Using distorted light to characterise surface layers
10 February 2012
Random isn't always random. PhD candidate Feng Liu of the EUV Lab group in DIFFER showed that the polarization of light is changed but not completely randomized after travelling through an optical fiber, as previously thought. Instead, the polarization coming out of the fiber is restricted by the symmetry of the fiber. This forms the basis for new instruments for characterizing surfaces. The research was published in the journal Optics Express.
The difficult thing about surfaces is figuring out what is going on. Everything of interest is, by definition, occurring in a layer that is just one atom or molecule thick, making signals weak and driving the development of more sensitive measurement techniques. One way to characterize surfaces is ellipsometry, using the direction of oscillation or polarization of light. The recipe for ellipsometry is very simple: take light with a very well known polarization, reflect that light off the surface, measure the polarization of the reflected light. Use the change in polarization in combination with models to determine what is on the surface.
But, ellipsometry is best performed when you have direct, line-of-sight access to the sample surface. In the most interesting circumstances – think of analyzing surfaces in reaction chambers or vacuum environments – we can't do that, because there are no windows at the right locations.
Diagram of the setup for an ellipsometry experiment – credit: Buntgarn / Wikimedia under Creative Commons Attributions ShareAlike 3.0
The team thought that polarization preserving optical fibers would allow us to deliver and collect the light with optical fibers and still measure the change in polarization. Unfortunately, it turns out that polarization preserving fibers do no such thing, and, in fact only preserve two out of the infinitely many, polarization states of light. Along the length of the fiber, the temperature and bending stress on the fiber induce refractive index changes. These change the polarization of light to such an extent that the output polarization of a polarization preserving fiber appears to be completely random.
An MSc student at DIFFER, Feng Liu, discovered that the large internal stress that allows a polarization preserving fiber to preserve two polarization states means that the output polarization states are not as random as they might seem. If the light entering the fiber has a fixed polarization relative to the axis of the internal stress of the fiber, then the amplitude of the two fields that describe all polarization states is fixed. The only thing that can vary is the phase between the two light fields. This restricts the polarization states to a circle on the sphere of all possible polarization states.
Ellipsometry with randomly varying polarisation states. The green blobs are the starting light polarizations. The black and blue blobs are the polarization states after reflection from a surface. The two arrows represent the changes in latitude and longitude.
Credit: Liu et.al., Optics Express
The team still can't predict where the polarization of the light exiting the fiber will fall on the circle, but the circle itself never changes. After reflection from the sample, the polarization states still fall on a circle, but the orientation of the circle on the sphere is different. By measuring this change in orientation, we obtain the change in polarization due to the sample. In other words, we have performed ellipsometry without ever really knowing the polarization of the light doing the measurement.
Once this is mastered, then a world of possibility opens up, where a single light source and polarization analyzer is linked to a fiber network, allowing industrial scale monitoring in environments where ellipsometry cannot normally be used.