Abstract

Most electron-beam microfabrication systems now in existence have been built ’in house’ for the production of special masks or devices on a relatively small scale. The fabrication of a variety of submicron structures by such systems is a well documented subject. However, the bulk of present day and foreseeable-future output from the semiconductor industry is based on devices with linewidths in excess of 2 μm. This is a function of the limitations of established processing techniques as much as the limitations of photolithography. It is proposed that a versatile electron-beam system can offer significant overall advantages in many applications in this field. A prime application is the saving of time in the proof of new complex circuit designs by using an electron-beam system as a 1× pattern generator so that the first set of masks for a new design can be made directly from a CAD system tape in one day. Such masks would be free from the repetitive defects and misalignments exhibited by the present optical-pattern-generator–image-repeater combination. It would be possible to include design variations on each layer without the intolerable time penalty incurred by making and changing optical reticles. Design optimization should therefore be possible with the minimum of mask remaking. In some circumstances, direct exposure of a single wafer could provide even faster availability of proving circuits, while eliminating the yield loss inherent in mask-alignment problems. A specification for an electron-beam system capable of performing the above functions should include a minimum practical linewidth of 2 μm, a maximum chip area of 10×10 mm and a maximum mask or wafer size of 100×100 mm. The system must be capable of exposing 25% of the maximum-size workpiece in under two hours. There is a requirement for automatic electronic registration on a ’per chip’ basis for both mask and wafer exposure, and for laser-interferometer stage monitoring to permit assembly of subfields if chip size and resolution requirements go beyond the capabilities of the basic system. The maximum beam current should be 2×10−7A at 15 keV, stable to ±2% over 1 h; electron energies should be 10, 15, 20, and 25 keV. The total image distortion (including misalignment of beam to target) should not exceed ±0.2 μm within each electronically scanned subfield of 2.5×2.5 mm. The system must accept standard inputs as presently used for optical pattern generators. Such a system could completely replace the present optical-pattern-generator–image-repeater combination and would satisfy the growing demand for near-perfect masks. It would also be able to directly expose devices whose dimensions and complexity produce alignment problems which even perfect masks could not resolve. Present device processing can achieve far more in terms of better yield and throughput by improving dimensional control on a layer-to-layer basis than it can by higher resolution lithography. The potential for electron-beam pattern generators to outperform existing optical image repeaters, contact printers, and projection printers in this respect appears to be the key to their wider use in the semiconductor industry.

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call