find a carbon peak at 270 eV. Neither carbon nor elements typical to resist residues such as sulfur were detected, thus indicating that the resist removal was complete.
 |
Figure 3. Auger spectrum following laser stripping of wafers after HDI (BF2+, 4 ? 1015 atoms/cm2, 80 keV). The vertical scale corresponds to the change in the number of Auger electrons/unit change in kinetic energy.
 |
Figure 4. SEM micrographs after laser stripping: a) after HDI, b) after poly etch, c) after metal etch, and d) after via etch.
Figure 4a, a typical scanning electron microscopy (SEM) micrograph of a laser-stripped HDI wafer, shows no residues on either the area covered with photoresist or in its vicinity. Furthermore, no metallurgical damage could be detected on wafers that were laser stripped within the appropriate process window for HDI wafers.
To obtain ultrasensitive concentrations (~109 atoms/cm2) of heavy metals and alkaline elements on the wafer following laser stripping, stripped wafers were characterized by total reflection x-ray fluorescence and atomic absorption spectroscopy, using vapor-phase decomposition. The level of impurity atoms such as Na, K, Ti, Cr, Mn, Fe, Ni, Cu, Zn, and Sn, is in the range of 5 ? 109 to 5 ? 1010 atoms/cm2. These results meet the current and near future requirements of the industry [15-16].
Results for stripping photoresist after RIE: metal, polysilicon, and via etch. A SEM micrograph (Fig. 4b) from a laser-stripped wafer after polysilicon etch shows no observable residues and thus complete removal of the top photoresist as well as the sidewall polymer. High-resolution scanning Auger nanoprobe analysis verified the results. This microanalysis did not detect any foreign elements on either the sidewall or on the top metallic line. In addition, ellipsometric measurements showed no oxide loss, indicating the high selectivity of the laser-removal method.
Figure 4c shows a SEM micrograph from a laser-stripped wafer after metal etch. As in the previous case, the top photoresist and the sidewall polymer were completely removed and no residues were observed. High-resolution scanning Auger nanoprobe analysis again confirmed the complete removal. Fig 4d shows SEM results after stripping a via-etched wafer (100% overetch into the metal). It also shows the removal of the sidewall polymer in the via hole and the top bulk resist.
Electrical tests. We performed extensive electrical tests (capacitance-voltage [C-V] and current-voltage [I-V] characteristics) to demonstrate that the flat-band voltage of the Si/SiO2 interface and the leakage current of source and drain junctions were not affected by the excimer laser UV radiation [14]. The experiments compared wafers after laser irradiation at conditions that were well above optimal process window to wafers after standard clean - plasma ash followed by piranha wet bath. The results did not show any difference between the laser-stripped wafers and the standard ones, indicating that laser radiation does not introduce any electrical damage on the wafers.
In addition, we compared laser-stripped HDI wafers and wafers stripped by regular plasma ashing, followed by wet processing with sulfuric acid. After stripping, the wafers were processed through all the subsequent production stages of the fabrication site. After completion, we tested the dies on the wafers electrically to look for charging
adiation damage or other failures that could be attributed to the laser-stripping process. The results clearly showed no difference between laser-stripped wafers and wafers stripped by plasma followed by a wet bath. We found no failures, either in terms of the general electrical test parameters, or specific charging
adiation related parameters.
Summary
The laser-stripping process, a combination of UV excimer laser ablation/etching and plasmaless reactive chemistry, provides an efficient, highly selective, single-step, dry-stripping process, with downright elimination of follow-up wet processing. This new processing method can also be used for dry cleaning wafers after various processing stages.n
Acknowledgment
The L-Stripper is a trademark of Oramir Semiconductor Equipment. The authors would like to thank G. Neumann, Y. Nemirovsky (Technion), P. Goldschmidt (Intel), J. Mathuni (Siemens), L.M. Loewenstein (Texas Instruments), D.L. Flamm (University of CA, Berkeley), and T. Francis (Applied Materials) for fruitful discussions; and N. Ariel, S. Polyakov, and J. Waldman for assistance with experiments.
References
1. D.L. Flamm, "Dry Plasma Resist Stripping, Part I," Solid State Technology, pp. 37-39, August 1992.
2. D.L. Flamm, "Dry Plasma Resist Stripping, Part II," Solid State Technology, pp. 43-48, September 1992.
3. D.L. Flamm, "Dry Plasma Resist Stripping, Part III," Solid State Technology, pp. 43-48, October 1992.
4. D.A. Toy, "Choose the Right Process to Strip Your Photoresist," Semiconductor International, pp. 82-87, February 1990.
5. P. Burggraaf, "Resist Implant Problems: Some Solved, Others Not," Semiconductor International, pp. 66-69, June 1992.
6. S. Fujimura, J. Konno, K. Hikazutani, H. Yano, "Ashing of Ion Implanted Resist Layer," Japanese Journal of Applied Physics, Vol. 28, No. 10, pp. 2130-2136, 1989.
7. P. Burggraaf, "What`s Driving Resist Dry Stripping?" Semiconductor International, pp. 61-64, November 1994.
8. P. Singer, "Plasma Ashing Moves into the Mainstream," Semiconductor International, pp. 83-88, August 1996.
9. R. Srinivasan, B. Braren, "Ultraviolet Laser Ablation of Organic Polymers," Chemical Reviews, Vol. 89, pp. 1303-1316, 1989.
10. D.E. Seeger, M.G. Rosenfeld, "Clearing Resist from Alignment Mark Areas Using an Excimer Laser," J. Vac. Sci. Technol., Vol. B6, No. 1, pp. 399-402, 1988.
11. O. Tehar-Zahav, "Photoresist Stripping by Excimer-laser Ablation," MSc Thesis, Hebrew University of Jerusalem, Israel, 1992.
12. B. Livshits, "Excimer Laser for Photoresist Stripping," Oramir Semiconductor Equipment Ltd., Haifa, Israel, unpublished, 1992.
13. B. Livshits, "Laser Stripping Problems Solution," Oramir Semiconductor Equipment Ltd., Haifa, Israel, unpublished, 1993.
14. M. Genut, O. Tehar-Zahav, E. Iskevitch, B. Livshits, "Excimer Laser Photoresist Stripping," Advances in Resist Technology and Processing XIII, Vol. 2724, pp. 601-612, SPIE, 1996.
15. T. Francis, "Controlling Process Equipment Contamination in the 90s," Semiconductor International, pp. 62-66, October 1993.
16. T. Francis, "Not by Particle Count Alone: Getting to the Root Cause of Contamination," Micro, pp. 69-78, July/August 1996.
BORIS LIVSHITS received his PhD degree in theoretical physics in 1963, and his D.Phys./Math Sci. degree in quantum electronics in 1973 in Moscow, and has more than 30 years of experience in chemical and laser physics. Before joining Oramir in 1992, he held a faculty position at the Institute of Chemical Physics, USSR Academy of Sciences. Livshits is presently chief scientist at Oramir, where he has developed the technology for five patents. He is a fellow of the New York Academy of Sciences.
OFER TEHAR-ZAHAV received his MSc degree in physics from the Hebrew University, Israel, in 1992, and is a DSc candidate at the Technion. Before joining Oramir in 1992 as a process engineer, he worked at Intel and Galram on initial experiments for the project described in this article. Tehar-Zahav is presently process development manager at Oramir and the coinventor of five of Oramir`s patents.
ELI ISKEVITCH received his MSc degree in materials engineering from the Technion, Israel, in 1990. Before joining Oramir in 1994 as a process engineer, he worked on superconducting thin films at Synergy, Jerusalem, and at the Technion. He is the coinventor of two of Oramir`s patents.
MENACHEM GENUT received his PhD in materials engineering from the Technion, Israel, in 1988. He is presently VP of R&D at Oramir. Before joining Oramir in 1992 as technology development manager, he worked as a research fellow at Carnegie-Mellon University, Pittsburgh, PA, and at the Weizmann Institute Of Science, Rechovot. Genut has published 18 papers in the field of semiconductor materials and processes, and is the coinventor of five of Oramir`s patents. Oramir Semiconductor Equipment Ltd., PO Box 14, Haifa 31000, Israel; ph 972/4-879-2528, fax 972/4-879-4220, e-mail [email protected].