We present an all-optical modulator realized on a silicon chip. The proposed modulator has nano scale dimensions and a high extinction ratio. Its operation principle is based on a spatially non-uniform variation of the absorption of a miniaturized, silicon waveguide - based Mach-Zehnder interferometer (MZI). The absorption variation is obtained by illuminating the MZI with visible light. Our modulator may be used as an interfacing link between microelectronic processing circuits and optical information transmission links. We provide details on the fabrication and the experimental characterization of the suggested device. Since the operation principle is not based on a high Finesse resonator, the modulator is less sensitive to wavelength changes and its operation rate is not connected to the time required for the optical response to reach steady state but rather to material related effects.
©2007 Optical Society of America
The development of miniaturized photonic devices is a challenging academic as well as industrial pursuit. Such devices may answer requirements of reduced size, reduced power dissipation, increased operation rate and better data processing performance [1–3], which are expected to lead the current microelectronic devices to more or less a dead end. Moreover, the present situation in which the information is created and processed using electrons (or holes) but transmitted over optical fibers as photons, requires integration of miniaturized photonic data processing devices on a silicon chip. Optical modulators are basic devices that serve as building blocks for various higher level circuits used in data processing. Being able to realize such devices on a silicon chip while operating/activating them with light and therefore reducing their power dissipation, and increasing their operation rate and robustness to noise level is a very important task,. However, while photonic structures that split, couple, add, subtract, bend and filter information channels [4–5], as well as other devices that are integrated with microelectronic devices, such as silicon low-loss waveguides [6–7], amplifiers [8–9] and lasers [10–11] are known, fast nano-scale all-optical modulators are not.
In this paper we present a nano-scale miniaturized waveguide Mach-Zehnder interferometer (MZI) realized on a silicon chip. By illuminating the MZI structure with visible light, one varies the Si refraction index and consequently the Si absorption coefficient for an Infra-Red (IR) information beam passing through it. The visible light illumination is the control command that turns the modulator to its ON and OFF states. In our case, we illuminate the structure in a spatially non-uniform way (e.g. by illuminating through a scattering medium and projecting speckles on the MZI), therefore breaking the symmetry or the balance between the two arms of the interferometer. The advantage of this approach is that it results in a high contrast, since when the arms are balanced, no output is obtained, whereas even the slightest misbalance will produce an output signal. Therefore, a very small scale device can produce the desired outcome. This approach differs from the one in which external visible illumination is applied directly on the optical waveguide that guides the IR beam , where one needs a long interaction length to obtain a high contrast, because there is no balancing mechanism..
The use of visible light to vary the absorption of Si was previously applied mainly to ring resonators. In contrast, the novelty we bring in this paper is as follows:
- We design a nano-scale device much smaller than in previous concepts using light induced Si absorption changes (device area of ~ 1μm2 instead of hundreds of square microns).
- We use spatially non-uniform illumination and obtain a high contrast modulator with better performance than a device based on direct illumination on a waveguide as in Ref. .
- Since we do not use a high Finesse resonator, our device is less sensitive to wavelength shifts and to spectral transmission and its operation rate is no longer connected to the time required for the optical response to reach steady state but rather to material related effects (although in general in Si, the free electron life time is usually longer than the life time of photons inside the resonator).
Therefore, in our concept we overcome the disadvantage of realizing optical switching by using relatively long structures [13–15] based on changing the refractive index in a MZI, by using a ring resonator coupled to a waveguide [16–17] or by using extremely high laser powers in a large or non planar structure  through which the modulated light is being guided. Note that there are other modulation approaches that have been demonstrated with III-V compound semiconductors [19–21] but not in silicon. Other concepts for realizing a fast (but not nano-size) electro-optical modulator in a silicon wafer may also be seen in Refs. [22–24].
The designed all-optical modulator having an area of less than ~1μm2 presented herein is very small and it is becoming comparable to the dimensions of transistors obtained in microelectronics. Note that ring resonator solutions as the one presented in Ref.  require a ring with a diameter of more than 20μm although some designs based upon nano-cavities have been suggested as well . In order to demonstrate experimental performance, we fabricated and characterized a larger modulator containing an MZI built from two 4μm by 1μm beam splitters connected via waveguides with a width of 200nm. The overall area of the device was less than 10μm2. The signal was carried by 1.55μm light and the modulation control was by 532nm light. The modulation control illuminated the beam splitters from above in an asymmetrical manner, which produced mobile electrons and holes that affect the refractive index n and absorption coefficient a of silicon [26–27] as follows:
where Δαe[cm-1] is the absorption coefficient due to ΔN[cm-3] (the change in the mobile electrons concentration) and Δαh[cm-1] is the absorption coefficient due to ΔP[cm-3] (the change in the mobile holes concentration). Δn is the change in the refractive index, Δne is the refractive index change due to electron concentration change and Δnh is the refractive index change due to hole concentration. One may see that for a given change in electron concentration there is a change in the absorption coefficient as well as a change in the refractive index. This absorption misbalances the two arms of the interferometer such that in destructive interference the two arms will still show opposite phase but different amplitude, therefore providing positive output energy. The overall interaction length in this device is three orders of magnitude shorter that that in high Finesse ring resonator-based devices (e.g. Ref. ) or conventional waveguides.
In section 2 we describe the operation principle of the proposed device and in section 3 we characterize the device experimentally. Conclusions are given in section 4.
2. Technical description
As outlined, the idea is to design a nano-scale miniaturized MZI and to vary its absorption coefficient by illuminating it with spatially non-uniform visible light. Although the device designed, fabricated and experimentally tested included a MZI with dimensions of less than 10μm2 (about 9 microns by 1 micron), we also present a design based on the same concept having dimensions of less than 2 by 0.5μm (i.e. area of ~1 μm2). It is this device that may be accurately called a “nano-scale” device. This device operates as an all-optical modulator in which the information is IR light guided through the device and the control is visible light. Experimentally, we illuminate the device from above with visible light at 532nm, using an appropriate laser and a scattering surface. This illumination generates speckles (non uniform spots of spatial interference) on top of the MZI structure. The speckles have dimensions close to the visible wavelength (i.e. ~0.5μm), creating spatial non uniformity that breaks the balance between the two arms of the MZI and hence realizing an ON/OFF control command. The imbalance is obtained due to the change of the refractive index n and the absorption coefficient a as explained above.
Figure 1(a) shows at the top a schematic sketch of the fabricated device including the waveguides used in order to couple light into it from the tunable fiber laser, and at the bottom a device with visible light illuminated from above. Another modular format (proposed but not yet implemented) is shown in Fig. 1(b), in which the visible control light is guided in an oxide or photo-resist layer up to the beam splitter of the interferometer. The light blue lines are silicon waveguides guiding the IR information beam, while the dark blue waveguide is the oxide or the photo-resist. Figure 1(b) also shows two options for the interferometer. The lower drawing resembles a conventional MZI configuration, while the upper one is a miniaturized configuration.
3. Experimental characterization
We fabricated the 10μm2 device on a Silicon-on-Insulator (SOI) chip with a 200nm-thick active (upper) Si layer. A light coupling waveguide having a diameter of 10μm and a funnel structure was fabricated at the edge of the wafer. The funnel has a shape that decreases gradually in diameter until it reaches the position on the silicon chip where the device itself is located (see Fig. 1(a)). At this point the funnel has a diameter of 200nm, assuring generation of a single mode in the silicon MZI arms. A special fiber to waveguide system was designed and fabricated to couple the IR light from an IR laser to the 10μm diameter waveguide. The coupling was done through a polarization maintaining fiber. The light coupling system included a tapered fiber that focuses the light beam into a spot of 5μm with a polarization perpendicular to the surface of the wafer. The exit of the MZI device was designed to mirror the entrance, i.e. as an inverted funnel, starting from a core of 200nm and gradually increasing in width to 10μm at the edge of the wafer. From there, the exit waveguide was coupled through a multimode fiber (with a core of 105μm) to an IR detector. The scanning electron microscope (SEM) image of the fabricated device including the input and output funnels may be seen in Fig. 2.
Figure 2(a) bottom shows a SEM view of the interferometer device and the two funnels (on the right). The 200nm waveguides are shown in zoom on the left. Figure 2(b) shows a zoom SEM view of one of the two beam splitters of the interferometer device. As mentioned, the overall length of this device is about 9 microns. Although this is small enough, our design concept can be applied for smaller devices as well. Figure 3 shows numerical simulations for the same MZI having dimensions of less than one square micron. Figure 3(a) shows the constructive interference state of the MZI and Fig. 3(b) shows its destructive interference state. The difference between the constructive and the destructive conditions was obtained by adding spatially non uniform absorption, which broke the balance between the two interfering beams within the MZI. The wavelength we used in the simulation was 1.55μm. The waveguide was fabricated in silicon.
Note that the device presented in Fig. 3 is a Multi Mode Interference (MMI) device, i.e. in the region of interference the waveguide is wider and the single mode condition is no longer fulfilled. The interference between the multiple spatial modes is responsible for having the destructive output.
When the balance is broken, the interference between the multiple modes is such that energy is obtained at the device output. At the output of the MMI device, we go back into the single mode waveguide and therefore some energy is back reflected. The MMI is also wavelength dependent, but this dependence is not as strong as resonators with high Finesse.
Figure 4 shows results of an experimental characterization of the fabricated device. The figure shows measurements of the spectral response of the fabricated device, under illumination from above by a speckles pattern at 532nm. One may see that the interferometer has a transmission peak (constructive interference) approximately at 1546nm. The device is wavelength dependent due to the MMI structure and the substrate refractive index. However, the bandwidth is still as wide as 10nm (while for instance with a resonator having a free spectral range of 100nm and a Finesse of 1000, the width is less than 0.1nm). Note that the spectral response does not change significantly after application of the illumination, which indicates that indeed the relevant effect occurring due to the external illumination is related to misbalanced absorption rather than to significant variation of the relative phase between two interfering optical paths (which would have caused a shift in the spectral chart).
Figure 5(a) shows the measured temporal response of the device. It shows that a response rate of above MHz is indeed feasible. The visible illumination applied to the device was at 532nm with pulses with duty cycle of 5%. The various colors designate the wavelength of the information. One may see that following the results of Fig. 4, at a wavelength in which the interferometer produces output (constructive interference as shown in Fig. 5(a) at 1543nm), the illumination causes an attenuation of the output IR information signal. At a wavelength where no constructive interference exists, such as 1556nm, one sees that the control light breaks the balance between the two arms expected to create destructive interference and therefore the output signal is high (when the illumination is ON). Another reason for the high energy output at a destructive interference wavelength is due to scattering and back coupling of the visible light to the detector. Note that this indicates that the real attenuation caused by the illumination (the “control” command) is even stronger, since it also overcomes the back -coupling of the scattered light.
Note that 1MHz is not the true limit for the operation rate, because we did not measure the true response rate due to equipment limitations, but rather saw that it is above the limit of 1MHz. The limit of 1MHz was imposed by the maximal spectral content obtained using the modulation rate of our visible laser, rather than the true limit for the operation rate of the device itself. In Fig. 5(b) one sees the measure of the pulses that the visible laser can produce when being modulated at various duty cycles. One sees that those pulses cannot go below the micro-second limit even for a duty cycle of 5%.
Figure 6 shows the difference between the case of illumination with visible light (at 532nm and a duty cycle of 5%) over the waveguide core (blue line) vs. direct illumination over the interferometer. One sees that illumination over the interferometer device produces about 2.5 times stronger contrast in comparison with direct illumination over the waveguide itself. This result emphasizes one of the advantages of our concept that uses an interferometer rather than just a simple waveguide.
Figure 7 shows a mapping of the power required in order to activate the all-optical modulator. The figure shows the measurements of the effect of the visible illumination power on the contrast of the modulation of the IR information beam (at 1546nm). The current supplied to the visible 532nm Nd:YAG laser is given in [mA]. 400mA is the minimal current required to obtain the modulation effect.
Figure 8 shows the required optical power of the illumination laser in units of actual optical power rather than in units of current supplied to the laser driver. As seen from the chart, the minimal activating average power for our experimental setting (e.g. the power of the coupled light) is 0.2mW. Note that this measurement was done using a visible laser operated at a frequency of 400Hz with a duty cycle of 5%. Therefore, the average power in the continuous case will be 4mW. The laser illuminated a spot area of more than 2000μm2. When the illumination is concentrated on area of the nano-scale device having dimensions of less than 1μm2 (e.g. spot of 0.5×0.5 μm2) the required power for our setting will be 500nW.
Figure 9 presents a picture of the silicon device illuminated from above with a green laser while IR light is being coupled into it. The picture also shows the projected speckle pattern that eventually generates the spatial non-uniformity responsible for the discussed operation principle.
Note that although the described device is a modulator realized in a planar waveguide, a similar approach may be realized inside optical fibers. For instance, following the approach described in Ref. , silicon wires can be inserted into photonic crystal fibers as well as regular fibers. For example, some fabrication attempts were done by A. K. George and Z. Zalevsky (while visiting) at the laboratory of Prof. P. Russell at Bath University, UK. Passing light through the silicon wire can realize an in-fiber all-optical modulator.
Figure 10 shows a fabrication attempt for such devices. Fig. 10(a) shows a cross section of a photonic crystal fiber with a silicon rod. Fig. 10(b) shows the cross section of a regular fiber with a silicon core. Fig. 10(c) shows a side view of the fiber of Fig. 10(b) showing the continuity of the silicon rod. In both Figs. 10(a) and 10(b), light was coupled to the silicon rods.
The purpose of the fabrication attempts illustrated in Fig. 10 is not to replace the ongoing research of the group at Bath on related topics, but rather just to indicate that the approach presented in this manuscript can eventually be implemented in order to generate in-fiber modulation devices.
This paper presents the numerical design, fabrication and experimental characterization of an all-optical modulator for Infra-Red light, where the modulator is activated with visible light. The modulator can have nano-scale dimensions and is constructed as a special compact Mach-Zehnder interferometer. The operation principle involves breaking the balance of the two arms of the interferometer by applying a spatially non uniform illumination of a visible control beam. The device has several advantages such as very small dimensions, fast optical response time and high contrast due to the balance obtained between the two arms of the interferometer. The required power to operate the device is relatively small, projected to be about 500nW for the final nano-scale version and with the described setting.
The authors from Bar-Ilan University would like to thank the Horowitz Fund of Bar-Ilan University for supporting this research. Zeev Zalevsky would like to acknowledge the contribution of Alan K. George and Profs. Philip St. J. Russell and Jonathan C. Knight to the generation of the outcome presented in Fig. 10. He would also like to acknowledge the UK Engineering and Physical Science Research Council for funding his previous visits to Bath, UK.
References and links
2. E. Yablonovich, “Photonic crystals: semiconductors of light,” Sci. Am. 285, 47–55 (2001).
3. T. F. Krauss, “Planar photonic crystal waveguide devices for integrated optics,” Phys. Status Solidi A 197, 688–702 (2003). [CrossRef]
4. M. Loncar, T. Doll, J. Vuc ∙ Kovic ∙, and A. Scherer, “Design and fabrication of silicon photonic crystal optical waveguides,” J. Lightwave Technol 18, 1402–1411 (2000). [CrossRef]
5. H. W. Tan, H. M. Van Driel, S. L. Schweizer, R. B. Wehrspohn, and U. Gösele, “Tuning a 2-D silicon photonic crystal using nonlinear optics,” Conf. on Laser and Electro-Optics 2004Vol. IFD2 (Optical Society of America, Washington DC, 2004).
6. K. K. Lee, D. R. Lim, and L. C. Kimerling, “Fabrication of ultra low-loss Si/SiO2 waveguides by roughness reduction,” Opt. Lett. 26, 1888–1890 (2001). [CrossRef]
8. R. L. Espinola, J. I. Dadap Jr., R. M. Osgood, S. J. McNab, and Y. A. Vlasov, “Raman amplification in ultra small silicon-on-insulator wire waveguides,” Opt Express 12, 3713–3718 (2004). [CrossRef] [PubMed]
12. S. Stepanov and S. Ruschin, “Modulation of light by light in silicon-on-insulator waveguides,” Appl. Phys. Lett. 83, 5151–5153 (2003). [CrossRef]
13. A. Liu, R. Jones, L. Liao, D. Samara-Rubio, D. Rubin, O. Cohen, R. Nicolaescu, and M. Paniccia, “A highspeed silicon optical modulator based on a metal-oxide-semiconductor capacitor,” Nature 427, 615–618 (2004). [CrossRef] [PubMed]
14. D. Samara-Rubio, L. Liao, A. Liu, R. Jones, M. Paniccia, D. Rubin, and O. Cohen, “A gigahertz silicon-on-insulator Mach-Zehnder modulator,” OSA Proceeding Series 2, 3–5 (2004).
18. H. K. Tsang, C. S Wong, T. K. Liang, I. E. Day, S. W. Roberts, A. Harpin, J. Drake, and M. Asghari, “Optical dispersion, two-photon absorption and self-phase modulation in silicon waveguides at 1.5 μm wavelength.,” Appl. Phys. Lett. 80, 416–418 (2002). [CrossRef]
19. M. Fetterman, C. P. Chao, and S. R. Forrest, “Fabrication and analysis of high-contrast InGaAsP-InP Mach-Zehnder modulation for use at 1.55 μm wavelength,” IEEE Photon. Technol. Lett. 8, 69–71 (1996). [CrossRef]
20. O. Leclerc, C. Duchet, P. Brindel, M. Goix, E. Grard, E. Maunand, and E. Desurvire. “Polarisation-independent InP push-pull Mach-Zehnder modulation for 20 Gbit/s soliton regeneration,” Electron. Lett. 34, 1011–1013 (1998). [CrossRef]
21. V. Van, T. A. Ibrahim, K. Ritter, P. P. Absil, F. G. Johnson, R. Grover, J. Goldhar, and P. T. Ho. “All-optical nonlinear switching in GaAs-AlGaAs microring resonators,” IEEE Photon. Technol. Lett. 14, 74–76 (2002). [CrossRef]
22. C. K. Tang and G. T. Reed, “Highly efficient optical phase modulator in SOI waveguides,” Electron. Lett. 31, 451–452 (1995). [CrossRef]
23. D. Marris-Morini, X. Le Roux, D. Pascal, L. Vivien, E. Cassan, J. M. Fedeli, J. F. Damlencourt, D. Bouville, J. Palomo, and S. Laval, “High speed all-silicon optical modulator,” J. Lumin. , 121, 387–390 (2006). [CrossRef]
24. D. Marris, E. Cassan, and L. Vivien, “Response time analysis of SiGeSi modulation-doped multiple-quantum-well structures for optical modulation,” Journal of Applied Physics 96, 6109–6112 (2004). [CrossRef]
25. S. F. Preble, V. R. Almeida, and M. Lipson, “Optically controlled photonic crystal nanocavity in silicon,” SPIE 49th Annual Meeting, Proceedings of SPIE Vol. 5511, 10–17 (2004).
26. R. A. Soref and B. R. Bennett, “Electro-optical effects in silicon,” IEEE. Journal of Quantum Electronics QE-23, 123–129 (1987). [CrossRef]
27. R. A. Soref and B. R. Bennet, “Kramers-Kronig analysis of E-O switching in silicon,” SPIE Integr. Opt. Circuit Eng., 704 (1986).
28. Z. Zalevsky, A. K. George, F. Luan, G. Bouwmans, P. Dainese, C. Cordeiro, and N. July, “Photonic crystal in-fiber devices,” Opt. Eng. 44, 125003 (2005). [CrossRef]