Charles L. Standley

Sputter-etching of heterogeneous surfaces

In conventional sputter etching, heterogeneous surfaces are eroded at generally unpredictable rates. The reasons for this are discussed and a solution to the problem is given: Based on control of redeposition, the technique involves the use of a device called a “catcher,” which is placed near the target of the sputtering chamber to trap re-emitted particles. Experiments are described which confirm the effectiveness of the approach. Introduction The technique of rf sputter-etching[ 11 is now widely used. Its chief attraction is that it will cause the erosion of materials without regard to whether they are insulating or conducting. The rf sputter-etch rates for a wide range of different materials are relatively close (within a factor of 10) so that many of the restrictions associated with chemical etching are not apparent. Furthermore, the sharpness of the edge that is eroded away is limited only by the sharpness of the mask since no undercutting can take place. Masks of several conventional photoresist materials have been successfully used for this purpose. To implement the technique, the substrate that is to be sputter-etched is made a part of the target of an rf sputtering system. Then, when the rf glow discharge is initiated, the entire target (including the substrate assembly) is eroded away at some rate dependent on a number of parameters such as rf power, gas pressure, etc. Most workers have found that it is good practice to make the body that is to be etched and the target on which it is placed of the same material, if possible, and relative rates for the sputter-etching of various materials have been measured [ 1 ] for such an arrangement. Several workers have noted that when the surface to be sputter-etched consists of more than one material, particularly if each has a markedly different sputtering rate, the resultant rate for the erosion of the entire surface cannot be readily predicted [2]. Thus, to construct a hypothetical situation, if material A is known to sputter at a different rate from material B, and a pattern is prepared that consists of alternating lines of A and B, it will be found that neither A nor B etches at the relative rates determined using surfaces of pure A and pure B. Further investigation would show that the actual rates observed depend on the type of pattern used as well as on the various puttering parameters employed. The purpose of this paper is to present an explanation of this phenomenon and to indicate how it can be avoided. Discussion It is now firmly established that during rf sputtering a significant fraction of the deposited material is re-emitted during the entire sputtering cycle[3]. The exact amount of this re-emission is a complex function of the relative areas of the target, the substrate assembly, and the chamber walls as well as of the rf and dc coupling between these respective surfaces[4]. This re-emission can have several causes, the principal ones being: 1 ) Resputtering of the deposited material as the result of bombardment by energetic ions from the discharge[4]; 2) resputtering due to bombardment by energetic neutrals present in the sputtering gas [ 3 51 ; 3) resputtering due to energetic negative ions which originate at the target surface and are accelerated across the Crookes dark space[4]; and 4) thermal re-emission [ 61. It has been found that even if re-emission due to mechanism 1 is largely eliminated (by seeing to it that no significant potential difference exists between the depositing film and the rf plasma), re-emission due to the other three causes is still generally around 30%[3]. Thus, in any rf sputter-etching system, it is possible that as much as 30% 67 JANUARY 1972 SPUTTER-ETCHING Figure 1 Sketch of the catcher. The backside is a flat plate. Figure 2 Sketch of sputtering chamber showing position and relative size of catcher.