Solvent-free plasma removal of etch polymers
03/01/1999
Solvent-free plasma removal of etch polymers
Jean-Francois Christaud, STMicroelectronics, Rousset, France
Wolfgang Helle, Craig Ranft, Cynthia Gapasin, Mattson Technology Inc., Fremont, California
Plasma energy, process chemistry, and wafer temperature are important factors in the removal of etch polymers in dry plasma strip systems. We have demonstrated complete removal of etch polymers using a unique, inductively coupled plasma strip system, which features switchable plasma energy modes and low-temperature process capability. Our results show that polymer removal without the use of wet chemicals or with greatly reduced chemical usage is possible in a production environment.
Polymer formation during etch processes is necessary to control critical dimensions of patterned structures [1]. Removal of these polymers can be difficult, however. In typical IC manufacturing, an oxygen plasma system first strips the photoresist. Then wet-chemical-processing systems remove the etch polymers, a process that often requires strong solvents [2], which are expensive due to high chemical, handling, and disposal costs, and hazardous to the environment.
A more cost-effective alternative to solvent cleaning is dry cleaning using a plasma strip system. The use of fluorine chemistries in standard strip systems has shown good results for the removal of some types of etch polymers such as poly etch and contact etch [3]. Polymer removal using standard strip systems, however, has been limited on other layers such as metal, passivation, and via because the polymers formed during these etches can contain inorganic elements, such as Al, that are not volatile in standard strip gases. At typical strip temperatures, fluorine can also attack the barrier and antireflective coating (ARC) materials (e.g., TiN) associated with these layers. Finally, if photoresist is stripped at standard temperatures before polymer removal (i.e., in situ strip after metal etch), the polymers can become even more difficult to remove, possibly due to oxidation of the inorganics.
Using an innovative, inductively coupled plasma (ICP) strip system manufactured by Mattson Technology, we have shown complete removal of etch polymers without using a solvent clean. Solvent-free cleaning processes have been implemented in production after both metal etch and passivation etch. We have also implemented a via process with significantly reduced solvent usage.
 |
Figure 1. ICPsm Source
Selectable-mode ICP (ICPSM) processing
The ICPSM is a modified version of Mattson`s standard ICP system [4] used for advanced residue removal applications. Figure 1 shows the characteristics of the process chamber. A key feature of the plasma source is the selectable grounded/floating control of the Faraday shield. In the standard ICP strip system, this shield is always grounded, leading to almost complete inductive coupling between the rf coil and the plasma. (The capacitive coupling of power between the coils and plasma is dissipated to ground.) This results in a plasma confined by the inductive field and running at low electron energy. Therefore, no electric field extends down to the wafer and ion generation is extremely low. In the ICPSM source, opening a relay allows the Faraday shield to operate in a floating mode. This leads to increased capacitive coupling to the plasma, which causes a higher electron energy level (more ionization) in the plasma and the electric field extending closer to
the wafer. Figure 2 shows the ion current measured slightly above the wafer when operating in the floating and grounded modes. Note that the increase when going from grounded mode to floating mode is about a factor of 4-5; however, the ion current increase is still much lower than what is typically seen in a high-density plasma etch system. Extra energy can thus be supplied for stubborn back-end-of-line polymer cleaning without causing plasma damage. Plasma damage tests such as antenna test structures, CHARM wafers, and surface charge analysis (surface photo voltage measurements) have shown damage-free results.
An additional feature of this system is its compatibility with fluorine-containing gas chemistries. Fluorine gases such as CF4 can be used in this system without degradation of chamber parts. Finally, a water-cooled wafer platen replaces the standard resistive heater, allowing for lower temperature control (typically 10-90?C) for enhanced selectivity and minimization of inorganic residues.
 |
Figure 2. Ion current measurements obtained using a Langmuir Probe near the wafer surface.
Etch and strip processes
For metal etch, an AlSiCu/TiON barrier stack is etched using a Cl2/BCl3/N2 chemistry in a state-of-the-art metal etcher. Resist is stripped in situ after etch using an on-board strip module. Polymer removal is achieved in the ICPSM system using a reducing gas chemistry containing a mixture of only CF4 and forming gas (N2-4%H2). The wafer temperature is set at 75?C and the ICPSM is run in the floating mode for the entire recipe. After ICPSM treatment, an overflow rinse system rinses the wafers in deionized (DI) water.
For passivation etch, a stack of SiON/PSG is etched in a SF6/CHF3/He blend. The resist and polymers are both removed in the ICPSM system using a two-step O2/CF4 recipe at 75?C. During the first step, the ICPSM is in the floating mode (Faraday shield relay open) for enhanced residue cleaning.
The Faraday shield is then grounded (relay closed) for the second step in which all the remaining resist and residues are cleared. After ICPSM treatment, wafers are DI-water rinsed in an overflow rinse system.
A Lam 4520 oxide etch system etches the vias. These vias are also round-etched to form a "wine-glass" shape. Dielectric materials include tetraethylorthosilicate (TEOS) and undoped silicate glass. Using an O2/CF4 recipe at 75?C, the ICPSM then strips and cleans the wafers. A short solvent clean follows.
 |
Figure 3. Patterened metal lines a) after etch/strip and b) after
Results and discussion
Figure 3 shows patterned metal lines after etch/in situ strip and after ICPSM/DI water rinse. The ICPSM completely removed the heavy polymer on both the sidewalls and tops of the metal lines. We believe the increased ion current at the wafer in the floating mode and the reducing chemistry are key to the success of this process. Experiments using the same process in the lower energy-grounded mode show that polymer removal is not as effective. Also, addition of oxygen in the process degrades polymer cleaning, possibly because of polymer oxidation.
In addition, polymer removal took place without damaging underlying layers. Figure 4 shows that no undercutting of the TiON barrier layer or the underlying TEOS has occurred, thus confirming that high selectivity is achievable at the low wafer temperatures in the ICPSM process chamber.
The metal clean process has been in production for a few months and shows equivalent or better yield vs. the previous solvent process. It has eliminated a costly hydroxylamine-based wet chemical clean step. Also, since the ICPSM process time is short, we can obtain very high throughput (160 wafers/hr).
 |
Figure 4. No undercut of metal lines as seen after ICPsm clean.
 |
Figure 5. The passivation layer after etch.
 |
Figure 6. The pad area is clean after the first step of the ICPsm process (floating mode/higher energy).
Figures 5, 6, and 7 show cleaning results for passivation etch. After etch, heavy polymer is seen on the pad area and sidewalls, in addition to what we believe is an oxide residue on the pad area (Fig. 5). This oxide residue is difficult to remove even with wet chemical treatments. After step 1 of the ICPSM process (floating, high CF4/O2) ratio, the pad area is clean (Fig. 6). Note that this step has good selectivity to resist, which protects the passivation layer from attack. During step 2 (grounded, low CF4/O2 ratio), the photoresist and the remaining polymer on the passivation layer are removed (Fig. 7). Refinement of the two-step process and the CF4/O2 ratios yielded a process-window-optimizing ash rate, passivation layer selectivity, and polymer removal efficiency.
 |
Figure 7. The passivation stack area (in circle) is clean after the second ICOsm process step (grounded/lower)
The passivation cleaning process has been in production for several months and shows equivalent or better yield vs. the solvent clean. It has also been well characterized by design of experiments. Again, the process eliminates an expensive wet chemical step and can obtain high throughput
(95 wafers/hr).
For via clean, the process of record had been a conventional high-temperature (200-290?C) oxygen plasma ash followed by a long solvent treatment. Much cleaner results are obtained when the high-temperature ash is replaced by the low-temperature ICPSM process. This can be attributed to less "hardening" or oxidation of the polymers at low temperature and to the addition of CF4. The solvent treatment time is reduced and the chemical usage is lowered by a factor of two. This process is now currently in production, attaining a throughput of 155 wafers/hr. Moving forward, feasibility has been shown for an all-dry clean process on anisotropic-type vias.
Conclusion
We have shown that dry plasma cleaning of difficult back-end-of-line etch polymers can be achieved in high-volume production. Robust, high-throughput cleaning processes have been designed for metal, passivation, and via etch polymers. Using the ICPSM plasma source, we have shown that increased plasma energy at the wafer enhances polymer removal and that low temperature control is important for obtaining high selectivity. Process chemistry is also important to balance cleaning efficiency, selectivity, and strip rates (throughput). These dry cleaning processes have resulted in significant cost savings over previous chemical treatments. For submicron technology using metal stacks with ARCs and Ti-based barrier layers, these processes, to our knowledge, are the only such processes currently in high-volume production in the world
Acknowledgment
The work discussed in this article was originally published in the proceedings of the Northern California Chapter of the American Vacuum Society (NCCAVS) Plasma Etch Users Group 4th International Workshop on Advanced Plasma Tools and Processing Engineering, May 26-27, 1998. In May 1998, SGS-Thomson announced that it had officially changed its name to STMicroelectronics.
References
1. H. Shan et al., "Effects of Deposition and Ion Scattering on Profile Control in Submicron Etch," J Electrochem. Soc., Vol. 141, p. 2904, 1994.
2. S. Marks et al., in Cleaning Technology in Semiconductor Device Manufacturing IV, ed. R.E. Novak, J. Ruzyllo, PV 95-20, p. 214, The Electrochemical Society Proceedings Series, Pennington, NJ, 1995.
3. D.L. Flamm, "Dry Plasma Resist Stripping Part II: Physical Processes," Solid State Technology, Vol. 35, No. 9, pp. 43-48, 1992.
4. S. Savas, "Advanced Photoresist Strip with a High-pressure ICP Source," Solid State Technology, Vol. 39, No. 10, pp. 123-128, 1996
JEAN-FRANCOIS CHRISTAUD received his BS in physics and chemistry from the Ecole Nationale Superieure de Chimie et Physique, Bordeaux, France. Christaud joined STMicroelectronics in 1994, and, since 1995, has been a process engineer at the Rousset wafer fab, in charge of all dry and wet stripping, residue cleaning, and wet etching. STMicroelectronics, Z.I. de Rousset, BP2, 13106 Rousset Cedex, France; ph 334/4225-8800 ext 8310, fax 334/4225-8868, e-mail jean-francois.
WOLFGANG HELLE received his MA in chemistry from Ludwig Maximilian University of Munich. After serving Lam Research as European process manager, he joined Mattson Technology GmbH, where he is sales and process manager for the entire product line.
CRAIG RANFT received his BS in chemical engineering from the University of California, Santa Barbara. He is the strip and LE product applications manager at Mattson Technology.
CYNTHIA GAPASIN received her BS in materials science from California State University, San Louis Obispo. She is the applications process engineer for strip and LE products at Mattson Technology.