In this thesis, a distributed feedback resonator (DFB) has been chosen as the laser cavity for most of the work on solid state organic semiconductor lasers because of its low lasing threshold and simple fabrication. Two different approaches are used for fabricating periodic resonators directly on silica substrates, namely e-Beam lithography and holography. Compared to holography, e-beam lithography allows design of high quality gratings with complex geometries, such as circular gratings. However, the overall grating area is normally small and is typically limited to 100 μm by 100 μm with the e-beam system in the department. Larger area patterns from a few millimetres to centimetres are possible only with an extremely high precision stepping motor to stitch together the fields. The whole fabrication is very time consuming and expensive, especially for large area resonators. Holography, on the
other hand, has the advantage of patterning large area gratings at low cost. The size of the resonator is dependent on the size of the laser beam. It is an interference based technique and excludes the complicated equipment required in e-beam lithography, maintaining the simple fabrication and low cost characteristics of organic semiconductors. Holography however offers limited control on the duty cycle and grating geometries. The DFB resonators, in this thesis, are all fabricated by holography to fulfil the requirement for the LED pumped organic semiconductor laser, which will be discussed in details in Chapter4.
The basis of the holographic grating fabrication is two-beam interference, as illustrated in Figure 3.5. A thin film of photoresist is exposed to two mutually coherent laser beams of wavelength λ separated by an angle of 2θ. The resulting in-plane wavevectors of these two beams are counter-propagating and form a standing wave interference pattern with period Λ given by:
sin 2 3-7
Figure 3.5. Holography made by two beam interference
Figure 3.6 outlines the general steps for making a DFB resonator on a silica substrate. First, a thin layer of UV sensitive photoresist is spin coated onto the substrate, followed by a pre-bake to evaporate any remaining solvent. Two beam interference method was used to define the periodic pattern in the photoresist as
described in Figure 3.5. Depending on the properties of the photoresist, a post-bake process could be required after exposure. A specific developer is then used to wash away the unwanted parts of the photoresist. At this stage, the periodic structures are formed in the photoresist and ready to be etched into the substrate. In the etching process, with the patterned photoresist as a mask, Reactive Ion Etching (RIE) is used to transfer the periodic structure into the substrate. To finalize the fabrication process, the remaining photoresist should be washed off in suitable solvents.
Figure 3.6. Steps for making a DFB resonator on silica substrate
The fabrication apparatus is shown in Figure 3.7. The laser source is the CW He-Cd laser with 325 nm output. The diameter of the laser beam is expanded to 16 mm with the use of a beam expander and a collimating lens to achieve a uniform intensity of 0.044 mW/cm2. A silica prism was used to reflect the laser beam into the heart of the interferometer, which comprises a beam splitter, two steering prisms and two rotatable mirrors. Refractive index match oil was used in between the prisms to seal the air gap to reduce reflection loss at each interface. The beam splitter separates the incident laser beam into the two interferometer arms with equal intensity. In each arm, the laser beam experiences two reflections before it gets out. The same numbers of reflections can ensure the output intensities in both arms are identical, which can maximize the contrast in the
interference pattern and thus the grating depth. The two output beams are reflected to the photoresist-coated substrate by a pair of highly reflective dichroic mirrors. The period of the grating can be varied by changing the rotating angle α of the mirrors: ) 2 2 sin( 2 3-8
Figure 3.7. Holography setup
The photoresist used is a positive photoresist SR1805 (Shipley), which is widely used in the fabrication of submicron structures [10-11]. The photoresist solution was diluted 1:1 with EC solvent to reduce the film thickness, which could benefit the etching process for obtaining deep gratings in the substrate. The solution was spin coated onto the silica substrate at 3250 r.p.m. to form a 170 nm thick film. The exposure dose, which is related to the exposure time, is critical and needs to be optimized to fabricate high quality gratings. Therefore, a range of exposure durations were investigated in order to optimize the grating depth and uniformity. The exposed samples were then developed in MF 319 for 2 seconds, followed by a rinse in deionised water. For positive photoresists, the exposed fraction of the
film is removed during developing. Figure 3.8 illustrates the two AFM scans of an optimized and a un-optimized gratings.
Figure 3.8. AFM image of (left) a un-optimized and (right) an optimized grating with period of 325 nm
The exposure time for the un-optimized structure in Figure 3.8 was 9 minutes, while it was 15 minutes for the optimized structure. It is obvious that in the un-optimized sample, the grating grooves are not as uniform as expected. Moreover, with depth in the range from 12 nm to 20 nm, the gratings are very shallow. Once optimized, the grating depth in the right scan can increase to 150 nm with relatively good uniformity. The grating depth is close to the film thickness of the photoresist, indicating that the exposed fraction was completely removed during developing. This is very important in terms of transferring the grating into the substrate. During RIE, the substrate is etched as the patterned photoresist is etched away. The final depth of the grating inside the silica substrate depends on the etching rate of the photoresist and the silica. If the grating depth in the photoresist is very shallow, the worst scenario is that the gratings in the photoresist could be etched away before the ions can reach the silica substrate, leaving no grating structure in the substrate. Even if the ions have the chance to react with the substrate, the grating depth in the substrate will be much smaller than that in the photoresist. Therefore, deeper gratings are preferred in the photoresist.
To perform RIE, the sample was placed inside a chamber of high vacuum which is gradually filled with appropriate gas, CHF3 in the case for silica substrate. The
applied high RF voltage leads to the ionization of the gas. The etching happens when the substrate reacts with the fluorine gas to form a gas form product. Figure 3.9 gives the AFM image of the etched substrate, based on the optimized gratings in Figure 3.8.
Figure 3.9. AFM scan of the etched substrate
After a fifteen-minute etch, the grating still preserved a similar shape as the photoresist gratings. However, the grating depth was reduced by a factor of 3 to 50 nm. The loss of depth could be due to the different etching rates of SR1805 and silica discussed above.