A high quality surface finish grinding process to produce total reflection mirror for x-ray fluorescence analysis

Total reflection x-ray fluorescence analysis is applied to trace element detection in liquid for effective environmental monitoring. This analytical approach requires x-ray total reflection mirrors. In order to achieve high sensitivity element detection, the mirrors require high surface quality for high x-ray reflectivity. Surface finishing for x-ray mirrors is typically conducted through a series of abrasive processes, such as grinding and polishing, and is thus time consuming. The purpose of this study is to streamline and enhance the surface finishing process based on unique high quality grinding techniques for the production of x-ray total reflection mirrors.


Introduction
X-ray mirrors, reflective optical elements that can reflect and focus x-ray (beams) of short wavelengths, are applied to various types of analyzers, including synchrotron radiation facilities [1], portable x-ray elemental analyzers [2], and astronomical x-ray telescopes [3]. In order to effectively characterize the optical and physical properties of x-rays, high surface precision at sub-nanometer (nm) levels of mean roughness [4] in various types of mirror geometry is necessary. Additionally, hard and brittle materials such as monocrystalline Si, quartz, SiC, and low heat-expansion glass, are typically used [5], which makes achieving the required surface precision challenging. Furthermore, finishing processes for x-ray mirrors [6,7] should be considered enough to be efficient.
The combination of ultraprecision grinding with electrolytic dressing [8] and specific polishing, such as elastic emission machining (EEM) [9], has been attempted in creating a mirror. The mirror will be shaped efficiently by grinding and both ground waviness and roughness will be diminished by EEM. Through this process combination, elliptical x-ray mirrors for x-ray free electron laser (X-FEL) facilities that can focus a trajectory of x-ray beam into an ultrafine spot of approximately 75 nm have been successfully produced [10].
Furthermore, portable x-ray elemental analyzers [2,11] have been constructed based on total reflection x-ray fluorescent (TXRF) analysis [12], which are used in multiple fields Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. that call for ultra trace analysis and high quality mirrors. Such mirrors are of plane (flat) geometry and small in size, requiring increased productivity due to their cost restrictions. The mirrors are also required to be accurately flat and smooth, at less than few nm in mean roughness. Materials for the mirrors can be chosen among Si, glassy carbon, and quartz glass. The strict requirements detailed above suggest that an innovative new finishing process that will maintain process number at a reasonable state, as well as attain the prescribed surface roughness and quality within a limited timeframe, is imperative. Therefore, in this study, we have developed a new grinding technique that will provide an efficient finishing process, which will allow for the practical manufacturing of total reflection mirrors dedicated to portable TXRF spectrometers.
2. Portable total reflection x-ray fluorescence spectrometer Figure 1 displays the developed portable TXRF spectrometer [11] and its configuration. When the incident x-rays are totally reflected on the x-ray mirror, atoms in the analyte are excited by both the incident and reflected x-rays, resulting in the emissions of fluorescent x-rays. Then, the emitted fluorescent x-rays will be analyzed via the x-ray detector, producing a spectrum of the analyte compositions. Figure 1 also includes an internal view of the portable TXRF spectrometer. An x-ray tube with a tantalum anode was employed as the x-ray source, and quartz glass was used for the mirror.

Grinding method, experimental set-up and conditions
A precision grinding with electrolytic dressing process [8] was applied to finish mirrors; figure 2 illustrates the experimental grinding set-up, which was developed in our previous research. A rotary surface grinder, with air hydrostatic bearings for the wheel and work piece spindles, was employed for the experiment. Detailed specifications of the set-up and grinding conditions are shown in table 1. The surface roughness (profile) was evaluated by the optical surface measuring instrument at five points on each work piece (mirror), and was then averaged to be compared to different wheels and finishing methods. After grinding, CMP (chemical mechanical polishing) was added to reduce grinding marks.

Grinding wheels
The application of several types of diamond grinding wheels were attempted for finishing the total reflection mirrors. Metal (cast iron) and metal resin compound bond grinding wheels were used for the experiments. Nano-diamond (ND) particles of ∼4.9 nm in diameter were mixed with #20000 diamond abrasives at 2% volume in order to reduce the ground surface roughness of mirrors. A #325 wheel was utilized for preparation of mirror blanks. Figure 3 exhibits the surface quality of blanks produced by the #325 wheel, which appears to have been generated in brittle fracture mode (surface roughess: 278 nm). Its external view suggests that its optical property is poor. Then, the arithmetical mean height (Sa) and maximum height of scale limited surface (Sz) data, which were produced with #4000 metal (cast iron) bond wheels and with #4000, #8000 metal resin bond wheels, were evaluated, as detailed in figure 4. Compared to metal (cast iron) bond wheels, metal resin bond wheels can more effectively reduce roughness. Sa and Sz were 9.94 nm and 346 nm, respectively, after applying a #4000 metal resin bond wheel; however, a #8000 metal resin bond wheel was able to achieve an Sa and Sz of 7.51 nm and 194 nm, respectively. The surface profile examples produced by the #4000 wheels with different bond materials are displayed in figure 5. The metal resin bond was once again able to produce a smoother surface roughness, as confirmed by figure 5(b). The surface profile produced by the #8000 metal resin wheel bond is shown in figure 6, which displays a high quality transparent surface.

Effect of nano-diamond contained grinding wheels
ND is of a single crystal structure, and is easily aggregated into its clusters, as demonstrated in figure 7 [13]. Through stirredmedia milling with microbeads, ND clusters have been dissociated into single particles, which were then dispersed as colloidal liquid. Afterwards, the ND colloidal liquid was used to fabricate ND (1% in volume) contained grindstones. The process is detailed in figure 8 [14]. For comparison, a grindstone without ND was also fabricated. Figure 9 illustrates a tribological testing method used to evaluate the friction coefficient and      its fluctuation on grindstones both with and without ND. Figure 10 [14] displays the difference in friction coefficients. The grindstone with ND exhibited lower friction coefficients with greater stability, compared to that without ND. These results suggest that an ND contained wheel can function with lower grinding force due to its superior tribological properties. Based on this finding, #20000 wheels containing ND were fabricated for the practical grinding trials of mirrors. Grinding experiments for mirrors were then conducted by applying the fabricated #20000 wheels both with and without ND. The results confirmed that the addition of ND decreased the ground surface roughness, as presented in figure 11. The average surface roughness value without ND was 5.84 nm. Those with 2% ND, however, were able to achieve a surface roughness value that was 1/3 of those without ND. Although further research is needed to investigate its mechanisms, for the purposes of this study, it can be concluded that ND contained wheels can be applied to produce higher quality mirrors.

Improvement of surface quality by simultaneous grinding
In order to improve the surface quality of mirrors through the grinding process, simultaneous grinding [15] has been attempted. As displayed in figure 12 (left), mirror work pieces have been simultaneously ground together with a cemented carbide (WC-Co, Co content: 6%) ring. As a mirror work piece was placed in the central area of the jig plate, the WC-Co ring was placed in the peripheral region of the same jig plate. Using this configuration, a mirror work piece was simultaneously ground with the WC-Co ring. Figure 12 (right) illustrates the surface profile produced from simultaneous grinding using a #8000 metal resin bond wheel, demonstrating its effect. The roughness example in Sa without simultaneous grinding was 6.52 nm ( figure 6); however, the one with simultaneous grinding achieved 1.65 nm. The roughness values with simultaneous grinding, even with the same wheels, were lower; average Sa of 1.72 and 1.46 using #20000 wheels with ND 2% and 4%, respectively, without simultaneous grinding, compared to average Sa of 1.38 and 1.12 using the same #20000 wheels with ND 2 and 4%, but with simultaneous grinding. The surface profile examples produced from the #20000 wheels containing ND 0% and 2% are shown in figure 13. Therefore, mirrors containing ND and produced under simultaneous grinding will have weaker grinding marks.

Further improvement of surface quality through additional CMP
After grinding using #8000 was completed, a series of CMP [16,17] processes were added for the purpose of removing     grinding marks. A single-sided polishing method ( figure 14) was employed in this experiment. For the first step, calcined ceria (CeO 2 ) abrasives were applied for 10 min to smooth the ground surface. Next, colloidal silica (SiO 2 ) and colloidal ceria were applied for polishing, for 2 min and 5 min respectively, for further smoothing. Lastly, colloidal silica was applied for 30 min for a final smoothing. Figures 15(a) and (b) display the surface profile and its top view, respectively, on the finished mirror. The mirror produced by the additional CMP processes detailed above was able to achieve an average of 0.85 nm in Sa .
7. Evaluation of produced x-ray total reflection mirrors X-ray reflectivity evaluation [18] (figure 16) was conducted on mirrors produced by #325 (blank), #8000, #20000+ND2% (hereinafter, #20000ND), and the additional CMP (hereinafter, CMP). The results are as displayed in figure 17. The mirrors produced by finer wheels exhibited higher reflectivity. CMP attained the highest reflectivity and #20000ND was a close      second. The blank produced the lowest. After learning the above results, mirrors produced from #8000, #20000ND, and CMP were evaluated on the developed portable TXRF spectrometer for ultra-trace analysis, which compares the detection limits of elements. Figure 18 illustrates the spectra of analytes containing 1 ng each of Mn and Co, obtained with #8000, #20000ND and CMP mirrors. These analytes were prepared by placing and drying 1 μℓ drops of a standard solution that contained 1 mg/ℓ each of Mn and Co on each mirror. As shown in figure 18(a), an Si line was detected because of the quartz glass mirror. An Ar line originating from air containing 0.9% of argon was also detected. Ta L lines originated from the x-ray tube with a tantalum anode. Ni, Cu lines are attributed to mirror contaminations. Figures 18(b) and (c) also display the detection of an Si line, an Ar line, and Ta L lines, all originating from their same respective sources as detailed above.
These analytes spectra indicate that the #8000 mirror demonstrate low detecting sensitivities for both elements. The detection limits of Mn and Co with the #20000ND mirror were 31 pg and 25 pg respectively, and those with the CMP mirror were 35 pg and 30 pg respectively. The difference in the detection limits between #20000ND and CMP mirrors may derive from variations that can arise in the measurement process. The results signify that the quality of the #20000ND mirror is sufficient to be practically applied to the portable spectrometer [11]. Figure 19 compares the finishing process among the three mirrors. Therefore, the #20000ND mirror can be fabricated through an adequate process combination of consistent grinding sequence to provide sufficiently high performance with both abbreviated process length and time.

Conclusions
This study aimed to establish a dedicated finishing process for creating high-quality surfaces for total reflection mirrors that can be applied to the portable TXRF spectrometer. The experiment has successfully demonstrated that the final finishing through simultaneous grinding of #20000ND in a consistent grinding sequence can both effectively and practically generate sufficiently high mirror quality, which enables detection limits of 31 pg and 25 pg for Mn and Co, respectively.