Sub-micron integrated grating couplers for singlemode planar optical waveguides

Size: px

Start display at page:

Download "Sub-micron integrated grating couplers for singlemode planar optical waveguides"

Rodger Matthews
6 years ago
Views:

1 Sub-micron integrated grating couplers for singlemode planar optical waveguides Colin M. Hayes, Marcelo B. Pereira, Baylor C. Brangers, Mustafa M. Aslan, Rodrigo S. Wiederkehr, and Sergio B. Mendes Department of Physics and Astronomy University of Louisville Louisville, KY Joseph H. Lake Department of Electrical and Computer Engineering University of Louisville Louisville, KY Abstract The fabrication of integrated grating couplers for planar optical waveguides is presented. Holographic gratings with spatial period of nm are produced in photoresist films using a Lloyd s mirror setup. The periodic modulations of the gratings are transferred to soda-lime glass and fused silica substrates through deep reactive-ion etching with a depth of approximately 70 nm. A single-mode waveguide (in the visible region) is created by depositing a film of Al 2 O 3 (180 nm) using an atomic layer deposition process. Characterization of several steps of the device fabrication was done by spectophotometry, atomic force microscope, and measurements of diffraction efficiency. Keywords-Waveguide grating coupler, diffractive components, surface-relief grating, inegrated optics, optical waveguide, holographic diffraction grating, ATR spectroscopy. I. INTRODUCTION Periodic modulation of surfaces and thin-films play an important role in several photonic and opto-electronic devices; they are critical elements present in distributed feedback (DFB) and distributed Bragg resonator (DBR) semiconductor lasers, in wavelength division multiplexing (WDM) add-drop filters for telecommunications, and in surface plasmon and optical waveguide devices for coupling free-space light beams into surface electromagnetic waves. Surface waves are valuable spectroscopic tools for highly sensitive investigations of interface phenomena and molecular thin-films including biological assemblies at the sub-monolayer levels - knowledge needed for elucidating biointerfacial processes, developing biomaterials for medical implants, creating new biomedical diagnostics and high throughput screening protocols for developing new pharmaceutical drugs [6] [7]. In this paper, we report our work in developing sub-micron integrated grating couplers for single-mode planar optical waveguides on planar glass substrates. The fabrication steps of these devices include surface cleaning of sample slides, spincoating of photoresist film, holographic exposure, photomask development, reactive ion-etching, and vacuum deposition of a transparent thin-film optical waveguide. All of these steps are discussed in detail below. Where applicable, theory will be discussed as well as any experimentation involved and our findings. II. FABRICATION The fabrication of single-mode planar optical waveguides and their integrated couplers is a detailed, multi-step process. Every step has many parameters, all of which affect the final device performance. In this section, we will detail each step, the fabrication parameters involved, and any calibration steps taken. A. Glass Surface Cleaning Slides of soda-lime glass are first inspected for defects, engraved with an identification number, and transported to the University of Louisville Micro/Nano Clean Room Facility. The slides are immersed in a 2% solution of surfactant (Microdetergent) dissolved in DI water and sonicated for 10 minutes at 60 C. The slides are then scrubbed with cotton and sonicated again. The scrubbing and sonication is repeated once again to improve the removal of surface contaminants before the slides are rinsed with DI water. The slides are placed in a Nano-strip bath (90% Sulfuric Acid (H 2 SO 4 ), 5% Peroxymonosulfuric Acid (H 2 SO 5 ), 5% DI Water, and less than 1% Hydrogen Peroxide (H 2 O 2 )) for 5 minutes at 90 C and thoroughly rinsed with DI water. To aid in the drying process, the slides are sonicated in methanol for 7 minutes at 70 C. Following this, the slides are blown dry with nitrogen and further dried in an oven for 1 hour at 110 C and allowed to cool back to room temperature for 30 minutes. B. Spin-Coating of Photoresist Film To optically write nanostructures on the glass substrates, we must apply a layer of photoresist. An adhesion promoter layer of hexamethyldisilazane (HMDS) is first applied to the slides by spinning at 4500 RPM for 30 seconds. The HMDS layer turns the hydrophilic soda-lime glass slide into a hydrophobic surface, allowing the hydrophobic photoresist to bond with the surface more easily. The application of the HMDS layer yields a great improvement in the quality of the applied photoresist layer. Without the application of the HMDS, defects in the NIH grant RR

photoresist are common and our observation is that the HMDS almost eliminates all defects in the photoresist. Next, a 1:1 solution of Shipley 1805 photoresist and J.T.

2 photoresist are common and our observation is that the HMDS almost eliminates all defects in the photoresist. Next, a 1:1 solution of Shipley 1805 photoresist and J.T. Baker BTS-220 thinner is spin-coated onto the slides at 4500 RPM for 30 seconds and baked for 30 minutes at 92 C. The speed for the spin-coating is a parameter that required calibration. We have applied photoresist at spin speeds from 2000 RPM to 6000 RPM and have found that the film thickness decreases from around 250 nm to around 160 nm. Fig. 1 shows a graphical representation of this trend. The film thickness has been determined using the method discussed in [2], which is summarized in section III C. When the interferometer is mounted on a rotation stage, we are given high-precision control over the spatial frequency of the interference pattern. Using this setup, we can easily generate patterns with a pitch between 850 nm and 230 nm using the 442 nm beam. If we use the other line from the He- Cd laser at 325 nm, we can achieve pitch sizes between 630 nm and 170 nm. Slides are typically exposed for 5 seconds, flipped, and exposed again to create two half-circle diffraction gratings that are 34 mm apart. This setup is based upon that found in [1]. Fig. 2 shows the actual setup used and Fig. 3 shows a schematic of it. Photoresist Spinning 250 Film Thickness (nm) Spin Speed (RPM) Figure 2. Picture of holographic exposure setup Figure 1. Correlation between spin-coat speed and film thickness C. Holographic Exposure In preparation for holographic exposure, the glass slides are coated with a black strippable paint, Universal Photonics s X- 59 Strip Coating, on the back (non-photoresist) side to remove unwanted reflection from the rear glass/air interface. The gratings are created with a holographic interference pattern using a Kimmon IK5651R-G 442-nm He-Cd laser. This beam is passed through a 40X microscope objective placed a distance equal to its focal length from a 10 µm pinhole. The resulting diffraction pattern is filtered with a 16 mm iris. An achromat doublet lens with a focal length of 40 cm is used for collimating the light beam with a diameter of 25.4 mm. This beam is then reduced to 20 mm with an iris before interacting reaching the Lloyd s mirror interferometer. The interferometer works by having half of the incident 20-mm beam diameter directly expose the photoresist-coated slide and the other half is incident upon an aluminum-coated prism pressed against the slide. The prism is arranged in such a way that the mirror side is perpendicular to the glass slide, creating an interference pattern such that its spatial frequency is given by: λ = 2sin Λ. (1) ( θ ) where Λ is the grating period, λ is the wavelength of laser beam, and θ is the incident angle. Figure 3. Schematic of holographic exposure setup D. Photomask Development Development is performed in a 1:4 solution of Shipley 351 Microposit Developer and DI water. Development times typically vary between 2 to 5 minutes. Development progress is monitored through the observation of the diffraction efficiency also as described in [1]. The exposed slide is mounted in the developer in Littrow configuration in relation to a 633-nm He-Ne laser aligned into the developer tank. The intensity of the negative-first-order transmitted beam is observed and plotted in a LabView program. Once an intensity peak can be seen, the development is halted by submersing the slide in DI water and blown dry with N 2. The peak where development is halted can be seen in Fig. 4 as it is where the intensity curve ends. Halting the process at this point allows for a 50% duty cycle in the grating photomask. After development, the slides are post-baked for 30 minutes at 115

3 C. Fig. 4 shows a typical intensity curve generated as development progresses and Fig, 5 shows a schematic of the development setup. Intensity ( W) Exposure Development Time (s) Figure 4. Diffraction efficiency as development progresses Figure 5. Schematic of development process monitoring E. Reactive- Ion Etching The next step is to transfer the sub-micron patterns created in the photoresist film to the glass substrate. This is done using an STS Multiplex Etcher deep reactive-ion etching (DRIE) tool. Samples are first bonded to a silicon substrate using an adhesive, Crystalbond. The chamber pressure is 5 mt, the chamber power is 500 W, and the plate power is 300 W. Samples are etched under 8 sccm of hydrogen (H 2 ) and 5 sccm of tetrafluoromethane (CF 4 ) for 30 seconds. After etching, the slides are cleaned with acetone to remove the adhesive and any residual photoresist. Current samples are etching to a depth of around 60 nm to 70 nm. The gas composition and etching time have also taken much calibration. Fig. 6 shows the effect of gas composition and etching time on the diffraction efficiency. Method of taking diffraction efficiency is discussed in section III C. Diffraction Efficiency 0.90% 0.80% 0.70% 0.60% 0.50% 0.40% 0.30% Reactive-Ion Etching 0.20% Etching Time (s) 5 sccm CF4 and 4 sccm H2 5 sccm CF4 and 8 sccm H2 7.5 sccm CF4 and 4 sccm H2 7.5 sccm CF4 and 8 sccm H2 Figure 6. Correlation of gas composition and etch time with measured diffraction efficiency It can be seen from the graph that there is a pronounced peak in diffraction efficiency with an etch time of 30 seconds for all gas compositions. Of the different gas compositions, it can also be seen that using 5 sccm of CF 4 and 8 sccm of H 2 provides a higher diffraction efficiency. We consider that a lack of directionality in the etching process of our current DRIE tool as the major cause of the etching depth of our surface relief gratings to be shallower than the one present in the photoresist films. With the DRIE, it is apparent that the trenches of the photoresist photomask are etched faster than the exposed glass troughs, resulting in reduced glass etching depth before the photoresist is completely etched away. We must be able to etch the glass to our goal depth before all photoresist is removed. If etching is continued after all photoresist is removed, diffraction efficiency lowers, as can be seen in Fig. 4. A more efficient process would etch the entire structure uniformly, allowing us to achieve a grating depth in glass similar to that which we have achieved in photoresist film. In the future, we will be using an ion-milling tool, which has a much greater directionality and is expected to allow us to etch the gratings to higher modulation depths on the glass surface. F. Waveguide Layer Coating After being cleaned, the slides are ready to have the waveguide layer deposited. This is done by an atomic layer deposition process using a Savannah 100 tool from Cambridge Nanotechnologies. We deposit a layer of aluminum oxide (Al 2 O 3 ) to a thickness of 180 nm, creating a single-mode planar waveguide. This process is performed under 300 C using a 5 second cycle duration over 2,066 cycles with water and trimethylauminum (TMA) as the precursors. Many factors influenced the choice of aluminum oxide as our waveguide layer. It is well known that a waveguide film must have a refractive index higher than that of the substrate in order to create a step-index waveguide. Since we will be using both soda-lime glass and fused silica as our substrates, this requires films with refractive index higher than The major constraint is that, for our aimed applications, the material must be highly transparent in the UV and the visible spectral range

4 with low propagation losses (below 1 db/cm). These constraints led us to choose aluminum oxide as a potential candidate. The film thickness is chosen to provide single mode operation for a broad spectral range. The waveguide structure must support a mode in the higher wavelength limit while allowing only a single mode in the short wavelength limit. For applications in the wavelength range of 550 nm to 350 nm, a thickness range of 179 nm to 201 nm is permissible, disregarding dispersion [3]. Fig. 7 shows a grating coupled light beam propagating inside a single-mode planar waveguide. Figure 8. Schematic of diffraction efficiency measurement setup In Littrow configuration, measuring the negative first order reflected beam presents difficulties as it is reflected back along the incident beam, as can be seen in Fig. 8. This problem is overcome through the use of a beam splitter on the incident beam. Part of the incident beam is passed to the grating while the remaining power is reflected to one side. The negative first order reflected beam is partially transmitted back towards the laser while the remaining power is reflected to the other side, allowing it to be measured. Application of Fresnel reflection/transmission equations is all that is required to convert the measured reflected light to the actual light reflected by the grating. Figure 7. Coupled light being guided in a single-mode planar optical waveguide III. CHARACTERIZATION To monitor the results of each stage of fabrication, we have been applying a variety of analytical tools. For example, by measuring the diffraction efficiency, the topology, and the film thickness, we can characterize several aspects of the device at different stages of the fabrication process. A. Diffraction Efficiency Diffraction efficiency in sub-micron gratings is a useful parameter to determine the depth of a surface modulation. Measuring the efficiency of the negative first order of the diffracted beam in transmission allow us to indirectly measure the grating depth through the use of a correlation obtained by modeling calculations or by direct measurements of the surface topology. Measuring the diffraction efficiency has the benefit that it is quick, simple, and non-destructive. To take the measurements, we mount the slide in a Littrow configuration and measure the intensity of the incident beam, zero-th order transmitted beam, zero-th order reflected beam, negative first order transmitted beam, and the negative first order reflected beam. Fig. 8 shows a schematic of the setup to take the diffraction efficiency measurements. B. Topology The topology of the diffraction gratings is measured by using a Park Scientific M5 atomic force microscope (AFM). The AFM is an efficient tool for characterizing the diffraction gratings we produce. With it, we can generate a 3D image of the grating and a 2D cross-section profile to determine the depth of the grating, which we relate to the diffraction efficiency. In addition, the 2D profile allows us to determine the roughness of the peaks and valleys of the gratings, which contributes to the unwanted scattering of incident light. With this tool, we have found that the depth of the modulations typically range between 35 nm and 60 nm. The RMS of the roughness of the peaks and valleys ranges between 0.3 nm and 5 nm with the average around 2.5 nm. The AFM is chosen over a Scanning Electron Microscope (SEM) due to the SEM s need for the sample to be coated with a metal film. Metalizing the samples takes time and removing the metal off of the samples after measurements could potentially damage the samples. The AFM is rather straightforward as it does not require a metal coating, but may also potentially be destructive in the measurement and cleaning processes. Fig. 9 shows the 3-Dimensional image produced by the AFM and Fig. 10 shows the 2-D profile associated with the same grating.

C. Film Thickness Film thickness of transparent films can been precisely determined with a spectrophotometer using the envelope method described in [2].

5 C. Film Thickness Film thickness of transparent films can been precisely determined with a spectrophotometer using the envelope method described in [2]. A transmission spectrum acquired with a Cary 300 spectrophotometer is taken and interference fringes appear, creating minima and maxima of transmission envelope curves. Using the transmission values in those envelope curves as well as the known refractive index of the substrate, we can find the index of refraction of the thin film, which is given by: where N is given by n 2 2 = N + N s (2) Figure 9. 3-D image produced by the AFM TM T N = 2s T T M m m s and s is the refractive index of the substrate, T M is the transmission maxima and T m is the transmission minima. Using the results of the refractive-index is along with the wavelengths associated with adjacent minima or maxima, the film thickness can be calculated by: 2 1 (3) Figure D profile produced by the AFM d λ1λ2 = (4) ( λ n λ n ) Correlating the grating depth obtained from the AFM as well as the diffraction efficiency allows us to calculate the grating depth by only taking efficiency measurements. This combination overcomes the destructiveness of the AFM. Our current experimental correlation shows diffraction efficiency growing monotonically in the grating depth ranges we have generated. Fig. 11 shows the current experimental correlation. Diffraction Efficiency 7% 6% 5% 4% 3% 2% 1% 0% Characterization Grating Depth (nm) Figure 11. Experimental correlation of diffraction efficiency and grating depth where λ 1 and λ 2 are the wavelengths associated with adjacent minima or maxima, and n 1 and n 2 are the refractive indices at those points. IV. CONCLUSION We have described here the fabrication steps of integrated grating couplers and single-mode planar optical waveguides aimed for UV-visible spectroscopic studies of surface adsorbed materials. The substrate cleaning, holographic exposure, photomask development, and waveguide layer coating were developed for our applications. Ongoing efforts focus on extending our current technology into the UV spectral region and combining the integrated optic chip described here with broadband light sources and multichannel array detectors for spectroscopic research of molecular thin-films and surface phenomena. In order to achieve this, we will use fused silica as our substrate, as soda-lime glass is not transparent in the UV spectral region. However, after all fabrication steps have been calibrated, this change should present little difficulty. We will also implement the mathematical diffraction grating formalism described in [5] to confirm our experimental results with diffraction gratings and search for a precise correlation of diffraction efficiency and grating depth and shape. Once developed, the waveguide device will allow for spectroscopic investigations of biological thin films based on planar single-mode surface waves of unprecedented sensitivity

6 in a novel, simple architecture. This tool will aid in the development of many studies and devices in the biomedical field. ACKNOWLEDGMENT We would like to acknowledge Mark Crain, Courtney Byard, and Nathan Webster for their assistance with this work. REFERENCES [1] L. Li, M. Xu, G.I. Stegeman, and C.T. Seaton, Proc. SPIE-Int. Soc. Opt. Eng 835, 1987, pp. 72. [2] R. Swanepoel, Determination of the thickness and optical constants of amorphous silicon, Journal of Physics E: Scientific Instruments, vol. 16, [3] Donald L. Lee, Electromagnetic Principles of Integrated Optics. New York: John Wiley & Sons, 1986, pp [4] Eugene Hecht, Optics, 4th ed. New York: Addison Wesley, 2001 [5] R. Petit, Electromagnetic Theory of Gratings, Topics in Current Phycics, vol. 22. Berlin: Springer-Verlag, [6] Martin Malmsten, ed., Biopolymers at Interfaces. New York: Marcel Dekker, Inc., [7] James F. Rusling, ed., Biomolecular Films. New York: Marcel Dekker, Inc., 2003.

Major Fabrication Steps in MOS Process Flow

Major Fabrication Steps in MOS Process Flow UV light Mask oxygen Silicon dioxide photoresist exposed photoresist oxide Silicon substrate Oxidation (Field oxide) Photoresist Coating Mask-Wafer Alignment