通过超柔顺探针对光刻胶进行无掩模、亚微米热化学图案化

This article introduces a scanning probe lithography technique in which ultracompliant thermal probes are used in the selective thermochemical patterning of commercially available photoresist. The micromachined single-probe and multiprobe arrays include a thin-fifilm metal resistive heater and sensor sandwiched between two layers of polyimide. The low spring constant s , 0.1 N/md and high thermal isolation provided by the polyimide shank is suitable for contact mode scanning across soft resists without force feedback control. The probes provide what is effectively a spatially localized postexposure bake that crosslinks the photoresist in the desired pattern, rendering it insoluble in developer. For 450-nm–1400-nm-thick AZ5214E (Clariant Corp.), line and dot features with sizes of 450 –1800 nm can be printed using probe powers of 13.5–18 mW, and durations of 1–60 s per pixel. Variation of feature sizes with process parameters is described.

With the increasing costs of photomasks—often in excess of one million dollars for state-of-the-art complementary metal-oxide-semiconductor processes—alternatives to optical lithography have been sought, particularly for lowvolume manufacturing and prototyping in conventional processes. As a maskless lithography approach, scanning probe lithography (SPL) offers the promise of high spatial resolution (which is not limited by the diffraction of light) and, for low-volume applications, substantial savings in mask and equipment costs.

It is interesting to note that as methods in scanning probe microscopy developed over the last few decades, most have been applied to lithography.1 Years after the invention of the scanning tunneling microscope (STM) in the 1960s, scanning probe lithography was born when an STM was later used to selectively oxidize hydrogen-terminated Si surfaces with nanometer resolution.2 Likewise, the atomic force microscope (AFM), originally used for mapping topography, was later modifified to “scratch” thin metals3 and soft polymers,4 again with resolution in the tens of nanometers. The near-fifield scanning optical microscope, which employs scanning probe tips to focus light, was used directly to optically pattern photoresist with resolutions beyond the far-field diffraction limit.5 Several other lithographic techniques have grown out of AFM and STM technology, two of which are electrostatic6 and dip-pen lithography.

Scanning thermal microscopy, introduced in 1986, has also found complementary efforts in lithography, but nearly all are based on thermomechanical indentation, that is, the use of heat and pressure in submicron tips to etch pits into polymers with glass transition points. Initally, heat was provided by an external laser pointed at a conventional AFM tip,8 and later by thin-film heaters embedded in the cantilever.9 Thermomechanical indentation has been pursued with interest in ultra-high-density data storage, the most notable device being the IBM “millipede.”10 In this work, we demonstrate that scanning thermal probes can be applied towards lithography by thermally catalyzing a chemical reaction in a submicron region (Fig. 1) rather than by thermomechanical indentation. We have demonstrated this principle on the positive tone photoresist AZ5214E (Clariant Corp.) in a process similar to image reversal. The use of localized heat to pattern this resist has been previously explored by focusing semiconductor lasers.11 Some potential advantages of the proposed method are that it is less prone to tip wear and produces less debris when compared to purely mechanical SPL techniques; however, it does require scanning over soft photoresist. An enabling device for this technique has been the ultracompliant probe technology, reported previously by our group,12,13 which minimizes damage to both the sample and the probe tip. Both single probes and eight-probe arrays are shown to provide thermochemical patterning of photoresist.

Fig1

Temperature calibration of the single probe is relatively straightforward because the fractional change in the probe resistances DR/Rd is directly proportional to the increase in tip temperatures D Td by a constant K. 12 The thin-film heater accounts for the majority of the probe’s resistance and results in a linear relationship between the measured resistance and tip temperature. To determine the constant K, the singleprobe cantilever was immersed in water and the current was ramped until bubble formation and rapid evaporation was observed. Water was found to boil at a 2.16% increase in probe resistance, and by correlating this to a 75 °C increase in tip temperature, K was found to be 288 ppm/ °C. Repeating the same experiment with isopropyl alcohol (boiling point 85 °C) resulted in a virtually identical figure. Using the above calibration, the tip temperature can be correlated to the measured resistance during probe operation. The probes can be heated to approximately 300 °C before high temperatures begin to damage the cantilever structures. It should be noted that the temperature versus power relationship differs depending on the degree of thermal isolation from the surrounding environment. A probe operating in proximity or in contact with the substrate requires about 30% more power than when suspended in air due to the heat loss to the substrate. Nevertheless, the typical input powers needed to reach temperatures in excess of 250 °C while in contact was , 20 mW.

To fully crosslink, AZ5214E requires temperatures of about 90 °C, but it was found that simply biasing the probe tip at this temperature was inadequate due to the thermal contact resistance between the tip and the photoresist. Except where indicated, the single probe was biased at 18 mW, corresponding to a tip temperature of 275 °C. In addition, the substrate was biased at 45 °C using a hot plate in order to further reduce the amount of heat that the probe needed to provide. Tests showed that when the contact force was made excessively high by flexing the cantilever towards the substrate, the increased pressure in combination with the high tip temperature resulted in thermomechanical indentations penetrating all the way through the resist, and no thermal patterns were observed after development. Conversely, weak contact forces resulted in poor thermal contact, also resulting in no pattern formation. An intermediate contact force of about 480 nN was found to be successful. The advantage of using flexible probes is again noted here in that once the probe is biased at some contact force, that force does not change signifificantly over the sample topography, thereby increasing the uniformity of the patterned structures.