Optical Lithography Forever...
Ruth DeJule, Associate Editor -- Semiconductor International, 9/1/1999
157 nm may be the most difficult optical technique; however, it is the least difficult post 193 nm technology,' stated keynote speaker Will Conley of Motorola-APRDL at the FSI Lithography Breakfast Forum in San Francisco. With enhancements, Conley and his co-authors anticipate the extension of optical lithography to 50 nm. This is shown in the Optical Extension Roadmap (see Table) that consists of extrapolated data from 248 nm experiments and simulations. Indicated in the Table is a numerical aperature (NA) of 0.9, clearly a hypothetical leap of faith, but it is intended to demonstrate the extremes of optical lithography needed to achieve these linewidths.
|
Optical Extension Roadmap | |||||||||||||
| Wavelength Feature Duty Cycle | 248 nm | 193 nm |
157 nm | ||||||||||
| 1:3 | 1:1 | 1:2 | 1:3 | 1:1 | 1:2 | 1:3 | 1:1 | 1:2 | |||||
| NA | 0.53 | 70 | 140 | 187 | 55 | 109 | 146 | 44 |
89 |
118 | |||
| 0.57 | 65 | 131 | 174 | 51 | 102 | 135 | 41 | 83 | 110 | ||||
| 0.60 | 62 | 124 | 165 | 48 | 97 | 129 | 39 |
79 |
105 | ||||
| 0.63 | 59 | 118 | 157 | 46 | 92 | 123 | 37 | 75 | 100 | ||||
| 0.70 | 53 | 106 | 142 | 41 | 83 | 110 | 34 | 67 | 90 | ||||
| 0.80 | 47 | 93 | 124 | 36 | 72 | 97 | 29 | 59 | 79 | ||||
| 0.90 | 41 | 83 | 110 | 32 | 64 | 86 | 26 | 52 | 70 | ||||
| factor | 0.25 | 0.50 | 0.33 | 0.25 | 0.50 | 0.33 | 0.25 | 0.50 | 0.33 | ||||
| nPitch (Ideal) | 0.5 | 0.5 | 1.0 | 0.5 | 0.5 | 1.0 | 0.5 | 0.5 | 1.0 | ||||
| nPitch (Full field) | 0.6 | 0.6 | 1.2 | 0.6 | 0.6 | 1.2 | 0.6 | 0.6 | 1.2 | ||||
| CH: contact hole | nPitch: Normalized pitch | ||||||||||||
| Full field: assumes 20% loss of workable resolution due to aberrations | |||||||||||||
| Blue: 130 nm node | Black 100 nm node | Gray: 70 nm node | Red: 50 nm node | ||||||||||
| SOURCE: Peterson Advanced Lithography | |||||||||||||
Though most stepper manufacturers are not as ambitious, all foresee optical technology at the 100 nm node. To achieve this, enhancements are being developed at all levels: the resist, mask and lithography tool. For the exposure tool, costs can be divided into actual system cost and cost of ownership (CoO). The optical system and light source are the most dominant CoO factors. Maintenance of the KrF laser source comprises 30% of CoO in DUV systems; and of this figure, 60% goes into maintaining the laser chamber, line-narrowing and monitor module. Laser source manufacturers like Cymer, Komatsu and Lambda Physik are addressing these issues. Komatsu, for example, has reduced CoO of its G20K KrF laser source by a factor of 2 since introduction of its previous model G10K. With a new design and new materials that have effectively prolonged the chamber life, Komatsu realized a reduction in chamber maintenance of 40% over the previous-generation laser.
A major challenge for optical lithography is the mask error factor (MEF). MEF is due to the loss of aerial image contrast, which increases process sensitivity in three major areas: variations from the mask, lens aberrations (such as focus, spherical, coma, astigmatism), and interference effects from light impinging on the resist-substrate interface from all angles defined by NA and from each diffraction pattern. It is the natural result of a low-quality aerial image that occurs when the exposure wavelength is larger than the feature size. 'For sub-wavelength size features, increasing the image contrast and tuning the resist for optimal response to that image is the only real way to reduce MEF,' said John Petersen of Petersen Advanced Lithography. Reducing MEF is typically accomplished through wavefront engineering techniques such as advanced lithography masks and exposure tool features such as off axis illumination to decrease or eliminate zero order light and possibly, in the future, pupil filters. However, shorter wavelengths, higher NA (the system's light gathering capability), partial coherence (sigma, how much light is entering the lens system) and reduced lens aberrations also will reduce MEF. But all come at a cost. For steppers, the primary cost drivers are NA and field size because both require larger, aberration-free lenses, according to Eric Johnson, vice president of technology at Nikon. The cost of the lens is a function of the cube of the NA, the cube of the field size, the square of aberration control and a linear function of distortion control.
Going to 157 nm lithography is likely to incur considerable expense. A vacuum or purged environment is likely, as well as an all CaF2, catadioptric lens system which combines both reflective and refractive elements. The157 nm lenses will require an estimated 3X improvement over 193 nm lenses, in terms of glass grade and performance, and the NA entry level for 157 nm is an extremely high 0.70, 0.75. But to Ultratech Stepper's Doug Anberg and most others in the industry, those costs are still easier to cope with, rather than tackling a quantum change to an NGL technology, which has an entirely new set of problems. Conferences like Lambda Physik's Second International UV Laser Symposium for 157 nm Applications in October help pave the way.
