Integrated Optical Directional Coupler on Silicon February 2009 Introduction Silicon is well understood and robust material for the development of electronic devices and circuits. During the last decade or so, it has also been found to be an excellent material for optical devices. Recently, many optoelectronics devices have been proposed and demonstrated successfully. The unique optical properties of SOI structures offers ability to integrate photonic devices to CMOS integrated circuit (IC) technology. Recently, we have successfully fabricated an integrated optical directional coupler which is one of the basic building blocks for realizing functional photonics devices. It has major applications in developing polarization splitters, power splitters, modulators/switches and wavelength (de-) multiplexers. Principle of Directional Couplers Directional coupler structure consists of two closely placed coupled waveguides which are parallel to each other as shown in Fig. 1. Symmetric Supermode

Input light Output light

Lc Asymmetric Supermode

Figure 1: Scheme of the parallel waveguide coupler The launching of light into anyone of the input waveguide results in the excitation of symmetric as well as antisymmetric supermode in the system of waveguides having propagation constants β a and β b respectively. It is possible to completely transfer the power from one waveguide to another if Lc = π / βa − βb , where Lc is the length of coupled region. By varying Lc , one can achieve the required power splitting ratio of the two output ports.

(

)

Fabrication and Experimental Results The directional couplers with various coupling lengths have been fabricated by photolithographic definition and subsequent reactive ion etching process. Fig. 2 shows schematic of a typical directional coupler along with some SEM images of critical regions. The width of individual waveguide is 5 µm and the separations between the waveguides are kept at 2.5 µm in the coupling region. Since our aim is to couple light from fiber to waveguide, input and output end of the parallel waveguiding regions are decoupled .by using two S-bend waveguides.

Experimental Optics Dept. of Electrical Engineering, IIT-Madras http://www.ee.iitm.ac.in/optics/

Device compactness can be improved to about 10 times by replacing the symmetric s-bend waveguide by asymmetrically etched s-bend waveguide. a

b

Bar Port Input port

Cross Port

c

d

Figure 2: Schematic of a directional coupler with the experimentally obtained mode profiles at the input as well as bar / cross ports. The S-bends can be either symmetrically or asymmetrically etched. SEM images of some critical regions are shown: (a) symmetrically etched s-bend, (b) asymmetrically etched s-bend, (c) waveguides approach towards the coupled region, and (d) coupled waveguide region. The experimental results of four different direction couplers with different power splitting ratios are shown in Table – 1. Table 1: Cross / Bar port output power (%) for different coupling length Lc Coupling Length Lc (µm)

Cross Port Power ( % )

Bar port Power ( % )

1400 1500 1550

60 80 90

40 20 10

1650

100

0

Conclusions Rib waveguide directional couplers with different coupling lengths have been successfully fabricated on SOI substrate and different splitting ratios have been obtained. Using this directional coupler structure, more complex integrated optical devices like interleaver, Mach-Zehnder modulator, add-drop multiplexers, etc are being designed to be fabricated in our laboratory in near future.

John P. George Nandita Dasgupta Bijoy Krishna Das

Distributed Anti-Stokes Raman Thermometry May 2009 Introduction Distributed Temperature Sensing (DTS) is a critical need for several applications including down-hole monitoring of oil and gas exploration, power cable and transmission line monitoring, leak detection in oil and gas pipelines and fire detection in tunnels. For such applications, optical fiber is an attractive choice because of its thread-like structure which can be laid across pipelines, tunnels and measure temperature for several tens of kilometers due to its low attenuation constant. Raman scattering in optical fibers is well suited for the DTS application. Raman scattering is an inelastic scattering process, where energy exchange takes place between the molecular system and the incident photons. In this process, vibrational energy of the molecules either feed off (Stokes scattering) or feed in to (anti-Stokes component) the photon energy, resulting in the generation of new frequency components (Fig. 1). A key attribute of the Raman scattering process is that the anti-Stokes scattering is 4

⎛ λ ⎞ ⎛ −ΔE ⎞ R(T ) = ⎜ s ⎟ exp ⎜ ⎟ ⎝ K BT ⎠ ⎝ λas ⎠ strongly dependent on temperature, whereas the Stokes scattering is not. Hence, by taking the ratio R(T) of the two components one can estimate the absolute value of temperature ‘T’ experienced by the fiber. λs and λas are the Stokes and anti-Stokes wavelength, ∆E is the peak vibrational energy, and KB is the Boltzmann constant.

Raman Thermometry (DART). A short duration of light pulse is launched into the sensing fiber and the time of flight between this pulse and the backscattered signals that reach the receiver can be used to estimate the precise location from where backscattering has occurred. The experimental setup used for Distributed anti-Stokes Raman Thermometry (DART) is illustrated in Fig. 2.

Figure. 2 Experimental setup of DTS

To understand the trade-off between cost and performance, we developed a model that computes the backscattered Stokes, anti-Stokes and Rayleigh spectrum components as a function of distance for a given temperature map. The transmitter, fiber, optical filter, and receiver parameters can be fed to the model and their effect on the system performance may be evaluated. In addition, the model also allows for the use of signal processing techniques such as averaging and filtering for further improving SNR. Using the model, we designed a system with a semiconductor pulsed laser, circulator and InGaAs avalanche photodiode. In Fig. 3, we show experimental and estimated (using model) distributed temperature maps, where three sections of fiber are maintained at different temperatures using two ovens and a refrigerator. An optical pulse of 85 mW and width of 80 ns was launched while the APD used a reverse bias voltage of 48 V. The laser pulses were fired at a repetition rate of 1 KHz and the backscattered trace corresponding to each pulse is averaged for 220K times to achieve a signal to noise improvement of 25 dB. The DART system has been demonstrated with 3 ºC accuracy. Temperature Map 76

Figure 5: Illustration of Raman scattering mechanisms and the corresponding energy levels.

Distributed Anti-Stokes Raman Thermometry The approach that we have followed for distributed temperature sensing is to use Raman scattering principle in tandem with optical time domain reflectometry (OTDR), a technique commonly referred to as Distributed Anti-Stokes Experimental Optics Dept. of Electrical Engineering, IIT-Madras http://www.ee.iitm.ac.in/optics/

Temperature [oC]

62 48 34 20

Exp. 63 5 63 Deg. Est. 63 5 63 Deg.

6 0

2

4 6 Distance(Km)

8

10

Fig. 3 Experimental and estimated temperature map

L. Bharath Kumar Balaji Srinvasan

Fiber Infrared Fourier Transform Spectrometer May 2009

  Introduction Fourier transform spectroscopy (FTS) is way of estimating the optical spectrum using an interferogram. The latter is produced at the output of an interferometer. By vibrating one of the mirrors of the interferometer (see Figure 1), the path difference between the beams is changed. This results in fringes forming at a rate determined by the wavelength of the source and the velocity of the mirror. This “downconversion” of the frequency means that the optical signal can now be measured on an oscilloscope as the Fourier Transform (FT) signal is in the Hertz or kilohertz range. FTS has several advantages over conventional spectrometers. Most importantly all wavelengths present in the source are measured simultaneously [1]. To ensure that the developed Fourier transform spectrometer is low cost, small and vibrates with high precision, a square gold plated mirror (2 mm×2 mm) was fitted on a compact disc (CD) pick-up head by removing its lens. Using optical fibre improves the robustness of the system and maximizes efficiency by reducing light losses that occur in bulk optic systems.

SOURCE

GOLD REFLECTOR

3 dB coupler

GRIN LENS VIBRATING MIRROR V

FIBER

DETECTOR

FUNCTION GENERATOR

DSO DSO

Figure 1: Experimental setup

light and 2d is the distance traveled by the mirror in one direction. •

The output FT pattern was stored on a digital oscilloscope, also shown in Figure 2. The arrows show the peaks of the FT frequencies corresponding to source wavelengths 1550 nm and 1565 nm.

Down-conversion of frequency Down–converted frequency of light is given by the frequency of fringes: fFT = (2 × vm× f ) / c where c is the velocity of light, vm the velocity of mirror, fFT is the frequency of FT signal, and f is the frequency of light . Experimental set up and procedure •

Two laser beams, at wavelengths of 1550 nm and 1565 nm, are fed to the spectrometer through a 3 dB coupler.

•

With path length matching and proper alignment of the mirror, we observe interference fringes with high visibility and a maximum coupling back of light returning to the system.

•

The mirror was vibrated with a sine wave of 19 Hz and a peak to peak voltage of 1.5 V. A characterization experiment showed that the traveled maximum distance at this value of voltage and frequency. The distance traveled by the mirror is related to the resolution of the spectrometer as: R = c/4d

Figure 2: FT output of interferometer. The arrows mark the two source wavelengths at 1550 nm and 1565 nm.

Results The resolution for the particular set-up used was found to be 15 nm. Many applications require much higher resolution than this and future work will look at improving resolution. The present setup could be used in Course Wavelength Division Multiplexing, where the wavelength spacing is about 20 nm.

where R is the resolution in frequency, c is the velocity of

Experimental Optics Dept. of Electrical Engineering, IIT-Madras http://www.ee.iitm.ac.in/optics/

Sanjay Dhabhai Shanti Bhattacharya

Broadband generation from a fiber ring laser August 2009

Introduction Broadband sources find applications in a variety of areas such as optical communication, optical coherence tomography, optical gyroscopes and in optical spectroscopy. Fiber laser sources with erbium doped fiber (EDF) as the gain medium are particularly useful due to their emission in the C (1530 nm - 1560 nm) and L band (1560 nm – 1620 nm). A unidirectional travelling wave ring cavity laser with a nonlinear element within the cavity is a preferred design for broadband sources using nonlinear effects and could be operated in the continuous wave (CW) or mode-locked mode in the filterless configuration, to achieve spectral broadening. Experimental Setup The schematic configuration of erbium doped fiber ring laser (EDFRL) is shown in Fig 1. A suitable length of EDF is pumped by a semiconductor laser diode operating at 980 nm through a wavelength division multiplexer (WDM). Nonlinear fiber and mode-locking elements can be introduced at points A and B.

Fig 2. Broadband generated from the laser in the C-band, and the demultiplexed output ( inter-channel spacing of 100 GHz)

Fig 2 shows a typical broadband output and the individual wavelengths carved out from the broadband, to demonstrate the application of such a design towards wavelength-division multiplexing. The amplitude fluctuations are minimal in this scheme since the nonlinear processes are aided by the intrinsic gain at the generated wavelengths inside the ring cavity. The broadband can be tuned with an appropriate intra-cavity filter.

Fig 1 Schematic of a typical Erbium doped fiber ring laser.

Results and Discussion A CW broadband source is designed by introducing a dispersion shifted fiber (DSF) as an intra-cavity element at position B of Fig. 1. The large intra-cavity field built up in the laser undergoes multiple propagation through the DSF, and this leads to an enhanced spectral broadening. The multiple four-wave mixing processes occurring between the closely spaced longitudinal modes in the DSF result in a spectral broadening throughout the gain spectrum of the EDF. The mechanism of spectral broadening is influenced by the smallsignal gain spectrum of the doped fiber and there exists certain optimal cavity parameters for achieving large spectral widths with relatively low pump powers.

Experimental Optics Dept. of Electrical Engineering, IIT-Madras http://www.ee.iitm.ac.in/optics/

Fig 3. Broadband generated in the L-band (P : pump power)

The design is extended to demonstrate a broadband in the L band by using an EDF of larger dopant concentration (shown in Fig 3). The extent of broadening can be further increased with the use of highly nonlinear fibers, coupled with larger pump powers.

Deepa Venkitesh (Work done at IITB)

Multi-wavelength generation using four wave mixing August 2009

Introduction Four-wave mixing (FWM) is a parametric process in which, electromagnetic fields of different frequencies, propagating simultaneously in a nonlinear medium, interact through the third order nonlinear optical susceptibility of the medium, resulting in the generation of new frequencies. An attractive technique to design a fiber-based multi-wavelength tunable source is through the four-wave mixing of two continuous wave (CW) sources operated at slightly different wavelengths. The separation between the generated wavelengths can be tuned by tuning the separation between the two mixing wavelengths.

(a) L = 5 km (b) L = 19 km

Experimental Setup The experimental setup used to observe multi-wavelength generation through FWM is shown in Fig 1.

λ

3 dB coupler

DFB laser s

λ

EDFA

1

BPF

2

Optical Spectrum Analyzer

DSF

Figure. 2. Spectrum at the output of DSF of lengths (a) 5 km (b) 19 km.

The efficiency of generation of the FWM products are estimated theoretically for the Stokes and the antiStokes wavelengths by including the dispersive and nonlinear contributions to phase mismatch , and these are found to match with the experimental results. The frequently used formulae for calculating the dispersive phase shift at the Stokes and anti Stokes wavelength are modified for larger wavelength separations. The conversion efficiencies are estimated and compare well with the experimental results.

Figure.1. Experimental setup to observe multi-wavelength generation using FWM

Continuous wave outputs from two narrow linewidth (~30 MHz) distributed feed back (DFB) lasers (center wavelengths - 1550.12 nm and 1550.92 nm with a tunability of 2 nm each) are combined using a 3 dB coupler and the combination is amplified through the erbium doped fiber amplifier (EDFA). The amplified signal passed through a band pass filter (BPF) of bandwidth 1 nm, is coupled to a dispersion shifted fiber (DSF) and the output is studied on an optical spectrum analyser.

Results and Discussion Figure 2 shows the Stokes and antiStokes wavelengths generated due to four wave mixing, in two lengths of DSF, for a wavelength separation of 0.8 nm. Smaller wavelength separations led to larger number of sidebands, demultiplexing of which would find applications in wavelength-division multiplexing systems.

Experimental Optics Dept. of Electrical Engineering, IIT-Madras http://www.ee.iitm.ac.in/optics/

Figure 3. Conversion efficiencies for the first order Stokes and antiStokes wavelengths: Calculated and experimental results

This experiment forms the technological base for all-optical wavelength conversion. If one of the inputs is modulated, conversion to a desired wavelength is possible with an appropriate design through a similar scheme. Since electronic polarization, which is an instantaneous effect, is responsible for the FWM process, the wavelength converters based on FWM are not limited by the bit rate of the systems. Such alloptical signal processing functionalities could be implemented with better efficiencies in highly nonlinear fibers.

Deepa Venkitesh (Work done at IITB)

Optical Time Domain Reflectometry November 2009

Introduction

The OTDR trace

Optical time domain reflectometry is a method we use to identify fault locations along the length of an optical fibre. Since most fibres are underground, and also cover long distances, it is important to be able to use an instrument at one end and be able to infer the locations of reflective (connectors) and non-reflective (splices) events. In addition, we can also measure the attenuation of an optical pulse along the length of the fibre.

An OTDR trace is a plot of the optical power received at the detector versus the distance from the source. In Fig. 3, we see an example of such a trace. The slope of the line, as calculated between the cursors [A] and [B], gives the loss (in dB/km) between two points along the fibre. We also notice reflective events which are spikes above the trace. These are typically caused by connectors, or reflections from the end of a fibre.

Rayleigh Scattering Attenuation in an optical fibre is primarily due to Rayleigh scattering of light from the molecules in the optical fibre. The effect is inversely proportional to λ4 , i.e. the longer the wavelength, the less is the optical attenuation. Water molecules exhibit strong optical absorption near 1400 nm. We are also limited by the availability of suitable semiconductor light sources and detectors. The combination of availability of optical components and the underlying physics of Rayleigh scattering has led to the development of communication systems in three windows, shown in Fig 1. Figure 3: Sample OTDR trace with 3 reflective events

Shree Krishnamoorthy, Harish Ravishankar S. Thiruthakkathevan, V. Laxminarayanan Anil Prabhakar

The trace is sampled at regular intervals and the distance of an event is determined by assuming a known velocity of propagation for an optical pulse in the fibre. Commercial Prospects

Figure 1: Attenuation in an optical fibre.

The attenuation at 1310 nm is about 0.3 dB/km, while at 1550 nm is about 0.2 dB/km. This translates into a reach of about 100 km at 1310 nm, and 150 km at 1550 nm, when using an instrument with a dynamic range of 30 dB (assuming no other losses along the way).

Figure 2: Schematic electronics of an OTDR

Experimental Optics Dept. of Electrical Engineering, IIT-Madras http://www.ee.iitm.ac.in/optics/

OTDRs are commercially available instruments and are widely used in the telecommunication industry. The most common form adopted is that of a MiniOTDR, which typically has about 34 dB of dynamic range at 1550 nm., with an event resolution of less than 4m. More recently, as cable TV operators switch to using optical fibres to deliver digital television channels to homes, they rely on handheld versions that are more suitable for field operations within about 10km, i.e., the OTDRs have a lower dynamic range. Photographs of both protoypes are shown in Fig 4.

Figure 4: MiniOTDR (left) and Handheld OTDR (right)

A. Prabhakar, B. Srinivasan, N. Chandrachoodan Venkatesh, Thiruthakkathevan, Chellamal, Jayavel Senthil, Priya, Lakshminarayanan, Albina