Donate now to keep alive!Help|Log In|Register|
Home PageMindat NewsThe Mindat ManualHistory of MindatCopyright StatusManagement TeamContact UsAdvertise on Mindat
Donate to MindatSponsor a PageSponsored PagesTop Available PagesMindat AdvertisersAdvertise on Mindat
The most common minerals on earthMineral PhotographyThe Elements and their Minerals
Minerals by PropertiesMinerals by ChemistryAdvanced Locality SearchRandom MineralRandom LocalitySearch by minIDLocalities Near MeSearch ArticlesSearch GlossaryMore Search Options
Search For:
Mineral Name:
Locality Name:
The Mindat ManualAdd a New PhotoRate PhotosLocality Edit ReportCoordinate Completion ReportAdd Glossary Item
StatisticsThe ElementsMember ListBooks & MagazinesMineral MuseumsMineral Shows & EventsThe Mindat DirectoryDevice Settings
Photo SearchPhoto GalleriesNew Photos TodayNew Photos YesterdayMembers' Photo GalleriesPast Photo of the Day Gallery

Construction and testing of a mini-XRF unit designed for mineral grain analysis

Last Updated: 16th Jun 2011

By Henry Barwood

Construction and testing of a vacuum mini-XRF unit designed for mineral grain analysis

In May of 2009 I acquired some “end of the year” research funds to upgrade the lab facilities at Troy. After much investigation, I decided to buy an Amptek X-123 X-ray Energy Dispersive spectrometer and a Mini-X X-ray generator, and try to build my own miniature XRF unit. The X-123 was configured with a 0.5 mil Beryllium window. I knew it was very fragile but it also offered the maximum sensitivity to low KeV radiation and the diode used was Peltier cooled so it required no liquid nitrogen cooling. The parts were purchased and delivered in December 2009. I unpacked everything and began mounting the two units on a base plate obtained from Amptek. I managed to perforate the diode’s Beryllium window within 2 hours of receiving it. I contacted Amptek and learned it would cost $2,500 for the repair (academic discount), which I did not have.

As luck would have it, I located some funds remaining on a Faculty Development Grant. It took 6 months, but I was able to pool those funds and some other money I had available and send the detector back for repair. It came back in July 2010 and I mounted it once again, with no damage this time. I spent 3 months building a sample holder and shield for the unit and in October 2910, I brought it online after some discussions with the Amptek software engineers. The software is not the “point and click” variety and you have to adjust various parameters to get it to work properly. The X-123 performed as advertised, but air absorption blocked emissions from elements below Potassium (around 3 KeV). While useful for research on metals, it was not sufficiently sensitive in the 0-3KeV range for mineral research. The unit languished in my lab for 6 months until I could design a vacuum enclosure for the X-ray source and the detector. Because they are mounted at a pre-determined angle, this proved quite difficult to accomplish without a fairly well stocked machining lab (which I decidedly do not have).

In April, 2011, I began work with some surplus sheets of 3/8 inch Plexiglass, a drill press, belt/disk sander and a table saw. My plan was to cut the vacuum chamber out of the Plexiglass so that ports for the X-ray and detector heads would fit into them. My method of construction was to keep trying until I got it correct. Holes were drilled in the Plexiglass and then fitted by sanding and cutting until they were at the correct angles and dimensions. The main forward wall of the chamber was cut to approximate dimensions and then bent to the correct (well, almost correct) angle by heating with a propane torch and bending by hand. Parts were glued with plastic cement and any leaks were sealed with Epoxy. Since I was building the thing as I thought of it, there was a lot of assembly followed by disassembly and correction.

By mid-April, 2011, I had the basic unit fabricated and began the process of molding seals and gaskets for it. The X-123 head was very carefully inserted in the port, pulled back slightly and a ring of RTV compound was spread about the base of the head. The assembly was then pressed back against the detection unit and allowed to vulcanize. Visual inspection showed a complete seal (at least it appeared that way). The Mini-X head was assembled in virtually the same way. After the gaskets had vulcanized, a quick pressure test revealed that there was a large hole in the X-123 gasket and the Mini-X gasket had not seated at all on the Plexiglass. The seal was so bad for the Mini-X that it could be removed entirely and the malformed gasket stripped away. The more complex and delicate X-123 head was repaired by cutting a section around the hole and filling it with fresh RTV compound. A rubber ring was pulled tightly around the Mini-X head and this was used to ensure that fresh RTV compound would actually be pressed into the opening for the X-ray head. Both solutions worked and a vacuum tight seal was achieved.

I tried to mold a gasket on the top plate that is removed to allow the sample to be inserted. Several attempts produced only marginal gaskets. I eventually learned to coat the Plexiglass plate with detergent as a mold release and was able to make a good, vacuum tight gasket. Once all this was completed, I hooked up a diaphragm pump and tried to pull a vacuum. The pump did not pull a good vacuum and I purchased a small, cheap vacuum pump (Chinese) used for air conditioning repair and replaced the diaphragm pump with it. I also decided I needed a vacuum release port and a vacuum gauge. It took a little over a week to find and hook up all this.

On May 7, after final fitting and checking, I drew a good vacuum. That evening I assembled the cords for the separate X-123 and Mini-X detector/generator power supplies. I hooked everything up and tried to turn on the Mini-X X-ray supply. That resulted in nothing but a warning window. I had to re-install the software and then it came on line. I got out my old Geiger counter so I could check to see how much scatter I was getting. Dead batteries! I checked my battery supply drawer and found AAA, AA and C, but no D (the size needed for the Geiger counter).

My patience gone, I decided to risk it and brought the supply up to 30KV and 15 uA. I thought I could locate the beam with a piece of fluorescent X-ray intensifier film. The proved impossible to do with the chamber configured the way I had built it. The exact composition of the phosphor screen was an unknown, but I decided to use the screen as a test sample. I collected a spectrum with no vacuum, and then turned on the vacuum. The gauge stayed at 0 inches of Mercury. I got a nice spectrum from the fluorescent screen but no low KeV peaks. The X-rays were turned off and I started checking the vacuum fittings. I quickly discovered that the valve I installed to shut off the vacuum from the pump had failed catastrophically. When I took the valve out so the vacuum fed straight through, and fired everything up again, the vacuum was fine.

At this point, I turned on the X-rays and again got strong peaks fluorescing off the screen. When I turned on the vacuum I got immediate low energy response. Because I was working without a safety net where the X-rays are concerned, I called it a night and shut down. The chamber worked, and all that was left was to locate the center of the X-ray beam using some Chinese dental X-ray film and to check where shielding would be required.

On 5/9/11 I mounted a small dental X-ray film on the sample holder and gave it a minimal exposure. These little packs are designed to develop almost instantly and can be ordered cheaply from China. They are advertised as being able to develop in full light, but I found that this is not the case. You do need a darkroom to open the individual pouches and inject the developer. Even with this I got a strong spot from the Mini-X after only 5 seconds exposure at 15 KV and 15 uA. I was glad that I checked the focus of the X-ray beam. It was actually about 4 mm lower than the center of the holder. This allowed me to make a mounting jig where I can insure that small grains are in the beam.

On 5/10/11 I decided that I needed the Geiger counter and after replacing the batteries and some fiddling around, I got it working again. I powered up, and found a fairly hot beam directly opposite the Mini-X source, so I shut down and got some Lead sheet I use for XRD work. Two layers of 1 mm sheet solved the problem. There is minimal scatter detected in any direction with the counter in direct contact with the chamber. Everything appeared to be good to go at this point.

Now came the hard part: Software. The ADMCA software that came with the X-123 will display the counts as they are acquired, but it will not do a peak search/match for specific elements in the spectrum. You also had to calibrate the X-123 before you could collect meaningful spectra. I assembled a test target with metal foil to use as known energy peaks. I selected Copper (8.05 KeV Ka) and Aluminum (1.49 KeV Ka) for the end points used for calibration. I also had to set the gain to produce a specific energy range. The range in KeV is determined by the gain. Higher gain means a narrower range. For a gain of 105 (maximum) you get a range of 0-10 KeV, for a gain of 55, you get 0-18 KeV. Lower gains will extend the range out even farther. The higher grains contribute noise, so it becomes a trade-off.

My testing was interrupted by a collecting trip to Arkansas. After returning, I worked with the software until I was able to collect calibrated spectra. They had lot of background, but were readable. Since I’m mainly concerned with identifying small grains, I tested a tiny piece of copper and got excellent peaks for the Alpha and Beta lines. Subsequent tests of some 0.5 mm flakes of “biotite” from the Arkansas syenites also showed identifiable peaks.

On 5/21 I ran the first “unknown”. Green prisms from some explosion breccias in the 3M Quarry in Arkansas that had been identified variously as epidote or diopside. A tiny grain about 0.5 mm was detached and examined. With a 15 KV 15 uA current and 3 minute collection time, I was able to resolve sharp peaks for Fe and Ca, thus indicating that it is likely epidote. Processing the data was a pain, but I was able to demonstrate that manual analysis of the chemical composition of microscopic grains was in reach. I ran several other samples inclusing a grain of "biotite" from the syenite pegmatite at Jones Mill Quarry.

My version of the ADMCA software does not run cleanly under Windows 7. I purchased the Windows 7 subroutine, Virtual XP, and was able to run ADMCA, the Mini-X controllers and XRS-FP . The XRS-FP program is a fundamental parameter quantitative software package that will analyze the raw data and return a lot of information, but I did not have time to learn how to use it at this point. Amptek provides W7 software for newer instruments, but it also requires a hardware upgrade, and I do not have the resources to have my unit upgraded.

On 5/23, I ran master calibrations on the various gain settings and reset some of the analytical thresholds to increase the unit’s sensitivity. Without adjustment of the background, I was able to identify Ca and Si peaks (and a lack of Al) in a sample of pectolite from Jones Mill Quarry in Arkansas. Another sample from the same quarry was tiny brown-yellow platelets on calcite/taeniolite matrix. An impure sample showed peaks for Ca and REE (unresolved La and Ce) and suggested that the crystals were either bastnaesite or parisite.

At this point I decided to rework the sample holder to accept small sheets of acetate film. I had used this technique before and it had very low background. This would also allow me to remove the sample intact and catalog it for later XRD/optical/etc. work. The Amptek software engineer also sent me instructions on how to set up a background subtraction routine in ADMCA. This method requires the collection and storage of background spectra for the various gain settings and then processing the live spectrum with the stored reference spectra. This method seems capable of extracting weak, low KeV peaks. My plans to test both areas were put on hold for a family vacation over the Memorial Day weekend.

On May 31, I made an adapter for the acetate sheets out of a Euro style micromount box. It fitted perfectly over the existing sample holder and required only cutting a slot and gluing into place. This formed a nice holder that an acetate sheet slid into. By gluing a tab on each square of acetate to attach a sample ID, the sample can be cataloged for future use. At least that was my original plan. It turned out that the thin acetate sheet I had was simply invisible against the plastic of the holder! I had to make a cardboard tab with a hole in it to mount the acetate so that the sample could reasonably be manipulated into the holder. Investigations with various papers for the sample holders showed that most of them contained Calcium Carbonate or Titanium Oxide whiteners that eliminated them as sample holder candidates. Vellum was found to have a low background; however, some varieties contained sulfate that interfered with the sample spectrum. Eventually, a composite holder was fabricated using a large hole in regular card stock with an acetate holder cemented to the opening. This removed the paper from the target area of the X-ray source and eliminated the coatings interferences.

Tests were performed over the next week using various target materials to learn the software, and to test the calibration and sensitivity of the X-123 detector. The vacuum system fabricated was designed to allow detection of elements down to Sodium on the periodic chart. The X-123 diode used a 0.5 mil Be window that is, technically, sensitive to these lower energies.

Various samples, including table salt, were examined. The lowest conclusively detectable energy was Aluminum at 1.49 KeV. Even with background subtraction and tuning of the MCA parameters, it was not possible to unquestionably detect elements below this level. Elements from Al to Cl are detectable with background subtraction and detector tuning. The sensitivity of the diode above 3 KeV is excellent and grains in the sub-millimeter range can easily be qualitatively analyzed from K to U. The loss of Na and Mg is a problem; however, many minerals can be identified by a combination of the detectable elements and their other properties.

The Amptek system is a useful piece of equipment, but not quite as useful as I had envisioned when I started this project. Once mastered, the ADMCA software provides:

Background subtraction
Calibration of the energies of the peaks
The ability to compare spectra from several samples simultaneously

There is not an easy way to search the peaks to determine the elements present and some knowledge of X-ray spectroscopy is needed to interpret the analyses.

At this point I decided to check additional low background sample holder materials. I ran a variety of papers from Vellum to card stock and found that all of them had coatings or additives that contributed additional lines to the background. One of the best mounting materials I discovered was the plastic used for thermal laminating pouches. It had zero spurious peaks, and was an easy material to handle. I also purchased some thicker acetate sheet that could be used with minimal background interference. The best glue I discovered was simple tacky white glue that does not contribute significantly to mounted mineral grains.

Searches via the Internet and published literature showed that there are two freeware XRF programs that are capable of manipulating the data saved from ADMCA (.mca files). These programs are PyMCA and DTSA-II. Both run under Windows 7 operated in 32 bit mode. PyMCA will load the .mca files directly and retain the embedded calibration. It will not easily perform search-match. DTSA-II will import ,mca files, but does not load the calibration. I have contacted the developer to determine how this bug can be fixed. The DTSA-II software will output publication quality files and will also perform rudimentary search- match operations. My experimentation with both these software packages will be the subject of a future report.

Henry Barwood

Article has been viewed at least 14333 times.


Good Luck Henry, great article. I would love to build one of those. Alfred Nier is one my favorites. A table-top XRF is a great tool to have. NASA's Rovers are another marvel of modern technology being used to ID minerals. I'd really like to go there in person though and dig myself... LOL... CHEERS.

Sam Cordero, Jr.
16th Jun 2011 3:43am
Hi Sam,

I had hoped for a greater analytical range. The diode is quite sensitive and can detect most elements with ease, but the lack of detectability below about 1.5 KeV is frustrating. Doubt I'll ever collect minerals on Mars, but it would be interesting to see what is there!

Henry Barwood
16th Jun 2011 12:14pm
I wonder if some handmade paper would work as a holder. It is a whole lot less likely to have an opacifier added to it.

David Von Bargen
17th Jun 2011 10:56am
Hi David,

Good suggestion. At this point, the frosted plastic sheet from the laminating pouches is working great. A single package of the stuff will provide enough mounts for the next 2-3 years. I've been busy separating minerals to analyze and hope to have some posts on results soon.

Henry Barwood
17th Jun 2011 3:28pm

In order to leave comments to this article, you must be registered
Mineral and/or Locality is an outreach project of the Hudson Institute of Mineralogy, a 501(c)(3) not-for-profit organization.
Copyright © and the Hudson Institute of Mineralogy 1993-2017, except where stated. relies on the contributions of thousands of members and supporters.
Privacy Policy - Terms & Conditions - Contact Us Current server date and time: December 14, 2017 09:54:58
Go to top of page