Specific Process Knowledge/Lithography/Coaters: Difference between revisions
| Line 249: | Line 249: | ||
*Softbake (contact or proximity) | *Softbake (contact or proximity) | ||
After priming. the wafer is centered on the coater chuck and held in place by vacuum, or in some cases pins. If static dispense is used, the wafer remains static during the ensuing resist dispense. In the case of dynamic dispense, the wafer rotates at low spin speed during the dispense. Using too high spin speed during dispense can cause surface wetting issues, while a too low spin speed causes the resist to flow onto the backside of the wafer. After dispense, a short spin at low spin speed may be used in order to spread the resist over the wafer surface before spin-off. | |||
===Spin-off=== | ===Spin-off=== | ||
The spin-off cycle determines the thickness of the resist coating. | The spin-off cycle determines the thickness of the resist coating. For a given resist, the thickness is primarily a function of the spin-off speed and the spin-off time, both following an inverse power-law (y=k*x^-a). The acceleration to the spin-off speed also influences the thickness, but the effect is dependent on previous steps. The spin-off is usually a simple spin at one speed, but it may be comprised of several steps at different spin speeds. After spin-off, the wafer is decelerated. | ||
The coated thickness, t, as a function of the spin-off speed, w, follows an inverse power-law, t = k * w<sup>-a</sup>. The constant, k, is a function of the resist viscosity and solid content, and the spin-off time. The exponent, a, is dependent on solvent evaporation, and is typically ~½ for UV resists. This means that from the thickness t<sub>1</sub> achieved at spin speed w<sub>1</sub>, one can estimate the spin speed w<sub>2</sub> needed to achieve thickness t<sub>2</sub> using the relation: <br> t<sub>1</sub>*w<sub>1</sub><sup>½</sup> = t<sub>2</sub>*w<sub>2</sub><sup>½</sup> => w<sub>2</sub> = w<sub>1</sub> * t<sub>1</sub><sup>2</sup>/t<sub>2</sub><sup>2</sup>. <br> For thick SU-8, however, a is observed to be ~1 (probably due to the low solvent content and/or the formation of skin). In this case, the relation simply becomes: <br> t<sub>1</sub>*w<sub>1</sub> = t<sub>2</sub>*w<sub>2</sub> => w<sub>2</sub> = w<sub>1</sub> * t<sub>1</sub>/t<sub>2</sub>. <br> | The coated thickness, t, as a function of the spin-off speed, w, follows an inverse power-law, t = k * w<sup>-a</sup>. The constant, k, is a function of the resist viscosity and solid content, and the spin-off time. The exponent, a, is dependent on solvent evaporation, and is typically ~½ for UV resists. This means that from the thickness t<sub>1</sub> achieved at spin speed w<sub>1</sub>, one can estimate the spin speed w<sub>2</sub> needed to achieve thickness t<sub>2</sub> using the relation: <br> t<sub>1</sub>*w<sub>1</sub><sup>½</sup> = t<sub>2</sub>*w<sub>2</sub><sup>½</sup> => w<sub>2</sub> = w<sub>1</sub> * t<sub>1</sub><sup>2</sup>/t<sub>2</sub><sup>2</sup>. <br> For thick SU-8, however, a is observed to be ~1 (probably due to the low solvent content and/or the formation of skin). In this case, the relation simply becomes: <br> t<sub>1</sub>*w<sub>1</sub> = t<sub>2</sub>*w<sub>2</sub> => w<sub>2</sub> = w<sub>1</sub> * t<sub>1</sub>/t<sub>2</sub>. <br> | ||
===Backside rinse=== | |||
Dependent on the spin speeds used in the various steps of the spin coating, resist may creep over the edge of the wafer and onto the backside. Also, some resists tend to leave fine strings of resist protruding from the edge of the wafer, or folded onto the backside, an effect sometimes referred to as "cotton candy". This resist will contaminate the soft bake hotplate, and thus subsequent wafers with resist. In a backside rinse step, solvent administered through a nozzle to the backside of the wafer while spinning at low or medium spin speed dissolves the resist and washes it away. After the rinse, a short spin at medium spin speed dries the wafer before the soft bake. During the backside rinse solvent inevitably creeps onto the front side of the wafer. This effect may be used to dissolve and subsequently remove an edge-bead, but it may also leave the rim of the wafer exposed. As an alternative to backside rinse, a wafer which is left dirty on the backside by the spin coat process may be soft baked in proximity in order to protect the hotplate from contamination. This leaves front side coating intact, but also leaves the backside dirty. | |||
===Edge bead=== | ===Edge bead=== | ||