Automated Spotsize Measurements

Automated Spotsize Measurements

BME 273 Project

April 25, 2000

Calley Hardin

Chris Baughman

Advisor: Dr. E. Duco Jansen

ABSTRACT

The Vanderbilt Free Electron Laser is a continuously tunable, source of pulsed IR radiation. The fact that the FEL can be tuned to specific wavelengths gives it many possible biomedical and clinical applications. Irradiance (Energy/area) is responsible for determining the amount and type of cell damage that will occur upon exposure to a laser. This means the spotsize of the beam at that point of contact is vital information in determining the physics of laser tissue ablation. The Vanderbilt Free Electron Laser is unique and in high demand for experimental use. For this reason, scheduling time is limited and highly valuable. Dr. Jansen is allotted only 8 hours of beam time per week. For this ultimate reason it is crucial to reduce any set-up or calibration time involved in experimentation. By automating the spotsize measurements we can significantly reduce the time required for this important data acquisition. A program for spotsize measurements was previously designed and has been in use for the past academic year. Our goal was to make this existing program obtain more data, in a more efficient and accurate manner. The program was to obtain spotsize data in two dimensions, within 10 microns without the use of software outside of LabView. The program was to be suitable for a spotsize of 0.3-0.25mm diameter beam. Finally, the program was to curve fit the data and output a visual representation of the beam profile. We were able to meet these requirements and produce a fully automated, efficient and reasonably accurate system for spotsize acquisition.

INTRODUCTION

Proposal:

The Vanderbilt Free-Electron Laser Center houses one of the most versatile lasers in the world. A free-electron laser (FEL) is so unique because the wavelength of the laser light is tunable. The Vanderbilt Free Electron Laser is a tunable pulsed laser. There has been many proposals that a tunable pulsed laser such as the FEL could have extensive biomedical and clinical applications (1). The effect of different wavelengths and radiation energies on tissue ablation is a question of great interest in the biomedical and clinical applications of the FEL (5). For example, has been postulated that tuned to a wavelength of 6.45 m, the FEL is well suited for tissue ablation and cutting (2). This wavelength corresponds to the Amide-II absorption band of protein. The theory is that the protein bond absorbs the radiation and the tissue is structurally weakened. In order to examine the validity of this hypothesis studies must be done concerning the physics and transfer of energy during tissue ablation. It is critical to know the energy input per unit area during ablation. By accurately measuring the spotsize of the laser beam at the point of ablation, the energy delivered to the system can be determined. This is just one of the many medical applications involving the importance of FEL spotsize measurements.

Melisa Moore and Brian Lenahan designed a system to automatically take spotsizes for the FEL last year as a BME272 senior design project (4). The system is fully functional and has been used in every beam shift since its completion. The system uses G-programming in LabView. In the process of using the program it became clear to Dr. Jansen and Mr. Mongin that some improvements could be made. Dr. Jansen observed that it would be extremely useful to have some additions made to the program that would enable the system to obtain data not only the horizontal direction but also the vertical. This would give a more complete approximation of the beam profile and enable the user to more easily locate the focal point. Figure 1 shows an estimated beam profile with measurements taken in both dimensions. It can be seen that this sort of data would clearly identify the focal point of the laser beam. Additionally, Mr. Mongin observed that it in the interest of time conservation and accuracy it would be useful to eliminate the need for external calculations and program the system to do its own mathematical analysis. Thus, a Spring 2000 BME 273 senior design project was proposed.

Goals:

The specific goals of the project were clearly outlined from the beginning. We were to make additions to the existing program with as little change to the original setup as possible. These additions consisted of three things. First we were to add a second dimension to the system. This would require an additional translation stage and an additional LabView program to be called upon by the existing program. Second, we were to research the mathematical capabilities of LabView and try to incorporate the tasks that were currently being performed by MATLAB and ORIGIN into the LabView program itself. Third, we were to graphically display these results on the front panel of the program to make the location of the focal point visually attainable and immediately obvious to the user. Given the fact that the laser is unique and in demand in the scientific community, and is thus often “borrowed” by other users, an additional task was to create a user’s manual for the system so that spotsize measurements could be obtained by any user in the Laser Contingency Room of the FEL Center.

Constraints:

The constraints of this project involved the properties of the FEL. The program needed to be accurate within 10 microns of the accepted values of spotsize. The program needed to accommodate a beam diameter of 0.3-0.25mm. The program also needed to be user friendly, i.e. a front panel that is self explanatory and reasonably simple.

METHODOLOGY

Spotsize Estimation:

The spotsize of a laser can be determined by employing the “knife-edge technique”. (3) Basically, the technique works by eclipsing a laser beam with a sharp knife-edge. As the knife-edge intersects the beam in a direction perpendicular to the propagation axis of the beam, a photodetector measures the intensity of the unmasked portion of the beam. If the laser has a Gaussian beam profile, as seen in Figure (3), a sigmoidal curve can be obtained. A Gaussian beam is one that is strongest at the very center of the beam and gets less and less powerful as you move away from the center. Figure (4) shows an example of the utilization of this technique on the free-electron laser. The y-axis on this graph is the total amount of energy that the detector reads and the x-axis is the distance the knife-edge has moved. Khosrofian and Garetz showed that the diameter of the beam could be determined by employing the following equations:

(1) ß-1= .552(X10 – X90)

(2) spotsize = 22 ß-1

X10 and X90 are the distances that the blade has traveled when the energy has dropped to 10% and 90% of its maximum value. Unfortunately, since the FEL’s maximum energy output is extremely volatile, determining exactly what maximum value one will use in determining X10 and X90 can be challenging. The old spotsize program that was developed in the Spring of 1999 by Melisa Moore and Brian Lenahan did not take this instability into account when calculating spotsizes. This flaw in the program made the program inaccurate in its calculation of the spotsizes. Thus, operators of the program have been exporting the data taken by the program to MATLAB and ORIGIN and making the spotsize calculations within these external programs. This became a very tedious and time-consuming task.

We determined that the best method for correcting for the instability of the beam was to curve fit the data to the standard sigmoidal equation:

(3) y = a/(b+e-cx)

, where a, b, and c are constants. The task of curve fitting the data had previously been accomplished by performing a non-linear, sigmoidal curve fit externally in ORIGIN. In order to curve fit the data within LabView, the Levenberg-Marquardt method of non-linear curve fitting was utilized. Basically, this method compares the data to an equation and a set of initial guess coefficients and minimizes the chi-square quantity until the fit converges on the data. The accuracy and efficiency of the program is very dependent on good initial guess coefficients and the number of iterations performed. If the coefficients are too far from their actual values then this method of curve fitting will not converge to a solution no matter how many iterations the program cycles through to find the answer.

The new curve fitting code was incorporated into the old program with complete success. The program is setup so that the user enters the initial guess coefficients as well as the number of iterations into the functional front panel of the program (Figure 5). Once the program has successfully fitted the curve and converged to a solution, it graphs the raw data as well as the curve fit on the graph output area in Figure (5). It then outputs the coefficients that it found for a, b, and c of equation (3) in the boxes labeled Best Guess Coef. The curve-fit program gives you the Mean Squared Error (MSE) of the curve as well a number representing any errors that may have occurred during the curve fitting process. The program then calculates the spotsize by taking an array of the fitted data and finding the maximum value of that array. The program then multiplies that maximum value by 0.9 and 0.1 to find the 90 and 10 percent y-values. In order to find the X10 and X90 values, we solved equation (3) for x and got equation (4).

(4) x = (ln ((a/y)-b))/(-c)

Since we now had the 90 and 10 percent y-values, we extracted the coefficients from the curve fitting program and plugged them into equation (4) with the appropriate y-value to find the X10 and X90 values. Once the X10 and X90 values are obtained in the program, they are plugged into equations (1) and (2) to find the spotsize of the beam. This spotsize is then displayed on the front panel as seen in Figure (5). A portion of the G-programming for the spotsize calculation is shown in Figure (6).

Addition of the Second Dimension:

Adding the second dimension to the existing program involved an in depth understanding of LabView and the elements of the existing program itself. It was determined that the best method was to create a new visual interface (vi) that called the original program which had already been revised to do the curve fitting. This new vi consisted of a FOR Loop. The input to the FOR Loop was the taken from the front panel where the user can input the number of spotsize measurements desired in the “Number of Measurements” box. The loop then calls the existing spotsize program, returns the system to its original position, moves in the vertical direction a given amount, and calls the existing spotsize program again. The loop continues in this manner until the desired number of measurements have been taken. Each time the spotsize program is called, the data is stored in an array outside the loop. Once the loop has finished all its measurements, the output array is sent to the front panel in the form of a plot. This plot gives the user an immediate answer as to where the focal point is relative to the original position of the knife-edge. Figure 7 shows the front panel including the input boxes, and both graphs.

Simulation and Testing:

During the beginning stages of development of our program, it became immediately obvious that testing would require an FEL simulator setup in order to complete our project. Experimental time on the FEL is very limited and scheduling time for testing our project proved very difficult. Thus, we put together the experimental setup shown in Figure 8.

Instead of the measuring the energy and subsequent spotsize of the FEL, we used a green Helium-Neon (HeNe) laser as shown in Figure 8. In the simulator the beam first passes through a fan which recreates the pulsatile nature of the FEL. Then, the beam is split with a piece of glass creating enough reflection to trigger the Energy Meter to take a reading. The PC in the setup controls both the MM3000 motion controller, but it also coordinates the Energy Meter data acquisition. This setup proved crucial to our project because it allowed us to work on the program as often as needed and on our own schedules. This simulation apparatus also proved to be extremely useful in determining the efficiency of the program. As can be seen from Figure 9, the HeNe set-up output a very noisy signal, yet the program was still able to curve fit the data and output a numerical spotsize.

RESULTS

Curve Fitting:

Over the course of our project we tested and perfected our program so that it was accurate and user-friendly. Figure 9 displays the output data on a test run of the HeNe simulation apparatus. Even though the data was extremely unstable, our program still was able to curve fit the data and output a reasonable value for the spotsize. In almost every instance the FEL is more stable than this particular signal and thus proving the program is more efficient than was actually required. Indeed, when the curve-fitting program was tested on the FEL seemed to perform efficiently and without error. Figure 10 shows a curve fit implementation on the FEL and a successful computation of the spotsize. One can see that even though there can be as much as a 10% fluctuation in the energy of the beam, the curve fitting function is able to make an approximation of the maximum power of the beam and fit the curve very smoothly.

Error Analysis:

After the completed program was completed and tested on the He-Ne simulator, we were able to get a beam shift on the FEL to test our program. As explained in the introduction, one of our constraints was that our system had to be able to measure spotsizes to an accuracy of 10 microns. However, what do we take the standard, or actual, values of the laser to be? We could take manual measurements or we could use the old system with external use of MATLAB and ORIGIN as the actual values. Since we were trying to replace the old system with a new one, we will be comparing our program’s results with results obtained with the old system. We did not take manual measurements of the FEL’s spotsize. During the shift we took several individual spotsizes to make sure that the program was working properly. Then, we executed the entire program for a short focal run of five spotsizes and then analyzed the same data with MATLAB and ORIGIN to examine how close our program reproduces the old method’s results. The results of this focal run are seen in Figure (11). We can see that our program reproduced results with an average error of 2.26 percent and a standard deviation of the error of 0.91. Notice that the new program sometimes outputs a bigger spotsize and at other times outputs a smaller spotsize than the old program. This is a natural byproduct of the erratic fluctuation of the FEL’s energy and the way the curve-fit converges on an answer and should not be considered unnatural.

Safety Issues:

The only safety issues relevant to this project are the dangers of undesirable tissue interactions between the skin and the laser or razorblade. The FEL is harmless coming out of the delivery arm. Since its diameter is about an inch, only a slight sensation of heat can be felt if you interrupt the beam in its path with a body part. However, once focused the beam becomes quite hot and dangerous. To avoid this, it is necessary to be aware of the beam path at all times and watch where you place your fingers or other extremities.

Often, a red HeNe laser is used in conjunction with the FEL to align the infrared beam. This way the FEL’s invisible beam path can be easily seen. The power output of the HeNe is so low that no harm can occur to the skin. However, the user must be careful to a void direct eye contact with this (or any other kind of) laser to avoid retinal damage. Safety goggles should be worn to avoid such dangers.

The edge of the razor blade is sharp to create a practical edge straight enough to uniformly slice the beam. The only time the razor should be handled is when it is attached it to the translation stage.

If unattenuated, the FEL’s power can be strong enough to cut through the razor blade in areas at or near the focal point of the beam’s path. If this were the case small sparks would begin to fly as the beam sliced through the moving blade. To avoid this situation, a polarizer is typically used such that a slight change in angle can attenuate the beam enough to correct this situation or avoid it entirely. Again, this can be avoided by the proper use of safety goggles.

Economics:

The economics of this project were reasonably low. There were no additional costs to Dr. Jansen in order for us to complete this project. All supplies were previously purchased. The only cost in this project was time. The engineering time used in brainstorming, developing, designing, implementing and testing. This required the assistance of a professor, a research assistant and two undergraduates. The actual FEL testing also required the help of the FEL technicians. As can be expected with any research or design project, the time invested is valuable and often not as abundant as desired. This is one of the largest benefits of having the Free Electron Laser Center located on the campus of a prestigious research institution such as Vanderbilt and located so closely to the University Hospital.