Energy Dispersive X-Ray Fluorescence Applications in the Surface Finishing Industry By A. R. Overman ASOMA INSTRUMENTS, INC.

Transcription

1 Energy Dispersive X-Ray Fluorescence Applications in the Surface Finishing ndustry By A. R. Overman ASOMA NSTRUMENTS, NC. 277J ~J Energy Dispersive X-Ray Fluorescence (EDXRF) is a very powerful and simple tool for the elemental analysis of a variety of surface finishing processes. This paper will address the basic theory of EDXRF analysis and typical at-line and On-Line instrumentation. EDXRF analysis can be performed in many stages of the surface finishing process, such as during the actual plating of the product or after a product is coated or anodized. Typical results from application studies of a large number of plating solutions and coatings show that EDXRF provides excellent analytical performance for a wide variety of surface finishing applications. Coatings must often meet strict specifications in order to ensure that it meets the end users performance requirements. f excessive coating materials are applied, then coating material is wasted and the product is out of specification. f the coating weight is too small, then the coating does not have the properties desired by the customer, such as corrosion resistance. n addition, for plating processes one can monitor the solution to prevent bath depletion, extend bath life, and lower costs, while ensuring a quality product. To monitor coating and plating processes, the analytical method employed should satisfy the following criteria: it must be accurate, fast, simple, require minimal sample preparation, and bc on-site to provide ti me1 y resu ts. Energy dispersive x-ray fluoresccnce (EDXRF) satisfies all of these requirements. Analytical precision is typically on the order of 1% rclative with an analysis time of 5 minutes or less. EDXRF is a very simple analytical technique that can analyze a variety of elements ranging from Mg to U in liquids, solids, powders, and coatings from ppm to % concentrations at a touch of a button. To analyze a plating solution, the operator pours the liquid into a sample pan, places it in the analyzer, and initiates the analysis, normally with just one button. Coatcd samples are cut into coupons and placcd in the benchtop analyzer or a remote analysis head is placed directly over the sample. Because no sample preparation is required, EDXRF can be easily adapted for on-line analysis of coatings and plating solutions. With on-line analysis, one can obtain real-time measurements to more tightly control the process. BASC THEORY n the simplified model of the Bohr atom1, the electrons orbit around the nucleus at discrete energies(figure 1). The electrons in the innermost orbit (the K shell) have the lowest energy. The electrons in the L shell have a slightly higher energy than those in the K shell, and so forth. The electrons will occupy the lowest energy orbital possible, therefore vacancies at the lower energies are filled first. n order for an electron to escape from an atom, it must have more energy than that of the outermost orbitals in the atom. f an incoming x-ray has sufficient energy to allow an electron to escape, the electron will absorb the x-ray and it is ejected, resulting in a vacancy in this orbital. An electron from a higher orbital will "fall" to fill this vacancy. For this to occur the electron must give up its extra energy in the form of an x-ray. 1 XRF Phenomenon L 1 Fig. 1 - Production of an x-ray. 1

2 Since the energy diffcrences between cach shell is unique to each element, by identifying the energy of the x-ray excited, one can identify the element. Low atomic numbered elements, such as Si and S, will emit lower energy x-rays. Higher atomic numbered elements, such as Au and U, will emit higher energy x-rays. t is the detection of these characteristic x-rays that is the basis of x-ray fluorescence. Bonding in molecules occurs at the outermost orbitals. Since XRF takes place in the innermost orbitals, (the K and L shells), the x-rays emitted are independent of chemical bonding, and thus, the chemical form. For example, the Cr atoms in hexavalent and trivalent chrome will emit identical x- rays and are indistinguishable EDXRF NSTRUMENTATON There are two classes of EDXRF analyzers available: At-Line and On-Line. At-Line analyzers provide the highest precision for a single sample resulting a snapshot of the process. On the other hand, on-line analyzers provide a full history of the process, establishing and monitoring trends and predicting process upsets. A t-line Analyzers All x-rays may be discerned by wavelength or energy. Energy dispersive x-ray fluorescence measures the energy of the x-ray in kilo electron volts (KeV) to identify the element. To excite the sample and detect the resulting x-rays, each analyzer needs a source of x-rays and a method of detection. Once the x-ray is detected it is sorted and stored in a multichannel analyzer (MCA) and processed by the built-in computer to provide a quantitative result. n addition, some analyzers also include special x & y staging for spot analysis. Spot analyzers are not addressed in this paper. Sample _ ntensity Electronics, MCA, and Processor Fig. 2 - Typical EDXKF analyzer. Energy (KeV) The source of x-rays may either be a radioisotope or an x-ray tube. Radioisotopes (Table 1) provide a very stable, fixed intensity of nearly monochromatic x-rays. Although radioisotopes decay, this is a predictable behavior which is compensated for by the software in the analyzer. X-ray tubes are adjustable to get optimized excitation. n benchtop instrumentation, the x-ray tube voltage is usually limited to 30kV and hence they do not excite the higher energy elements that 241Am can. For lighter elements, x-ray tube systems typically provide the best analytical performance available. Table 1 Radioisotopes Used n EDXRF sotope Half-Life Radiation Elemental The EDXRF analyzers in these studies are equipped with a proportional counter to detect the x- rays. When the x-rays enter the counter, they ionize the gas, and the resulting ions are accelerated to create a pulse. The intensity of the pulse determines the energy of the x-rays. There are several advantages to using proportional counters: they can handle higher count rates for improved precision; they do not require liquid nitrogen cooling; they are portable; and they are inexpensive. Once the pulse is created, it is amplified and the energy is discriminated. The multi-channel analyzer (MCA) is used to store and sort the x-rays by energy. Use of an MCA allows the analyzer to simultaneously detect the different elements in a sample. When the analysis is complete, a computer processes the spectra to get a net count rate for the elements. f an analyzer has been calibrated, the net count rate is automatically used to calculate the concentration or coating weight of the element. On-Lirie AHO.V=& Because of it's nondestructive nature, EDXRF is ideal for on-line analysis. ndustries seeking S09000 certification are looking towards on-line analysis to improve quality control in the process. On-line analyzers include all of the components of atline equipment with the addition of a sample 28 2

3 presentation system and a variety of continuous data outputs for process control. Data outputs are available in the form of RS232, RS422, 4-20 ma, and 0-1OV analog. These results may in turn be used to control coating or plating process. Most coating and plating applications meet the sample requirements for on-line analysis. The samples should be uniform, with little matrix variation; there should be relatively few elements to be analyzed; and concentrations should be higher than 50 ppm, depending on the element to be analyzed. Coatings are the simplest application of online analysis. A remote analysis head may be positioned over the coil at a constant distance from the sample. The position may be fixed or scanning depending on the type of system used. For plating solution analysis, a representative stream of the sample is sent through a flowcell at a low pressure, typically 5 psi. The flowcell is a chamber with a measurement window in contact with the analysis head. The window used in the flowcell is one of the more critical components of the system. f the window is not strong enough, it will not withstand the pressures of the stream. f the flowcell is too thick, it absorbs the x-rays reducing performance for the analysis. For typical streams at 5 psi, the flowcell window is thin enough to analyze elements from P to U. f lighter elements need to be analyzed, special windowless systems may be used to present the sample to the analysis head. An on-line analyzer may be equipped to analyze one stream or up to 16 different streams for a large number of plating solutions. Frequently two cabinets are used for multi-stream systems. Onc cabinet houses the measuremcnt flow cells and the other houses the electronics. For remole control of several process, the measurement flow cell may be located convenient to several plating solutions, and the electronics cabinet can be as far away as 30m for remote control. Because the distance between a plating solution and the cabinet may be large, the sample must reach the analyzer quickly in order to be representative of the process. n these instances, a fast pass loop may be employed. The fast pass loop takes the stream at a high flow rate, 8 to 20 pm, so the sample reaches the analyzer quickly and is still representative of the process. Once the stream reaches the analyzer, a variable restriction forces a small fraction of the sample through the x-ray flow cell for analysis at a low pressure and flow rate, typically 1 pm. Plating solutions are normally corrosive, and hence present many sampling challenges. The sampling system should be resistant to the acid, not plate out, and the analysis cabinet should be able to withstand the environment of the plating shop. Teflon and acid resistant tubing and flow cells are used in the analyzer to prevent plating out in the line. n cases where salts may precipitate at lower temperatures, the lines may be heated. There are several types of enclosures available for on-line systems: Nema 12,4, or 4X. These enclosures are splash resistant, watertight, and water-tightkorrosion proof, respectively. n addition, these enclosures may be purged to prevent corrosive fumes from damaging the analytical equipement. Leak detectors and automatic shutoff valves may be included to detect spills in the analysis chamber and stop the flow. QUANTTATVE ANALYSS Saniple Reqtrirenrents Before a quantitative analysis is performed, the analyzer must be calibrated with standards representative of the sample matrix. f the standards used to calibrate are not of the same matrix as the samples to be analyzed, then a bias from the matrix is introduced. Also these standards must be well assayed, since the error in the assay will add to the uncertainty resulting in an inferior calibration. n addition, sampling errors are introduced, if the process is not homogenous and samples are not representative. For best results, the standards should evenly cover the anticipated concentration range. Spectral Deconvohitions Before the net x-rays of the element of interest (EO) are used for calculation of concentration, the spectrum must be processed to remove unwanted x-rays. The unwanted x-rays have two common forms: backscatter and overlap from ncarby elements. Backscatter refers to the x-rays that originate from the radioisotope or x-ray tube and are reflected back from the sample with varying energies. To correct for this, a sample with none of the analyte present is analyzed. The analyzer then automatically determines a relationship between the number of x- rays that have the same energy as the EO1 and the number of x-rays in another designated backscatter region in the spectrum. This ratio is then used to strip the backscatter from the EO1 region of the spectrum. f there are elements of similar x-ray energies present in the sample, these elements are also excited. Often, the peaks of adjacent or interfering elements will overlap affecting the count rate for the analyte. To correct for this, a sample containing only the EO1 and another sample containing only the interfering element are analyzed separately. The overlap between 3 29

4 ~. Concamtion Fig. 3 - Net x-ray intensity versus coat weight or concentration. the elements is then deconvoluted automatically by the analyzer software to provide the net number of x-rays from the EO. Net X-ray ntensity vs. Concentration or Coating Weight At low concentrations, the intensity of the x- ray is proportional to the concentration. As the concentration or coat weight increases, less and less x- rays are transmitted to the detector and the intensity approaches a horizontal asymptote representing the maximum intensity. Linear or hyperbolic equations are normally used to represent this behavior (Figure 3). Upon analyzing a suite of standards, a least squares fit is performed to determine the fit. EDXRF has a significant advantage over other analytical techniques for the measurement of Fig. 4 - Attenuation of substrate x-rays. coating weight, because the intensity of the coating x- rays is not dependent on the substrate. ndirect Measurement of the Coat Weight When there are no elements in the coating to excite, EDXRF can measure the coat weight by exciting elements in the substrate. The substrate has the maximum intensity when there is no coating. As the coating increases, the intensity of the substrate x- rays decreases and approaches zero at an infinite coating thickness (Figure 4). Thus the coat weight canbe determined by measuring the attenuation of the substrate x-rays. A bsorption/erihancetnerit Effects f other elements are fluoresced and have sufficiently high energy x-rays, they can excite the lower energy elements in the sample. For example, gold plating solutions contain a potassium salt. The ;r Without Matrix Correction for K a With Matrix Correction for K 1 - /<A t + --A /,,F + + Given Gold (G/L) Figure 5 - Gold Calibration with and without a matrix correction for potassium salt content. 30 4

5 energy of a gold x-ray is approximately 9.7 KeV while the energy needed to excite potassium is only KeV. Therefore, the gold x-ray may be absorbed by the potassium atoms. Since the composition of the plating solution is constantly changing as salts are being added, the amount of gold x-rays absorbed changes with the potassium concentration. Equivalently, the resulting potassium x-ray intensity is enhanced by the gold concentration. This is called an absorption/ enhancement effect. Modem EDXRF analyzers include a software feature that measures the potassium x-rays in addition to the gold x-rays and adjusts the concentration reading accordingly. Correlation data for a calibration with an absorptiodenhancement correction is significantly better than without the correction (Fibwre 5). Another common example of this effect includes sulfur in nickel and copper platings solutions. TYPCAL APPLCATONS OF EDXRF Precision is a measure of how repeatable the readings are. Bias on the other hand is the difference between the actual concentration and the measured concentration. Absolute precision is the variation of a reading in the units of measurement. Absolute precision is typically reported as the standard deviation. For example, if a measurement is 1000 ppm +/- 10 ppm of Fe, then the standard deviation is 10 ppm. A measurement will vary by as much as one standard deviation 65% of the time. Ninety five percent of all measurements on a single sample will fall within two standard deviations of the average. For 99% confidence, three standard deviations should be used. n the previous example, 99% of the measurements would fall withing /- 30 ppm. The relative precision is simply the absolute precision divided by the concentration. Returning to the example, the relative precision would be 10 ppd1000 ppm or 1% relative. Precision and Bias Sources of Error. Comparisons of analytical techniques are Error is the total measurement of how well a often based on accuracy, and accuracy is a method works. t encompasses all forms of error combination of two terms: precision and bias. incorporating the precision and the bias. This includes Sample Type 5 31

6 tmniun b&g on 0 os Gvsr tg/m2l Fig. 6 - Typical correlation plot for a zirconium conversion coating on aluminum Gvsr (dd.4 Fig. 7 - Typical correlation plot for an anodize, coating. error from the operator, the standards, the sampling system, and the machine. f sources of errors are known, one can see the cumulative effects. Although the analyzers and technique are designed so any bias due to instruments is negligible, there is still instrument error in the form of the standard deviation. The standard deviation is largely related to the intensity of the signal and the analysis time. The data presented in the following table refers to the precision related to the instrument, hence the instrument error. Kesiilts A correlation plot of given versus measured is shown in Figure 6 for zirconium conversion coatings on aluminum. These samples were cut to fit the analysis aperture, however, on-line systems would analyze the coil nondestructively. Figure 7 also shows ' excellent correlation for the calibration of an EDXRF analyzer for the analysis os anodized coatings. The results in Table 2 on the previous page were obtained by calibrating an EDXRF analyzer with well assayed standards covering the expected concentration ranges. Statistical analysis were then performed to determine the precision or repeatability. As the table shows, the precision is typically 1% relative. Typical analysis times are on the order of 100 seconds, and sample preparation was negligible. The results indicate that EDXKF can be applied equally well to plating solutions, conversion coatings and anodized coatings. normally only taking 1 to 5 minutes. n addition, the sample is not altered so it can be returned to the product, this is particularly advantages for on-line analysis. On-site analysis allows you to react quickly to upsets in the process. t's also very accurate, normally a precision on the order of 1 to 2 percent relative is obtainable. t's versatility allows it be used equally well in the laboratory or on the factory floor or embedded in the process for a large number of elements ranging from Mg to U. n addition, these analyzers are designed to be very low maintenance, typically requiring nothing more than infrequent changing of paper or window film. REFERENCES 1. Bertin, E. P., Principles and P ractice of X-Ray Spectrometric Analysis, Plenum Press, New York 2. - London, (1970) Little, S.R. and Schindler, J.S., "Applications of Low Cost XRF Technology to On-Line Analysis", 37th Annual Denver X-ray Conference, Gunning, Geoffry, "Energy Dispersive X-ray Spectrometers for the Elemental Analysis of Plating Solutions", AESF SUR/FN Overman, A.R., "Energy Dispersive X-ray Spectrometers for the Elemental Analysis of Plating Baths", AESF SUR/FN 93. SUMMARY EDXRF analyzers arc simple to operate, requiring only a push of a button. Sample preparation is normally only pouring the sample into a sample cup and in the case of powders, manually pressing the sample. There are no hazardous wastes to dispose of or fumes to deal with. Because of this XKF is ideally suited for on-line analysis. t's a very fast technique 32 6