Loss measurements

In this section waveguide loss and the factors that affect it are described. Why this is important and various ways that it can be measured are also discussed; including finally a description of the white light method used to measure a waveguide’s cut-off wavelength.

Losses of channel waveguides can be separated into two types of loss. The first type is the coupling loss suffered in coupling external light into the waveguide. This is made up of three parts; a mode mismatch between the input beam and the shape of the mode being considered, a diffuse reflection and refraction due to imperfect end-face polishing and a specular Fresnel reflection at the interface due to the change in refractive index. The second type of loss is the propagation loss. This is due to absorption or scattering of the guided beam in the waveguide and is proportional to the length of the sample (measured in dB/cm).

In this project, the larger the propagation loss of the waveguide, the more limited the length of the waveguide that can be used. The loss will also determine the power of laser source required to deliver the necessary power to a certain position along the waveguide. Although very high power lasers are available at the wavelength used (1066nm), the more powerful the laser, the higher the cost.

Waveguide losses have been measured in many ways with no method dominating. The two losses are generally measured together and the coupling loss inferred from the propa- gation loss. For example, the propagation loss can be measured by the cut-back method [139]. Here either many similar waveguides are fabricated with the only difference being in their path length. The total losses of these waveguides is plotted as a function of

their lengths. The loss should scale linearly with length with the intercept giving the coupling loss. Another method is by observing the scatter of light along the waveguide as a function of its position along the waveguide. The intensity of light scattered as a function of position along the waveguide should fit an exponential curve with the decay rate giving the propagation loss. This can be done by imaging with a camera if the substrate is lossy, or it can be done with photographic film close to the waveguide [140]. In this project propagation loss is calculated by comparing the difference in loss of waveguide of approximately 4cm length to a length of 3cm by repolishing one end of the waveguide. The difference in the loss gives the propagation loss directly.

The loss of a waveguide at a single wavelength is discussed above. However a measure of the loss at a range of wavelengths gives more detailed information about a waveguide. As has been seen in chapter 3 the number of modes a waveguide supports is dependent on its size. For an asymmetric waveguide there is a wavelength above which a waveg- uide will cease to guide the fundamental mode. This is called the waveguide’s ‘cut-off’ wavelength, at which point the theoretical loss of the waveguide will be infinite. This cut-off wavelength can be measured and used to help characterise a waveguide.

The experiment involves coupling radiation from a broad-band light source into a waveg- uide and measuring the loss of the waveguide over a range of wavelengths using a monochromator or optical spectrum analyser (OSA) to select narrow wavelength ranges. The apparatus used is shown in figure 5.6. White light from a bright tungsten source is coupled into standard SMF fibre with a 40x lens. Initially white light from the fibre is collimated with a lens positioned such that a narrow beam is incident on the monochram- ator slit. The computer controls the monochromator, slowly scanning in steps through a defined wavelength range. The output from a photodetector is simultaneously measured by a lock-in amplifier (LIA) and this is logged by the computer. An optical chopper positioned midway between the output of the fibre and the monochromator provides the reference signal for the lock-in amplifier (LIA). The measurement is then repeated for the case of the light being coupled through the sample waveguide and the output of the waveguide being directed into the monochromator. The loss (in dB) as a function of wavelength can be calculated using the following equation:

Loss(λ) =−10×log₁₀Pwg(λ)

Pf(λ) , (5.3)

wherePwg is the power measured out of the waveguide andPf was the power measured

in the fibre.

Depending on the cut-off wavelength of a waveguide either a silicon (range∼400nm and

∼950nm) or an InGaAs (range∼950nm and∼1700nm) detector was used. The InGaAs detector was preferred between the wavelengths of 900nm and 1100nm due to its greater sensitivity with lower noise at the wavelength to be used for trapping experiments. In

white light source mirror lens SM PM fiber lens chopper _aperture monochromator detector LIA PC waveguide polariser

Figure 5.6: Experimental apparatus used for white light measurements also showing

data interconnects (− →).

this project the cut-off wavelength is defined as the wavelength at which the loss rises to 3dB greater than the background level of loss.