Non-dispersive x-ray fluorescence (NDXRF) got its start in the 1920’s when Ross and other experimenters discovered that they could isolate an x-ray line for an element by using two filters made of different elements over two detectors. One filter absorbs the elements x-rays, while the other transmits them.
The difference in counts between the two matched detectors with balanced filters is the net intensity and is related to that elements concentration. When combined with earlier work that demonstrated that elements could be measured by measuring total x-ray intensities from some simple samples, a new and powerful method was born. Unfortunately it was almost 50 years later when small microprocessor based analyzers were built in the 1970’s that NDXRF started to make a commercial impact.
NDXRF has the least expensive hardware of any of the XRF methods, because it only requires a few low costs components. It needs an x-ray source, usually either a radioisotope such as Fe-55, Cd-109, Cm-244, Am-241 of Co-57, or a small x-ray tube. And it requires a detector such as an ionization chamber or Geiger-Mueller Counter, which does not need to be energy dispersive. While Ross used two detectors, the more common approach is to use a single detector and use a filter wheel or tray to position the filters over the detector in sequence. In addition to the Ross method a single filter (Hull Method) or no filter at all may be used to measure some elements..
In commercial devices it is most common to see a proportional counter used as the detector since it is a low resolution EDXRF detector. The advantage of a proportional counter is that it can be configured to not count the backscattered source x-rays making the overall background counts substantially lower. At the same time a proportional counter instrument may be used for EDXRF analysis, making it a hybrid EDX/NDX instrument. With x-ray tube source devices, x-ray tube filters may be used in combination with specially selected target anodes to produce optimal sources for exciting the elements in a sample. The non-dispersive XRF method is very powerful, and cases where an appropriate filter pair exists and can successful isolate an elements wavelength it is often possible to match the performance of a WDX analyzer at a tenth the cost using 100 times less source intensity. .
One of the most common applications is measuring phosphorus, sulfur and chlorine in oil. Generally either a Fe-55 radioisotope, or an x-ray tube with either a Pd, Ag, or Ti target is used to excite those elements. By looking at the absorption edges for various materials it is easily seen that chlorine has an absorption edge above chlorine in energy, sulfur has and absorption edge between chlorine and sulfur, and phosphorus has an absorption edge between sulfur and phosphorus. X-rays do not readily excite hydrogen and carbon in the base matrix and the detector windows readily absorb their x-rays, and so they aren’t measured.
In this case chlorine can be measured by using chlorine as a transmitting filter and sulfur as an absorbing filter. The difference between the counts of x-rays off the oil sample and through the filter correlates to the chlorine concentration. The filers are electronically balanced by measuring the intensity on a blank and introducing a coefficient that when multiplied by one intensity yields zero net count. Similarly a sulfur and phosphorus pair of filters can be used to measure sulfur. A single filter can be used to measure phosphorus since there are usually no measurable elements below phosphorus that would produce counts.
The matter is confused somewhat because it is difficult to produce good sulfur and phosphorus filters, so usually an element with an L absorption edge at the appropriate energy is used instead, Mo or Nb for S, and Zr for P. Since the heavy metals are denser the filters are usually much thinner than their K absorption edge counterparts.