Gold bowtie is made on silicon nitride waveguide based on glass substrate. At first, silicon nitride film is deposited on glass substrate by using plasma-enhanced chemical vapor deposition (PECVD) at temperature of 200°. The film thickness is 300 nm. Then a 240 nm polymethyl methacrylate (PMMA) layer is spun on the sample by using spin coater. The waveguide pattern is defined on the PMMA layer by using electron beam lithography (EBL).
And the waveguide defined on PMMA is transferred to silicon nitride layer by using reactive ion etching (RIE). The residual PMMA layer can be removed by acetone (ACE). This process flow for fabricating the silicon nitride waveguide is illustrated in Fig. 3-1.
Fig. 3-1 The first part of fabrication process.
33
For making gold bowtie structure, PMMA layer is spun on the sample again followed by a conducting layer (ESPACER). This layer can solve the problem of charge accumulation which will make the resultant pattern distorted in EBL process. The pattern for the bowtie is also written by using EBL on PMMA. During the exposure step, we use predefine marks to do alignment process, by which the bowtie pattern can be written right on the waveguide. Actually there is a standard deviation of about 3 μm in alignment process. Therefore we have to make an array of the bowtie to compensate the deviation. After the exposure, the ESPACER layer can be removed by DI water rinsing. Au layer of 30 nm thickness is then deposited at a rate of 0.4 Å /s by thermal evaporation. Then, the PMMA resist and the overlaying Au layer with PMMA underneath can be removed by lift-off process. This is a process conducted conveniently by immersing the sample in ACE at room temperature for 4 hours. Finally there is only bowtie structure laid on the waveguide with no PMMA remaining. Whole this part of process flow is illustrated in Fig. 3-2.
Fig. 3-2 The second part of fabrication process.
34
The above process is implemented from bottom to top. It is intuitive that we make the waveguide first and then put the bowtie on it. However after the waveguide is completed it becomes much easier for the electron to accumulate on the sample surface. And the resultant pattern in EBL process will be distorted severely. Therefore the bowtie fabricated in this process has no sharp apexes as we expect (Fig. 3-3 (a)). In the contrary, we can also reverse the process to do it from top to bottom. That is fabricating the gold bowtie, by EBL and lift-off process, on an intact silicon nitride film first. And then a PMMA layer is spun on the sample followed by EBL and RIE etching to define silicon nitride waveguide on the substrate. This reverse process has several advantages. First, charges can be grounded by the intact silicon nitride film and thus the bowtie can be made with clear shape as what we design without distortion. Fig. 3-3 (b) shows SEM picture of the fabricated device by using the reverse process.
It is obvious that the apexes are sharper than that fabricated by using normal process. Second, we can choose bowtie of appropriate parameters and fabricate waveguide structure underneath it. That will increase workability of the fabricated sample.
Fig. 3-3 The top view SEM pictures of these structure (a) general process (b) reverse process.
We check surface roughness and thickness of the gold bowtie by atomic force microscope (AFM) and P-10 surface profiler. Fig. 3-4 (a) shows top view AFM picture of the bowtie array
35
and white line indicates the cross section view we take as shown in Fig. 3-4 (c). Vertical distances between positions indicated by the red triangles and by the green triangles are measured respectively. We can see the measured thickness of bowtie is very close to 30 nm as what we expect and RMS of surface roughness is very low as only 2.37 nm. The profile shown in Fig. 3-4 (a) seems no gap is there between the two triangles. The gap cannot be resolved by AFM measurement because the picture composes only 256 lines and the line resolution is 7.81 nm. For clarification, we show the SEM picture of the same bowtie array to confirm that the geometry is what we design (Fig. 3-4 (b)). For P-10 surface profiler measurement, we prepare a test sample with a tape sticks on it as mask. Then the tape is removed from the sample surface after Au film deposition. Then we measure the height difference between the region with Au film and the region of bare substrate to get thickness of the film. The average thickness of Au film among four test samples is 30.2 nm.
Fig. 3-4 (a) Top view AFM picture. (b)Top view SEM picture. (c)Line cross section AFM picture (white line in Fig.3-4(a)).
36
After we finish the above fabrication process, it is necessary to make an entrance for coupling light at one end of the waveguide. Generally, we used to cleave the sample to make the waveguide entrance show up at edge of the sample. But there are a lot of difficulties for cleaving the glass substrate because there is no crystallographic orientation on it, and it is too hard to be cleaved. So we use a process of mechanical grinding and polish to make the waveguide entrance appear for light coupling. The grinding sheet is classified by its roughness.
The sheet of lower number is rougher and the sheet of higher number is finer. We use sheet of 80 to initiate the process. It can make the sample edge close to end of the waveguide quickly.
Then we use sheets finer by finer to make the waveguide end appear progressively and remove the damaged region. Finally we use sheet of 4000 to polish the edge for better coupling.
37