2.5.2 Projection Printing

The introduction of projection printing systems revolutionized the semiconductor industry during the 1970s. The first systems were scanning projection aligners using spherical reflective mirrors for image formation. The invention of all-reflective systems was a real breakthrough because at that time the production of high-quality quartz lenses was hardly feasible and extremely expensive. Nowadays, scanning systems are still in use for technologies with minimum feature sizes down to 1.0 $\mu$m, whereby modern tools use excimer lasers as light source [19]. However, the numerical aperture and thus also the resolution is limited making scanning systems less competitive.

Today's predominant printing technique is a step-and-repeat projection system. A schematic of a typical modern projection system is shown in Figure 2.4. These systems are either all refractive or catadioptric, i.e., a configuration consisting of both lens and mirrors. Usually a reduction is built in. Early systems used 10:1 reduction lenses, 5:1 or 2:1 lenses have become more common due to limitations in mask sizes that can be fabricated with the needed precision. Reducing the image size has the great advantage that the features on the reticle do not have to be as small as the final image and are therefore much easier to fabricate. Another advantage is that mask defects and imperfections are also reduced in size and hence become less severe. Only a small region of the wafer, also called field, is exposed at a time (typically 0.5-3 cm²). Between exposures the wafer must be mechanically moved to the next field, hence the common name steppers.

State of the art I-line steppers including several resolution enhancement techniques are used for the fabrication of 0.35 $\mu$m technology. Excimer-based steppers at a wavelength of 248 nm are already in use for pilot production of 0.25 $\mu$m devices, and optical printing with a wavelength of 193 nm is seriously considered for the 0.18 $\mu$m generation [20]. Hand in hand with the trend towards smaller wavelengths, the numerical aperture of the lens is increased. The technical challenge is to produce a very large lens with little aberration and high transparency in the DUV regime. Here, continual progress has been achieved, too. State of the art lenses have evolved from numerical apertures of 0.28 in 1978, to 0.38 in 1985, to roughly 0.65 today. This trend will continue. Optical designers are vigorously developing systems with numerical apertures as high as 0.8 [13]. To handle the resulting extremely shallow depth of focus, e.g., DOF = 0.3 $\mu$m for $\lambda$ = 193 nm and NA = 0.8, new systems for alignment and focusing have been developed. The first steppers used global alignment and focusing. Modern systems are capable of sight by sight control and automatic adjustment of alignment and focus at every field of the wafer. This also makes the use of large diameter wafers more feasible. The primary disadvantage lies in a reduced throughput that typically lies at 20-50 wafers/hour. A recently proposed method referred to as in-lens filtering or apodization enhances the depth of focus by placing a special amplitude and phase filter in the pupil plane of the stepper [21]. Such a configuration is shown in Figure 2.4. For certain patterns like contact holes the depth of focus can be greatly increased, but only at the expense of reduced peak intensity and increased energy in the sidelobes of the aerial image [22]. However, up to now pupil plane filtering has primarily been of theoretical interest since the pupil plane in microlithographic lenses is usually somewhere inaccessible in the optical path and cannot be reached unless the lens is disassembled. Furthermore different mask types require different filters for optimum performance. Thus the in-lens filter cannot be a simple fixed optical element, which makes this approach hardly practicable for factory floor processes where a high number (up to 25) of different masks are applied during the fabrication process of an IC.