Mueller Matrix Imaging Microscope Using Nanostructured Thin Film Mirrors
Alexander Ruder
Author
04/05/2021
Added
91
Plays
Description
The operation and calibration of a Mueller matrix imaging microscope is demonstrated using nanostructured thin film mirrors for polarization state generation and analysis.
Searchable Transcript
Toggle between list and paragraph view.
- [00:00:00.030]Hello.
- [00:00:00.420]My name is Alex Ruder and I'm a PhD student in the department of electrical and
- [00:00:03.990]computer engineering at UNL.
- [00:00:05.670]My advisor is professor Mathias Schubert and this presentation is about a Mueller
- [00:00:09.570]matrix imaging microscope that we've developed that uses nanostructured thin film
- [00:00:13.530]mirrors.
- [00:00:14.760]I'm first going to go over some background information on Stokes vectors and
- [00:00:18.150]Mueller matrices, and various Mueller matrix instrument principles.
- [00:00:21.840]And then I'm going to provide an overview of the instrument construction it's
- [00:00:25.350]optical model and the calibration method.
- [00:00:27.990]Last we're going to take a look at some Mueller matrix images that were recorded
- [00:00:31.320]with the instrument of different anisotropic samples.
- [00:00:35.310]So the Stokes.
- [00:00:35.880]Vector is a four element column vector that describes the polarization
- [00:00:39.180]properties of a beam of light.
- [00:00:41.400]And the Mueller matrix is a four by four matrix that fully describes how an
- [00:00:45.210]optical element at a given wavelength and orientation will convert the
- [00:00:48.930]polarization of an incident beam of light. By multiplying a known stoke
- [00:00:53.580]vector with a known Mueller matrix
- [00:00:55.170]the resulting modified Stokes vector can be determined.
- [00:00:58.620]The Mueller matrix of a sample for a given orientation.
- [00:01:02.190]can be measured by varying the input polarization state,
- [00:01:05.250]and then observing the resulting polarization properties after the sample.
- [00:01:09.960]But because most detectors simply measure intensity and not polarization the
- [00:01:13.620]polarization properties after the sample also have to be varied in order to recast
- [00:01:18.450]the change in polarization into a change in intensity,
- [00:01:22.620]the group of optical components before a sample that generate the instant
- [00:01:27.270]polarization state are known as the polarization state generator or PSG.
- [00:01:32.460]The group of components after the sample are known as the polarization state
- [00:01:36.180]analyzer or detector, PSA, or PSD,
- [00:01:39.900]the general optical path of a mule matrix polarimeter begins with the source,
- [00:01:44.100]the PSG, the sample,
- [00:01:45.630]and finally the PSA because the the detector only records intensity
- [00:01:50.190]values. The first row of the PSA,
- [00:01:52.440]Mueller matrix only needs to be considered similarly carrying out the
- [00:01:56.100]multiplication between the source Stokes vector and PSG Mueller matrix results
- [00:02:00.570]in a column vector,
- [00:02:02.010]the PSG and PSA vectors are noted here as G and D vectors.
- [00:02:06.720]The Mueller matrix can then be rearranged as a vector and the effect of
- [00:02:11.370]the different configurations of G and D are rearranged as a matrix known as the
- [00:02:16.170]instrument matrix,
- [00:02:17.670]carrying out the vector matrix multiplication of the Mueller vector and
- [00:02:22.620]instrument matrix results in a vector of intensity.
- [00:02:26.220]This gives us a system of linear equations and a minimum of 16,
- [00:02:30.810]well chosen configurations of G and D are needed to solve the system.
- [00:02:35.370]Typically more are used, which results in an overdetermined system.
- [00:02:38.790]We can then solve
- [00:02:40.890]this system using ordinary least squares.
- [00:02:43.890]And in the case of a spatially varying Mueller matrix polarimeter this system is
- [00:02:48.390]solved for each pixel. The optical path of the instrument begins at the source,
- [00:02:52.710]which is a fiber collimated lens that is fed with a fiber
- [00:02:55.500]coupled green led. The collimated beam is reflected from the first anisotropic
- [00:03:00.160]mirror and passes through a field stop to limit the beam diameter.
- [00:03:03.490]The beam then passes through the sample and is collected by an objective lens.
- [00:03:08.260]The beam exits the objective lens and is reflected from the second
- [00:03:11.590]anisotropic mirror,
- [00:03:12.790]which is then collected by an achromatic tube lens that forms an image on the
- [00:03:16.990]sensor of a monochrome camera.
- [00:03:19.000]The mirrors are made using electron beam of operation. An optically thick
- [00:03:22.600]titanium base layer was deposited onto thick quartz
- [00:03:25.510]substrates. Glancing angle deposition was used to grow a thin
- [00:03:30.580]titanium slanted calmer than film on top of this optically
- [00:03:33.730]thick base layer. These are images of a similar titanium,
- [00:03:37.120]SCTF grown on a Silicon substrate,
- [00:03:40.420]the slanted columnar structure of the film results in strong optical anisotropy,
- [00:03:45.340]especially as a function of the, as azimuthal angle of the mirror.
- [00:03:48.970]Both anisotropic mirrors are attached to stepper motors that rotate
- [00:03:52.480]continuously at different speeds. In this case,
- [00:03:55.330]the first mirror completes one revolution during the measurement
- [00:03:57.790]and the second mirror completes five revolutions. An encoder attached to the
- [00:04:01.780]second rotating anisotropic mirror is used to generate a clock signal that is
- [00:04:05.830]fed the camera. This serves as a frame acquisition trigger pulse,
- [00:04:09.700]and by triggering frame acquisitions this way,
- [00:04:12.760]the different angular positions of each rotating mirror are precisely known for
- [00:04:16.900]each frame of
- [00:04:17.420]the measurement. In the current setup 128 frames are recorded for a single
- [00:04:22.420]measurement.
- [00:04:22.960]The time varying Stokes parameters for each mirror are modeled as a fourth order
- [00:04:27.010]Fourier series where each vector element has unique Fourier coefficients.
- [00:04:32.230]The coefficients are treated as images,
- [00:04:34.210]which results in a per-pixel model of the instrument,
- [00:04:37.660]carrying out the vector matrix vector multiplication gives a stack of
- [00:04:42.430]intensity images where, in this case, consists of 128 images.
- [00:04:46.870]The Fourier coefficient images are initially unknown and must be determined using a
- [00:04:51.340]calibration process. Several especially homogenous samples with different,
- [00:04:55.930]mostly ideal, Mueller matrices are measured. In this case,
- [00:04:59.470]we took measurements of air as well as a polarizer and a wave plate with the
- [00:05:04.090]polarizer and wave plate measured at different rotation angles. Regression
- [00:05:09.070]analysis is then used to find the best fitting Fourier coefficient images that
- [00:05:13.150]result in the closest match generated intensity to the
- [00:05:18.000]measured calibration image stacks.
- [00:05:20.910]The resulting coefficient images look like for both PSG,
- [00:05:24.120]which is the bottom four rows. And the PSD,
- [00:05:27.660]which is the upper four rows. On the left side
- [00:05:31.290]You can see that, at first,
- [00:05:32.700]it doesn't look like there's significant spatial variation between each of the
- [00:05:36.510]elements when they're all set to the same scale.
- [00:05:39.030]But if we allow each frame to rescale (on the right side),
- [00:05:42.450]but keep the zero value color mapped to white, you can see that there is spatial
- [00:05:46.890]variation across a significant number of the elements here in the coefficient
- [00:05:51.240]images.
- [00:05:51.630]Now the instrument is ready to accurately measure Mueller matrices of
- [00:05:55.800]unknown samples. As a sanity check,
- [00:05:58.580]the recorded intensity measurement for the straight-through air measurement is
- [00:06:02.330]determined. Because air doesn't appreciably alter the polarization
- [00:06:06.560]properties of light over short distances,
- [00:06:08.600]the Mueller matrix should ideally be an identity matrix,
- [00:06:11.840]which is close to what we see here. Across the diagonal
- [00:06:15.080]the values are very close to one while there is optical activity and the off
- [00:06:19.340]diagonal elements in the image,
- [00:06:21.230]the values of each of these frames have been scaled up by twenty-five times.
- [00:06:24.680]So the resulting measurement is actually quite close to what it should be.
- [00:06:28.700]Moving on to other samples. this is a patterned titanium slanted
- [00:06:31.640]columnar thin-film deposited onto a glass slide,
- [00:06:35.180]the same type of structure that the rotating anisotropic mirrors are made out
- [00:06:39.620]of. Um, you can see that there's significant optical activity in the one-two
- [00:06:43.490]and two-one elements of the Mueller matrix.
- [00:06:46.490]And if we take the same region in the sample and rotate it in the plane by
- [00:06:51.320]45 degrees, then repeat the measurement.
- [00:06:54.230]We can see that the optical activity propagates from the one-two
- [00:06:58.640]and two-one Mueller matrix elements into the one-three and three-one
- [00:07:03.500]matrix elements. Last,
- [00:07:05.690]we have a birefringent resolution target from Thorlabs. Observing this
- [00:07:10.490]without polarizers, it just looks like a piece of glass.
- [00:07:13.400]But when this resolution target is placed between two crossed polarizers,
- [00:07:17.210]the pattern,
- [00:07:18.260]on it can be seen and rotating the sample between the cross polarizers causes
- [00:07:22.910]the pattern to invert as it's rotated.
- [00:07:26.060]So the sample acts as a half wave plate with a homogenous phase shift over the
- [00:07:30.320]sample plane,
- [00:07:31.220]but different fast axis orientations in the background and pattern regions.
- [00:07:35.960]This is reflected in these Mueller matrix images.
- [00:07:39.350]We can see element on one has almost no optical activity,
- [00:07:42.950]which indicates that the sample isn't strongly absorbing or scattering the light
- [00:07:47.780]and the near lack of activity in the first row in and column also indicates
- [00:07:52.220]that the sample isn't highly polarizing. So in conclusion,
- [00:07:55.700]we were able to use nanostructure thin film mirrors as polarization,
- [00:07:59.690]state generator and analyzer and Mueller matrix imaging microscope,
- [00:08:04.190]A spatially varying Fourier series approach was used to calibrate the instrument
- [00:08:08.690]and model the optical properties of each of the mirrors,
- [00:08:12.870]especially varying samples were imaged with good results.
- [00:08:15.890]And in the instrument shows a promising optical simplification for Wheeler
- [00:08:20.750]matrix imaging.
- [00:08:21.920]Thank you for listening to my presentation and thank you to all of my
- [00:08:24.980]collaborators that helped make this work possible.
The screen size you are trying to search captions on is too small!
You can always jump over to MediaHub and check it out there.
Log in to post comments
Embed
Copy the following code into your page
HTML
<div style="padding-top: 56.25%; overflow: hidden; position:relative; -webkit-box-flex: 1; flex-grow: 1;"> <iframe style="bottom: 0; left: 0; position: absolute; right: 0; top: 0; border: 0; height: 100%; width: 100%;" src="https://mediahub.unl.edu/media/16370?format=iframe&autoplay=0" title="Video Player: Mueller Matrix Imaging Microscope Using Nanostructured Thin Film Mirrors" allowfullscreen ></iframe> </div>
Comments
0 Comments