Not applicable.
Not applicable.
Not applicable.
This invention relates to a method for curing epoxy-based photoresist, and a photoresist cured by this method. More particularly, this invention relates to a method and apparatus for curing epoxy-based photoresists more completely, at lower temperatures, and more quickly than the prior art methods.
Microelectromechanical systems (MEMS) are integrated micro devices which may be fabricated using integrated circuit batch processing techniques. MEMS devices have a variety of applications including sensing, controlling and actuating on a micro scale. Accordingly, MEMS devices often include a moveable component such as a sensor or actuator. Often, the sensors or actuators include cantilevered beams which are caused to move by an impulse, such as a current or an acceleration.
For example, MEMS thermal switches are known, wherein one hot beam expands relative to an adjacent cool beam. By coupling the hot beam to the cool beam with a tether, the cool beam is caused to deflect.
Similarly, applying a voltage between terminals 230 and 240 causes heat to be generated in circuit 220, which drives flexor beam 210 in the direction 265 shown in
As mentioned above, circuit 120 may be coupled to flexor beam 110 by dielectric tethers 150, and circuit 220 is coupled to flexor beam 210 by dielectric tethers 250. Accordingly, dielectric tethers 150 and 250 must not only have good insulating properties to keep the current from flowing from circuits 120 and 220 into flexor beams 110 and 210, but dielectric tethers 150 and 250 must also have satisfactory mechanical properties, including good stiffness and elasticity. The stiffness transmits the motion of the circuit 120 or 220 to flexor beams 110 and 210, and the elasticity assures that the flexor beam returns approximately to its original position upon cooling of the circuit 120 and 220.
One dielectric material which is particularly convenient to use in such applications is photoresist, because it is easy to pattern into structures 150 and 250 and it is usually insulating. When cured completely, photoresist may also have satisfactory electrical and mechanical characteristics. However, it is necessary to completely cure the photoresist, in order to convert it from its viscoelastic pre-cured condition to its mechanically rigid cured condition.
Prior art methods for complete curing of photoresist, however, require raising the photoresist to just below the point where it starts to decompose. This procedure can lead to some decomposition of the photoresist, and the cracking or delamination of the other films such as flexor beams 110 and 210 and circuits 120 and 220, as well as dielectric beams 150 and 250, from the substrate.
One prior art method for curing epoxy-based photoresist structures 150 and 250 after patterning and development of the structures 150 and 250 is set forth below. This method may be particularly applicable to the epoxy-based photoresist, SU8:
1) Blanket exposing the photoresist;
2) Partially curing the photoresist by heating at 150 degrees centigrade for 60 minutes (total process time, including ramp up and ramp down is 100 minutes);
3) Releasing the photoresist structures from the substrate by removing any sacrificial layers upon which the photoresist is deposited; and
4) Final curing of the photoresist, carried out at 210 degrees centigrade for 30 minutes (total process time, including ramp up and ramp down is 80 minutes).
Ideally, the temperature for step (4) would be even higher, around 240 degrees centigrade, which is the maximum glass transition temperature for epoxy-based SU8 photoresist for a fully cross-linked sample. However, at these temperatures. widespread cracking may occur in the dielectric films and at the interfaces of various films that are part of the devices, which can lead to shorting of the conductive circuits 120 and 220 to the flexor beams 110 and 210, delamination and other problems.
Accordingly, process temperatures of at least 210 centigrade are required in the prior art process for a duration of at least 30 minutes. The prior art process also requires two heating and cooling steps at graduated temperatures. In terms of effectiveness, the prior art process also cures about 80% of the photoresist film. However, because the film is still partially uncured, it continues to possess some viscoelastic properties, rather than the perfectly elastic properties desired.
Accordingly, a method is desired which leads to more complete curing of the photoresist at lower temperatures and for a shorter time, and with a simpler process than the prior art method.
The properties of epoxy-based photoresists, such as SU8 developed by IBM Corporation of Armonk, N.Y., are such that as the degree of cross-linking of the material increases, it increases the temperature at which free monomers can shift their position within the material in order to be in a better position to react-with other monomers. Therefore, as the degree of crosslinking in the material increases, the temperature required to continue the crosslinking reaction continually increases. Prior art methods for curing SU8 involved two or more time-steps at successively increasing temperatures, while holding those temperatures for at least about 30 minutes in each time step, in order to achieve an acceptable degree of curing.
The systems and methods described here are an optimized method for curing epoxy-based photoresists such as SU8, and the photoresist cured by this process. The method includes heating the photoresist with a continuously varying temperature profile, wherein the temperature of the film is slowly and continuously increased over time. The curing procedure is applied after he photoresist has been placed and patterned on a wafer, and any underlying sacrificial layers have been removed. By ramping the temperature of the wafer and photoresist, molecules in the film are continually heated to maintain mobility, even as they polymerize the film. This continually keeps the crosslinking reaction progressing, even as the degree of crosslinking increases. After the maximum temperature in the heating phase is reached for a period, the photoresist is cooled, so that the exposure of the photoresist to the maximum temperature is reduced or minimized.
Differential Scanning Calorimetry of the epoxy-based photoresist shows that the material is more completely cured after this process than it is after a multiple discrete time-step curing process. In particular, DSC analysis indicates that the photoresist film is about 86% cured with the continuously varying temperature method as compared to the 82% curing percentage obtained using the discrete step method of the prior art.
One embodiment of the systems and methods provides a ramped temperature gradient which may be about 1.6 degrees centigrade per minute for a total ramp time of about 100 minutes, raising the temperature of the photoresist and substrate from about 40 degrees centigrade to about 200 degrees centigrade. The substrate is then held at 200 C for 30 minutes. Accordingly, the maximum temperature reached by the ramp is only 200 degrees centigrade, well under the maximum glass transition temperature of about 240 degrees centigrade for a fully cross-linked sample of SU8. The total processing time is also shorter, about 160 minutes as compared to about 180 minutes using the prior art procedure.
In another embodiment, the continuously varying temperature function may be non-linear, and have a concave upward or convex downward shape, which may further optimize the process in terms of process time and maximum temperature required.
These and other features and advantages are described in, or are apparent from, the following detailed description.
Various exemplary details are described with reference to the following figures, wherein:
The systems and methods described herein may be particularly applicable to microelectromechanical systems (MEMS) thermal switching devices, such as that depicted in
Photoresist, in its cured state, is a crosslinked polymeric material. Commercially available photoresists may consist of the monomers dissolved in a solvent, at varying concentrations. To apply a layer of photoresist to a substrate, a specified quantity of the photoresist solution is poured onto the substrate, which is then rotated at high speed. The rotation throws off any excess material, leaving a uniform, thin layer of the photoresist solution retained on the substrate.
The substrate may now be heated to a predefined temperature to evaporate away the carrier solvent, leaving a layer of photoresist on the surface. Exposing the photoresist layer through a lithographic mask to ultraviolet light, for example, generates photoacids in the photoresist. These photoacids catalyse the crosslinking reaction of the photoresist monomers when the substrate temperature is elevated. This generates areas of linked and unlinked photoresist on the substrate. The regions of linked (or unlinked) photoresist can be dissolved away in a developer to leave a desired pattern of unlinked (or linked) photoresist molecules on the substrate, depending on whether the photoresist is a positive photoresist (or a negative photoresist).
Thus, after patterning and developing the desired structures, it is desired to obtain advantageous mechanical properties from the photoresist regions left behind, and therefore, the photoresist needs to be cured as completely as possible. That is, the monomers in the photoresist need to be encouraged to react among themselves to form a dense network of molecules linked to each other by covalent bonds.
The profile in
In general, the glass transition temperature is coincident with the baking temperature of a sample, that is, as the molecules are heated to a point at which they are free to move, they react with other monomers, becoming cross-linked and thereby increasing the glass transition temperature. For this reason, prior art processes recommended holding the sample temperature at a level near the maximum glass transition temperature, which is about 240 degrees centigrade for a fully cross-linked sample. In contrast, the systems and methods described here gradually raise the temperature of the sample, so that previously immobilized molecules become free to move at the lowest temperature possible. Accordingly, the maximum temperature reached during the continuously varying heating phase, is about 200 degrees centigrade, and is substantially below the maximum glass transition temperature of a fully cross-linked SU8 sample. This provides more complete cross-linking, as described in more detail below, as well as minimizes the exposure time of the sample to high temperatures, which may otherwise damage the films by cracking or decomposition. Furthermore, as shown in
The photoresist may be held at a maximum temperature for a period of time after which a cooling phase may be applied. For example, the photoresist may be held at the 200 degree centigrade temperature for at least about 15 minutes, and more preferably, about 30 minutes, before the cooling phase is applied. The cooling phase may also be continuously varying, as shown in
The method begins in step S100, and proceeds to step S200, wherein photoresist solution is spun onto the surface of the substrate. The photoresist may be, for example, SU8. In step S300, the photoresist is soft baked to evaporate the solvent from the photoresist solution. The evaporation of the solvent may result in a photoresist film in a thickness of, for example, about 13 μm to obtain structures 150 and 250. In step S400, the photoresist is exposed through a mask which may be patterned according to the structures 150 and 250 to be formed on the thermal switch 10. The exposure may generate photoacids which catalyze the cross-linking reaction in the photoresist. In step S500, the photoresist is baked again, to cross-link the exposed portions of the photoresist. This post-exposure bake step S500 may include heating the photoresist to a temperature of about 95 degrees centigrade for about 5 minutes. After the post-exposure bake, the photoresist may be developed in step S600. If the photoresist is a positive photoresist, the exposed portions of the photoresist are dissolved in developer in step S600. If the photoresist is a negative photoresist, the unexposed portions are dissolved in developer in step S600. This step leaves only the desired structures of photoresist, which are then cured as completely as possible. SU8 is a negative photoresist, and the developer solvent for SU8 may be, for example, ethyl lactate or diacetone alcohol. After developing, the photoresist and substrate may be rinsed with acetone to removed any residual organic solvent.
To complete the curing, the remaining photoresist is blanket exposed in step S700, which is by application of broad spectrum illumination, for example, the I-line and G-line radiation from a mercury lamp. Here again, the term “blanket exposure,” should be understood to mean that the radiation is not transmitted through a lithographic mask, but is instead allowed to illuminate the entire surface of the photoresist and substrate. The photoresist is then baked once again to complete its curing in step S800, using, for example, the ramp curing method illustrated in
After curing the remaining photoresist, the film may be analyzed using a differential scanning calorimeter, or DSC. This step is shown as step S900 in
The differential scanning calorimeter device is known in the art, as a device used for among other things, understanding the curing reactions in thermosetting polymers. To use the differential scanning calorimeter, a small polymer sample in a hermetically sealed pan may be taken through a closely controlled, programmable temperature sequence, during which the heat output or input required to take the sample through the temperature sequence is measured. A typical temperature ramp for the differential scanning calorimeter may be, for example, from 0 degrees centigrade to about 300 degrees centigrade at a rate of about 5 degrees centigrade per minute. Exothermic or endothermic reactions in the sample appear as peaks and valleys in the energy input measurement as a function of temperature, respectively.
The differential scanning calorimeter therefore may monitor the heat produced in a sample at a given temperature. Since SU8 curing is known to be an exothermic reaction, the amount of heat produced may be indicative of the relative amount of incompletely cured photoresist in the film. Therefore, the lower the total energy output during the temperature sequence, the more stable or completely cured the SU8 sample is likely to be.
In addition to more complete curing of the photoresist, the ramp curing method has several additional advantages over the prior art method. The maximum temperature to which the sample is exposed is somewhat reduced, from 210 degrees centigrade to only about 200 degrees centigrade. This temperature, 200 degrees centigrade, is substantially below the maximum glass transition temperature of a fully cross-linked sample, about 240 degrees centigrade. By reducing the maximum temperature to which the films are exposed, problems with cracking and delamination of the films nay be reduced. A slow ramp up and ramp down of the temperature also reduces damage to the substrate from the thermal shock that may be caused if the temperature is raised more quickly. Therefore, the devices cured using the continuously varying temperature profile may have a higher yield than devices produced using the prior art method, and a lower scrap rate.
In addition, the processing time is shortened, from about 180 minutes total for the prior art method to about 160 minutes total for the continuously varying temperature profile method. This may increase production speed and efficiency by reducing the process time required to produce the photoresist structures 150 and 250, for example.
Finally, the single step curing of the continuously varying temperature profile method significantly simplifies the process flow used to make the structures 150 and 250, as compared to the multi-step process of the prior art shown in
Another feature of the systems and methods described here is the use of broad spectrum illumination for the blanket expose step S700 in
Another aspect of the systems and methods, is the use of a convection oven, rather than -a localized heater such as a hot plate, for example, to heat the photoresist and substrate. A convection oven provides more uniform heating of the photoresist and substrate, such that the temperature profile throughout the film is more consistent and uniform. If a hot plate, for example, is used instead, it is likely that the portions of the photoresist in closer proximity to the hot plate will be heated to a hotter temperature than portions of the film further away from the hot plate. Therefore, the portions of the photoresist closer to the hot plate may begin to decompose, whereas the portions further away from the hot plate may not be fully cured. Thus, the use of a convection oven may also improve the uniformity and consistency of the resulting photoresist structures 150 and 250. Accordingly, the means for heating the photoresist to a maximum temperature with a continuously varying temperature profile may be a convection oven.
A number of alternative embodiments of the linear ramp profile shown in
Also shown in
Using the continuously varying temperature profiles of
It should be understood that the cooling phase of the methods shown in
While various details have been described in conjunction with the exemplary implementations outlined above, various alternatives, modifications, variations, improvements, and/or substantial equivalents, whether known or that are or may be presently unforeseen, may become apparent upon reviewing the foregoing disclosure. For example, not all of the steps of the method shown in