B.2 Recycling Technologies X-Ray Fluorescence Measurements of Molten Aluminum Elemental Composition Leigh C. Duren (MS Candidate Industrial Intern) Advisors: D. Apelian & D. Backman Sponsor: wte Corporation (www.wte.com) Report 07 #2 Sponsorship This project is sponsored by an Advanced Technology Program (ATP) grant to wte Corportation of Bedford, MA. wte Corporation is a 25 year old value added recycling company that operates both plastics and metals recycling and refining operations. wte Corporation has committed significant resources to the development of new, high speed sorting and identification technologies for the metals recycling and production market. Over the past 10 years wte Corporation and their partners have raised nearly $10.0 million in federal funds to develop this technology. ATP is a division of the National Institute of Standards and Technology (NIST), which provides funding for high-risk technologies. The mission of ATP is to accelerate the development of innovative technologies for broad national benefit through partnership with the private sector. The next section explains the vision and national benefits that are the goals of this project. Project Motivation The current method for verifying melt chemistry is quite time consuming and leaves much room for improvement. Consider the steps involved in the process. A small sample of the molten metal is ladled from the molten bath and poured into a steel mold according to ASTM E716-94 Sampling Aluminum and Aluminum Alloys for Spectrochemical Analysis. Depending on the size and homogeneity of the melt this sample could represent only a very small fraction of the overall melt composition, but this is the industry standard. Next, the sample solidifies and is transported to a machine shop where it is machined to a predetermined depth, also according to ASTM E716-94. Finally, the samples are delivered to a chemical testing laboratory where they are subjected to spark optical emission spectroscopy (OES) as per ASTM E1251-04 (Analysis of Aluminum and Aluminum Alloys by Atomic Emission Spectrometry). If the chemistry of the sample falls within the tolerance of the target chemistry the alloy is produced. Alternatively, if the chemistry is out of range alloy additions are made to the liquid melt, which must be given time to homogenize and the process is repeated. 1
Today s trends for just-in-time manufacturing need a faster way of verifying melt composition to save time, energy, and money. As a secondary scrap producer wte Corporation is interested in increasing the amount of secondary scrap utilized in the processing. The more that secondary scrap is utilized in the process, the greater the potential for numerous iterations as described above; an addition of secondary scrap has the potential to correct one compositional variance while simultaneously creating another variance out of specification tolerance. Therefore, there is a reluctance to use the greatest amount of secondary scrap possible. Developing a technology that could provide in situ analysis of melt composition has the potential to increase the use of secondary scrap because it will provide real time feedback on melt composition, as well as possibly increasing accuracy and precision, and will save time, energy, and money. With the low profit margins that many aluminum foundries are facing, cost saving technology is attractive. X-Ray Fluorescence (XRF) and Laser Induced Breakdown Spectroscopy (LIBS) are two technologies that have been identified by wte Corporation as potential candidates for the in situ analysis of molten aluminum. LIBS, a form of OES, has been shown to be successful at measuring the composition of molten aluminum but is still in the prototype stages by Energy Research Company (ERCo) of Staten Island, NY. At the time of this proposal there was no commercial XRF measurement of molten aluminum and there is no science or art reported on the subject. Furthermore, no integrated LIBS and XRF technology has been reported for liquid metals. It is the ultimate goal of this project in its entirety to integrate LIBS and XRF technology to provide accurate and precise in situ analysis of liquid metals, beginning with molten aluminum. Materials Strategies Inc. (founded by Prof. Apelian), wte, and ERCo have established Melt Cognition LLC to commercialize this technology. A Brief Background on XRF When a material is bombarded with x-rays of sufficient energy, the x-rays strike an inner electron, which is ejected from the atom and filled by a higher energy electron. When the higher energy electron fills the vacancy, a characteristic fluorescent x-ray is emitted, termed x-ray fluorescence. For energy dispersive x-ray fluorescence (EDXRF), the type used in this work, the energies of the fluorescent x-rays are detected and processed by a multi-channel analyzer (MCA). The MCA records the number of counts for each incoming energy, which produces a spectrum that can be used for qualitative and quantitative analysis. XRF is widely used for elemental and chemical analysis especially of solid-state metals, glass, ceramics, and even for the detection of air and water pollutants. Objectives This research seeks to contribute to developing XRF technology for in situ compositional analysis of molten aluminum. These contributions are expected to: 2
Establish a method of instrumentation and data analysis for XRF to determine aluminum melt composition; Investigate alloys within the aluminum 380 series and determine the statistical variation of the major alloying elements Si, Fe, Cu, and Zn on XRF measurements taken under ideal conditions; After determining the statistical variation under ideal conditions, vary the experimental factors that degrade the quality of the measurement to determine how each influences accuracy and precision. These factors include temperature of the melt and melt chamber atmosphere; Understand how characteristic emission line intensities are converted to elemental concentration by use of background subtraction, least squares fitting, background correlation functions, separation of peak overlaps etc., in order to determine the most suitable calibration procedure. Methodology The method of instrumentation and data analysis was developed based on first principles and lessons learned from past programs at wte Corporation. The instrumentation faces many technical challenges as this technology uses x-ray equipment in a way that it has never been used before. Induction Atmospheres (IA) of Rochester, NY was contracted to build a custom induction furnace to which the necessary x-ray equipment was fitted. The first, and most obvious challenge, is the high temperatures that both the x-ray tube and detector are exposed to as the melting temperature of Al is 660 o C. In order to keep the x-ray tube and detector at working temperatures the components were placed at a safe distance away from the hot zone of the induction furnace and an additional protective sheath was fitted to the detector. Thermal profiles of the chamber were performed to ensure that temperatures destructive to the components were never reached. The components of the x-ray system were also isolated from the harsh environment of excess vibration and the susception of electromagnetic (EM) fields. The water pump was relocated outside of the induction unit to minimize vibration and the protective sheath for the detector was designed to also isolate it from the EM fields. Isolation from vibration and EM fields was verified by acquiring data with no x-rays present; in the presence of vibration and EM fields noise is detected and recorded by the MCA. X-rays are attenuated very quickly in an atmosphere of air so atmosphere control in the furnace allows the user to acquire data in a vacuum or inert atmosphere of He. An oxygen sensor provides the user with additional information on the level of O 2 in the furnace. Atmospheric control is also used to prevent the formation of any oxides, dross, or slag on the melt surface. Experiments have shown that levels of 0ppm O 2 can be obtained through a vacuum cycle followed by a purge cycle of He. 3
Quantitative X-Ray spectroscopy is very sensitive to the geometry of the x-ray tube, sample of interest, and detector. To accommodate for varying sample heights, the induction furnace was designed to allow the user to adjust the height of the sample while the x-ray tube and detector were placed at fixed, optimum settings. Experiments were performed to find the optimum sample height. Lastly, and perhaps most importantly, redundant safety interlocks are part of the system so that the user is unable to accidentally expose himself to x-ray radiation. The induction chamber must be completely closed so that no x-rays can escape from the bottom of the chamber; the site port into the chamber cannot be open while x-rays are active; the x-ray activation control is removable and stored separately from the experimental unit. After the design and fabrication of the experimental unit, experiments began. The objectives listed above were accomplished through a rigorous set of orthogonal fractional factorial experiments varying the major alloying elements of the aluminum 380 alloy series Si, Fe, Cu, and Zn, while other alloying elements were held constant. The melt chemistry was verified using the current industry standard, spark OES. Using an optimized system, XRF spectra was obtained for alloys within the 380 alloy series space as defined in the design of experiments (DOE) in the form of channels vs counts, as shown in Figure 1. 3500 Raw Data Output for XRF of Liquid 380 Series Alloy 3000 2500 s t n u o C 2000 1500 1000 500 0 100 200 300 400 500 600 700 800 900 1000 Channel Figure 1: Data obtained from an alloy within the 380 Alloy Series Results The major objective of this thesis is to understand how characteristic emission line intensities are converted to elemental concentration by use of background subtraction, least squares fitting, background correlation functions, separation of peak overlaps etc., in order to determine the most suitable calibration procedure. To achieve this objective 4
a MatLab routine was developed by the team to first calibrate the data to energy from channels based on known peaks, assuming the peaks follow a normal distribution- a valid assumption for x-ray spectroscopy. Next, the MatLab routine uses background subtraction methods, least squares curve fitting, and separation of peaks to fit normal peaks to the peaks of interest (Al, Si, Fe, Cu, and Zn). Finally, a relationship between the known chemistry and the raw peak height, the calculated peak height, and the area under the peak is developed in order to find the most suitable calibration procedure for the 380 Alloy Series in the liquid state. The effect of temperature on experimental results will also be investigated by varying melt temperature within the range used in commercial foundries. The completion of this project will provide the necessary data for analyzing the 380 alloy series in the liquid state necessary for developing the first in situ chemical analysis of molten metal. Additionally, it will generate many ideas and paths for further work in the field of molten metal analysis for the scientific community. 5