The prelab (with unique parameters for each individual lab group section) is REQUIRED to be completed by each student and submitted on the date specified (and at a time PRIOR to the scheduled start of the lab session). Students who fail to submit the prelab or submit it late (late prelab submissions are worth ZERO) will be penalized -20% on the lab. If all students in the entire group fail to submit a complete design with a predicted spectrum matching the required design, the entire group will not be admitted into the lab, will not be able to fabricate an MDM device, will be marked as absent, and all members of that lab group will forfeit 50% of their marks in this lab immediately! NO EXCEPTIONS!
Typical of the work performed by a vacuum technician in industry, the tooling factors for a crystal depostion monitor will be determined by using optical means
In this lab, optical interference filters will be fabricated in order to calibrate a crystal deposition monitor to various locations in a thermal deposition chamber. Because of the geometry of the chamber, substrates closer to the deposition filament than the monitor crystal will receive a thicker deposit than is read on the monitor and those further away will receive a thinner deposit. In terms of filter performance, thicker dielectrics will be shifter towards longer wavelengths (red) than the design value and thinner dielectrics will be shifted towards shorter (violet) wavelengths. A unique tooling factor is hence required for each location in the chamber to ensure that a filter produced at that specific location is accurate - this procedure is required when a new monitor is installed in any chamber, or when the geometry of the chamber (e.g. location of the monitor or substrate holder) is changed ... tooling factor, unlike other parameters, is not something that can be found in a book: it must be experimentally determined. It is unique for every installation and every vacuum tech has performed this procedure.
The geometry of the chamber, and the monitor, is approximately as follows:
So, substrate #2 is further away from the deposition monitor crystal and so will receive a thinner layer (and be spectroscopically violet-shifted) than substrate #1 next to the monitor crystal itself ... not that the tooling factor given is entirely accurate for _that_ location either (!).
In this lab you will also use commercial thin-film design software (FilmStar from FTG Software) to model the filters and determine thicknesses of filter layers
The PRELAB is due, at latest, fifteen minutes before the lab: as a group, along with the lab instructor, you will choose one design to fabricate. If prelabs are not submitted by PRIOR to fifteen minutes before the scheduled start of the lab period, it is overdue and is worth ZERO. Print it out LONG before the lab and DO NOT come late to the lab because you were printing the prelab ... this is unacceptable and will result in an immediate ZERO on the prelab regardless - LATE ARRIVALS are NOT accepted!
Design the filter based on your specific lab group as follows:
|Lab Group||Lab Date||Center Wavelength (nm)||Aluminum Thickness (Å) Each layer||Initial TFi|
|Wednesday 10:30 Group A||2018/04/28||520||250||120.0|
|Thursday 9:30 Group A||2018/04/29||530||250||120.0|
|Wednesday 12:30 Group A||2018/04/28||515||250||120.0|
FAILURE TO USE THE CORRECT WAVELENGTH IN YOUR FILMSTAR DESIGNS WILL RESULT IN AUTOMATIC FAILURE OF THE LAB - If the wavelength in your designs does not match that of your assigned lab group it will be considered plagiarism and a mark of zero will result for the entire lab
Filmstar models MUST BE UNIQUE TO EACH STUDENT. Plagiarized models (including the prelab) will result in a mark of ZERO for the entire lab and a trip to the Chair's office for academic dishonesty resulting in possible expulsion.
Failure to submit this prelab on the date and time shown above, or one without the name embedded right into the graph, will result in an immediate -20% penalty on the entire lab mark. NO LATE SUBMISSIONS OF THE PRELAB will be accepted. Students arriving even ONE MINUTE late will be assigned a mark of ZERO on the prelab! To be safe, submit the prelab EARLY to your lab instructor.
A major component of this lab is the use of FilmStar, a commercial thin-film design package from FTG software. You are required to use FilmStar to design the initial filter as well as determine the actual dielectric layer thickness from spectroscopic results.
Download and install the FilmStar package - it has been tested under compatibility mode (XP SP3) up to Windows 10 64-bit. If you are unable to run it on your own PC, you may run it on college PC's in V14A. The installation password is a second, password-protected file within the installation zip file provided. After installation of the package, you will find a number of example designs which come with FilmStar. The easiest way to start is with this FAW file (click to download) which outlines a basic design for an MDM filter with a transmission of 15% at 500nm - you must modify the file for your actual device to be constructed. Ensure FilmStar is in Film Archive mode (from the File Menu) and open this file. Press "Calculate" in the lower right corner, select a wavelength of 500nm, and view the spectrographic curve for the filter. The Layers Editor (in the Design menu) will detail the required thickness of each layer: in the basic design given the first layer of aluminum (T) is 123.9 Angstroms, the second layer of cryolite (F) is 0.91449 wavelengths (3313.4 Angstroms when displayed as a Physical Thickness - from the design menu), and the final layer of aluminum (T) is 118.8 Angstroms. When a design is complete, a report may be generated via the Report Generator on the file menu - for this initial model you may view the report here which outlines all required layer thicknesses. Before submission of the prelab, display all thicknesses in PHYSICAL (nm) units since the monitor used in the cleanroom lab works in physical units.
Note #1: that the aluminum layer is in PHYSICAL thickness (nm) while the dielectric thicknesses are in OPTICAL thickness (with a Quarter-wave optical thickness appearing as '0.25' - this is λ/n): convert both to Angstroms before the lab (FilmStar can do this via a menu function - Design menu). This is required prior to the lab since our deposition monitor uses units of Angstroms.
Note #2: When using the evaluation version of FilmStar, use an 'F' layer for sodium-hex ('F' is for MgF2, n=1.38, which is very close to to the actual value of 1.33). The evaluation version supports only a few materials ("Show Indices" from the Layers Editor).
Note #3: For a review of the basics of filter theory, see the notes on this course home page as well as Chapter 6 of FUndamentals of Light Sources and Lasers by Csele. Thicker, more reflective, aluminum layers will lead to higher Finesse and hence a sharper spectral peak (making determination of that wavelength easier) however % transmission will be reduced as those layers become thicker.
For the final lab report, you will be required to include an unedited FilmStar REPORT outlining the initial design with your target design wavelength (as fabricated in the lab) as well as 'corrected' designs which match the results achieved in the lab. From the FILE menu, select REPORT to generate this document - it will contain the predicted spectroscopic output as well.
To modify the design you may either tediously manually edit the layers in a trial-and-error manner (and recalculate each time until the spectroscopic peak is within 1% of both the amplitude and wavelength required),use the built-in optimize feature, or use the built-in interactor (preferred). If using the optimize feature, begin by setting the target wavelength in the Calculate option (again plotting the current design) then, from the "Optimize" menu, select a target for your new design (e.g. Target Type = Equality, Type = % Transmission, Target Wavelength = Your Design Wavelength, Transmission, Tolerance = 1%, Polarization = Random). Set the variables to be used in the optimization (Variables). Now, select the "Optimize" function and run at least 100 iterations. Calculate again and see how the curve has changed - hopefully it matches your required design - and updated layer thicknesses may now be determined from the layers editor. It may be necessary to perform the optimization in two steps optimizing the dielectric layer (2) first then the aluminum layers (1 and 3) next.
The interactor, as shown briefly in the lecture, is preferred as it allows you to see what happens when a layer is made thicker (i.e. 'what if' scenarios). To use the interactor simply select 'Interactor' mode from the 'Evaluate' menu. Generally, you optimize either the dielectric layer or the two aluminum layers (together) alternately ... one affects the other! When done, deselect the 'Interactor' mode, convert to physical thicknesses, and print the report as usual.
The reverse process is also possible: given the spectrographic response of a fabricated filter, one may use FilmStar to determine the actual dielectric thickness of the filter to give this response. Assuming you know the wavelength peak of a filter from a spectrophotometer you may use this as your new design wavelength to determine the actual thickness - this is required while analyzing substrates in the lab.
NOTE: This lab will be completed on the Bendix high vacuum evaporator system in which layer thickness will be monitored using an Inficon XTM/2. Density, Z-Ratio, and Tooling Factor must be programmed into the unit for each material used.Filament Configuration (Current as of 2016/03)
Review the MSDS Safety Sheet for cryolite.
This metal-dielectric Fabry-Perot filter will be fabricated using aluminum and sodium-hexafluoroaluminate (Na3AlF6, also called 'cryolite') with n = 1.33 and density = 2.9 g/cm3. This material will be evaporated using a very tight helix able to hold the loose powder.
Filter Design Reference: Thin Film Optical Filters, H. Angus Macleod, IOP, London, ISBN 0-7503-0688-2 (2001). Chapters 5 and 7.
Prepare the vacuum system for deposition. The diffusion pump will be preheated in advance of the lab to allow degassing of the fluid and ensure proper operation of the pump. If required, clean six substrates (these substrates have never been touched by human hands so a simple wipe with methanol should suffice if required at all).
Load the substrates into the round holders ensuring two are adjacent to the monitor crystal and several others are located at various distances from the dielectric source filament, in a line from the cryolite filament. At least SIX substrates must be loaded and used for analysis.
Ensure the crystal oscillator is running and in-range by checking the XTAL indicator on the Inficon XTM. Press the TC/CHK button on the Inficon to see how much life has elapsed on the crystal. Load two filaments with aluminum: one primary and one backup in the event of failure. Two aluminum layers of under 300 Angstroms each are required so large amounts are not required. Load another filament with sodium hexafluoroaluminate powder. This material is toxic (see the MSDS safety sheet via the link on this page)! Put a clean-room rag under the filament and press the powder gently into the filament to load it. The entire filament must be full of powder to ensure a decent supply is available. Dispose of the rag and change your gloves immediately. Finally, if magnesium fluoride (MgF2) is to be used as an overcoat (the configuration varies year-to-year) to prevent oxidation of the filter layers then this source must also be loaded.
Pump the vacuum system down (refer to the previous lab) being sure to fill the cold trap as required before crossover. When ultimate high vacuum has been achieved (Better than 8*10-6 torr) set the thin film monitor for the INCORRECT tooling factor of '152.0%' or '132.0%' and deposit a layer of aluminum, a layer of Na3AlF6, and another layer of aluminum as per your original design. If configured for this, and time allowing, complete the filter with a thin (200 Angstrom) layer of magnesium fluoride to protect the outer aluminum layer from oxidation and mechanical damage during handling - this will also demonstrate a limitation of the thermal deposition technique. Before each deposit of aluminum, and with the shutter closed, ramp the filament current SLOWLY to both outgas that filament and heat it slowly to avoid thermal stress. For the dielectric, outgas the Na3AlF6 filament gently (VERY gently – it should barely glow) as well. Be sure to program the Inficon XTM monitor's density, z-ratio, and tooling factor before each deposit and zero the display.
After each layer of material is deposited note the final layer thickness which was ACTUALLY deposited (It is quite likely that the actual deposit will vary somewhat from the design target). The actual thickness of the cryolite layer, as read from the monitor during deposition, becomes "Tx" in the calcualtions required (see below). After all layers are deposited bring the chamber back to atmosphere, note the locations of each substrate as each is removed, and remove the substrates.
Quantitative analysis requires the use of a spectrophotometer. Using the Lambda-3B spectrophotometer in the spectroscopy lab (V14A) determine the center frequency of each filter. Follow the Instructions given in the SOP to scan the transmission curve of each filter. Install a glass blank in the reference beam (you should know why this is required). For each filter, capture the spectrum data and include a transmission plot (generated using ExcelTM of each filter in your lab report. Be sure to scan the entire range from 350 to 900nm for each filter and record the wavelength of maximum transmission and the percentage transmission at that specific wavelength for each filter.
As an alternative to the Lambda-3B, the GeneSys-20 spectrophotometer can also be used to determine the transmission of the filters. Custom software allows scanning of the background (with a glass blank installed) then successive scanning of filters. Each scan takes approximately eight minutes to complete. Instructions will be provided in the lab if use of the spectrophotometer is necessary.
You may well obtain multiple peaks in the spectrum scanned - is this expected based on the discussions in class (see the transmission spectrum of original design)?
It is the responsibility of students in the group to analyze the filters in V14A outside of your regular lab time. Time is available during any of the professor's office hours as well as during other lab sections (i.e. while a different group is in the cleanroom, you may utilize V14A for analysis). Failure to complete spectroscopic analysis on your own time is not an excuse for failure to complete the lab.
Now that you have the spectrum of each filter (eight minimum), analyze each filter by changing the FilmStar model slightly so that the center wavelength predicted by the changed model matches that actually found for your filter - this will determine the actual thickness of the dielectric layer. You may modify the model manually (by changing the thickness of the dielectric layer in the layer editor - Design menu - and recalculating the output to see if it matches the wavelength of maximum transmission) or through use the interactor option (by far the easiest). Alternately, optimization targets may be selected in the OPTIMIZE menu. (See the notes above on using the FilmStar package). Follow the basic procedure outlined in the lecture to match the wavelengths of the major transmission peaks. The actual percentage transmission may not be an exact match since the tooling factor of the aluminum layer will also change by position, but the wavelength of maximum transmission must be a match within 1% ... and as noted in the lecture secondary peaks (representing 1/2 λ and 3/2 λ) must match as well.
Remember that each filter is an interferometer and so an integral number of "half wavelengths" must fit into the cavity. It is possible, with filters far from the monitor, that the deposit will be so thin the center wavelength is pushed into the UV portion of the spectrum and so will not be observed however the broader "half wavelength" peak will appear ... you can see this for yourself by using the interactor on FilmStar to make the central layer thin (try half the initial thickness). Watch as the "design" peak (full wavelength) moves into the UV beyond recordable range. You must logic this one out based on position in the chamber (as demonstrated in the lecture).
You need a 'corrected' FilmStar model (one in which the wavelength peak for the model matches that found using the spectrophotometer) for each filter analyzed (at absolute minimum at least three but very likely eight as discussed in the lecture). The only excuse for not analyzing a filter is a complete lack of peaks in the entire spectrum scanned in which case a spectrophotometer output will still be required in the report to shown this. From each FilmStar model, the physical thickness becomes "Tm" in the calculations outlined below. Print/Save a FilmStar report for each as this is required for the report.
Armed with the actual physical thickness of the dielectric layers for various substrates, as determined spectroscopically, and knowing the thickness as read from the monitor (ideally, the same as the target thickness from the original FilmStar design but possibly different), proceed to compute the tooling factor for the XTM monitor for each substrate location as follows:
In this excerpt from the manual, Tfi=152% and Tm the thickness of the dielectric layer determined using FilmStar (i.e. when the model is corrected for the actual center wavelength found).
Calculate the corrected tooling factor for ALL substrates that show a peak within the measurable range of the spectrophotometer - at least three substrates (one central near the monitor crystal) are required but you will likely have more. In each case, be sure to generate a FilmStar REPORT which shows the center wavelength corrected to match the spectroscopic results observed and the actual thickness of the dielectric layer required to yield those results. For each substrate NOT analyzed (six substrates were required, minimum), explain WHY it was not possible (the only excuse is where the center wavelength falls outside values which may be analyzed - detail the center wavelengths found for each of those substrates ... if a substrate is omitted without just cause marks will be deducted since laziness is not an excuse for not analyzing a substrate). Again, include a graph of each observed transmission spectrum for each filter.
Be sure the tooling factors are LOGICAL and increase/decrease as the distance from the source filaments do as expected. As well, the range of tooling factors will range from about 70% to 150%. If a factor is found outside this range consider if you have the correct peak (FSR) for the filter!
Prepare a formal, word-processed, report (submitted in a folder) covering the lab as follows:
FilmStar models (for analysis) are to be completed individually. All members of a lab group will have the same spectroscopic data but the models WILL differ depending on how the optimizations were completed. Do NOT share FilmStar models as this will be considered plagiarism. It is expected that ALL students will have unique models and slightly different results!
Outline the basic procedure followed to deposit the filter including key parameters such as pressure. Do not detail "what knob was turned when" but simply outline the basics such as "the chamber was pumped to ___ torr" and "a layer of cryolite was deposited to a thickness of ___ as read on the deposition monitor". Include values programmed into the XTM monitor as well. This procedure should be complete enough that someone with your procedure, along with the SOP for the machine, should be able to duplicate your results precisely. Test results are only valid if reproducible!
Outline the analysis of each filter and the results. If you have eight filters, several pages (grouped together) will be required for each filter consisting of:
You will have eight sections (one per filter), each containing several pages as outlined above. Duplicate the diagram showing the location of the substrate for each substrate involved. For substrates NOT analyzed (i.e. if there were no spectroscopic peaks found), provide spectrum results and explain WHY each was rejected for use in this report! Note: if a filter shows a complete peak but was not analyzed a penalty will be applied: analyze all filters which show a reasonable full- or half-wave peak.
For the central substrate (the first one analyzed), SHOW ALL WORK for all steps and be explicit in outlining the METHODOLODY used - for example outline the calculation of actual thickness from the center wavelength found. EXPLAIN the procedure followed in WORDS allowing a user who has never performed this analysis before to follow it! Assuming the first substrate analyzed was done explicitly, the other substrates may be analyzed in a more abbreviated manner.
WARNING: To ensure FilmStar models are completed by each student and are not plagiarized (which will result in a mark of ZERO for the entire lab), print/capture the report directly WITHOUT EDITING. The FilmStar report must appear as it would directly from the FilmStar package ... text, graphics, or any parameter appearing on the report must be original and unedited. It may, however, be scaled to fit on a page with other text (i.e. cut/past of the entire report without editing is allowed)
CLEARLY Summarize the new experimentally-found tooling factors and the corresponding locations in the chamber with a diagram of the substrate holder (showing NORTH as the operator position) and a corresponding chart (e.g. show 'N' on the diagram of the substrate holder and in an accompanying chart show both the theoretical and experimental tooling factors for that substrate). This large diagram would serve as the operator's calibration chart allowing the operator to perform accurate depositions at any location in the chamber (for this one specific filament at least).
As well as a summary diagram (above), a summary chart of all results (for all substrates) must be provided with the following columns: distance to the filament, expected tooling factor (based on r-squared calculations), peak wavelength measured (on the spectrophotometer), thickness of the cryolite layer (from the FilmStar model), and calculated tooling factor.
The r-squared calculations are simply a ratio of the distance to the monitor, squared, divided by the distance to the substrate, squared all multiplied by the old tooling factor. Substrates closer to the filament than the monitor will have larger tooling factors, substrates further from the filament than the monitor will have smaller tooling factors. This same general pattern should be found with your optically-calibrated results.
Report OrganizationA typical report, then, will consist of the following:
FINALLY ... examine the marking scheme for the lab (found on the course page). This will also serves as a checklist to ensure you have done everything required for full marks.