2020 Volume 60 Issue 3 Pages 610-612
This paper describes a specified measuring system for glow discharge emission spectrograph, which can provide a spatial/radial distribution of analyte atoms on the sample surface, while the conventional system provides only the information on the elemental distribution in depth direction. For this purpose, a spectrometer system consisting of an image spectrograph and an intensified charge coupled device (ICCD) detector was employed. The delay time and gate width of the ICCD detector was principally selected to improve the spatial resolution of the emitting zone. The objective of this paper was to determine an optimized set of the experimental parameter for better spatial resolution. The best spatial resolution was obtained when the gate width was 1 μs and the delay time was 60 μs. Better spatial resolution was obtained at narrower gate width, because the re-emission from the analyte atoms could be observed to a less extent when the observation was conducted more instantly just after start of the pulsed discharge.
Glow discharge optical emission spectrometry (GD-OES),1,2) enabling a spatially-resolved spectral image to be provided, will become a promising analytical method for defect evaluation in the steel-making process. It can be realized with a particular spectrometer system, where the emitted radiation is directly collimated on the grating of the spectrometer and the plasma image after wavelength dispersion is projected on a position-sensitive detector. Gamez et al. first conducted spatially-resolved elemental analysis in GD-OES by using an imaging spectrometer system with a pulsed radio-frequency power supply, indicating that small particles embedded in an organic sample could be separately detected in the elemental mapping.3) His research group also published a review article about the image measurement in GD-OES, discussing on the lateral resolution related to several experimental parameters for operating glow discharge.4)
We have published a paper with respect to spatial variations in the emission intensity over the whole region of the glow discharge plasma.5) A CCD detector was used for the imaging spectrometer system in our previous study.6,7) However, the CCD detector was not able to measure the emission intensity for an extremely short time just after the sputtering of the sample atoms, because it had a poor sensitivity. In order to improve this disadvantage, we replace the detector with an intensified charge coupled device (ICCD) detector, which enables much shorter gating with the enough sensitivity, thus suited for a time-resolved measurement. This paper explains variations in the spatial resolution of an emitting zone when the gate width and delay time of the ICCD detector was principally controlled.
The experimental system was similar to that employed in our previous study,5) except that the detector was replaced with an ICCD device and that the power supply with a pulsed power amplifier. Table 1 lists the apparatuses, together with their operating conditions.
| Excitation source | Grimm-style9) |
| diameter of hollow anade distance between the electrodes | 8.0 mm 0.2–0.4 mm5,6) |
| Spectrograph | Model 12580 (BunkoKeiki Corp., Japan) |
| Slit width | 65 μm (fixed) |
| ICCD detector | DH734-18F-0 (Andor Technology Ltd., UK) |
| exposure time delay time gate width | 10 s (fixed) 0 to 150 μs 1 to 10 μs |
| Pulsed DC amplifier | HEOPT-1B60-L1 (Matsusada Prec. Ltd., Japan) |
| Function generator | AGF-2005 (GW Instek, Taiwan) |
| waveform frequency | square at duty ratio of 50% 250 Hz (fixed) |
| Plasma gas | Argon (>99.999%) |
| gas pressure | 530 Pa (fixed) |
| Analytical line | Cu II 224.7 nm (the most intense line)8) |
The sample was a square-shaped copper chip having a width of 1.0 mm stuck on a pure nickel plate. The copper chip was made of a conductive copper tape with a thickness of 0.35 mm (CUS-13T, Takachi Ltd., Japan). The nickel plate was set to a glow discharge source so that the copper chip could be placed at the center position of the plasma.
Figure 1 shows typical emission images of the Cu II 224.7 nm line for the copper chip, when the gate width of the ICCD detector was 1 μs (a) and 5 μs (b). The peak voltage of the glow discharge plasma was −800 V and the argon pressure was 4 Torr (530 Pa). A pulsed voltage was applied to the glow discharge source at a frequency of 250 Hz and duty ratio of 50%. In this image, 1 pixel corresponded to 43 μm in the actual size. In this study, the lateral resolution was defined from a half width at half maximum (HWHM) of the intensity profile.7) The calculated resolution in x axis was 1.43 mm when the gate width of the ICCD detector was set to be 1 μs, and it was 1.73 mm when the gate width was 5 μs. It was evident that the size of emission image with the ICCD gate width of 5 μs was larger than that of 1 μs. Therefore, it can be inferred that the influence of diffusion of the sample atoms became obviously when the gate width became larger.
Emission images of a copper-chip specimen when the gate width of the ICCD detector was set to be 1 μs (a) and 5 μs (b).
Figure 2 shows a transient variation in the emission intensity as a function of the delay time of the ICCD detector, and the corresponding trigger pulse, which is controlled with a square-formed timing pulse of the function generator at a frequency of 250 Hz and a duty ratio of 50%. The emission intensities and their standard deviations were estimated for triplicate measurements. There appeared an interval of about 100 μs form the start of the timing pulse until the emission signal completely rose. After the emission signal reached the steady/constant stage, sample atoms could diffuse over a certain area of the resulting plasma sufficiently; as a result, the lateral resolution for the imaging would become worse. Therefore, it is considered that the lateral resolution can be improved if the emission light is detected just before the diffusion of the sample atoms. In our previous paper, we have reported a response performance of the power amplifier employed in the study.10) The power supply took ca. 47.5 μs from the start of the timing pulse until the pulsed voltage has been elevated up to 50% as high as the peak voltage,10) and thus about 100 μs was needed for loading of the full voltage. This behavior was similar to the rising-up variation of the emission intensity, as shown in Fig. 2. Therefore, it was considered that the intensity variation would be mainly attributed to the performance of the power supply.
Temporal variation in the emission intensity of Cu II 224.7 nm together with the trigger pulse at a frequency of 250 Hz and a duty ratio of 50%. Error bars were estimated from the standard deviations when the intensities were averaged for triplicate measurements. (Online version in color.)
It is important in a time-resolved measurement of the emission signal to synchronize the waveform of pulsed voltages with the gate width and delay time of the ICCD detector. On/off periods of the plasma could be changed when adjusting the duty ratio of the voltage waveform. Furthermore, the gate width of the ICCD detector can be adjusted to be narrower to detect only the emission signal just after analyte atoms are sputtered and not to detect the re-emission caused by their diffusion in the plasma. However, the intensity became much more reduced at shorter gate widths; thus a compromise condition had to be determined.
3.2. Delay Time of ICCD DetectorFigure 3 shows HWHM values of the emission image, along with the emission intensity, when the delay time of the ICCD detector varied from 50 μs to 150 μs. In this measurement, the frequency was set to be 250 Hz and the duty ratio was set to be 50%, thus making the on/off periods of the plasma to be kept constant. The HWHM values were reduced when the delay time became smaller. The emission intensity increased with an increase in the delay time and then it became saturated after 100 μs because the discharge power would be fully loaded as depicted in Fig. 2. It was thus found that the lateral resolution could be improved as the delay time was reduced down to 50 μs; however, further smaller delay times would not work sufficiently due to the rising-up characteristics of the emission signal (see Fig. 2). When sputtered atoms diffuse in the plasma with re-emitting by the repeated excitations of them, the lateral resolution would deteriorate due to the emitting portions different from the original one; therefore, it is important to select the delay time within each transient signal.
Variations in HWHM values of the emission image and in the emission intensity of Cu II 224.7 nm when the delay time of the ICCD detector varied from 50 μs to 150 μs. The gate width was fixed to be 1 μs and the peak voltage of discharge to be −800 V. (Online version in color.)
Figure 4 shows HWHM values of the emission image, along with the emission intensity, when the gate width of the ICCD detector varied from 1 μs to 10 μs, indicating that the lateral resolution was improved with decreasing the gate width. The best lateral resolution obtained was 1.43 mm when the gate width was 1 μs and the delay time was 60 μs. However, when the gate width was set to less than 1 μs, the lateral resolution could not be calculated exactly, because the emission intensity became very weak. This result implies that, when the delay time has been optimized, the gate width should be selected as small as possible, because the emission of sputtered atoms can be observed more instantly such that they would be located just above their original portions of the sample surface. As a result, the influence of the re-emitting would be reduced at smaller gate widths.
Variations in HWHM values of the emission image and in the emission intensity of Cu II 224.7 nm when the gate width of the ICCD detector varied from 1 μs to 10 μs. The delay time was fixed to be 60 μs and the peak voltage of discharge to be −800 V. (Online version in color.)
By using the optimized measuring conditions for the gate width as well as the delay time, the lateral resolution was obtained to be 1.43 mm. This value was larger than the actual sample size, meaning that the diffusion of sample atoms could not be controlled completely under the measuring conditions, and thus the emitting zone was over the size of the specimen. Furthermore, this measurement required a long exposure time of the ICCD detector to obtain an emission image after wavelength dispersion. The emission intensity was accumulated by repeating nominally 2500-times discharge pulses during the exposure time of 10 s at the discharge frequency of 250 Hz. As shown in Fig. 2, some variance in the emission intensity was included during the averaging of the signal. Therefore, transient responses of the emission signal would not be exactly the same for each discharge pulse.
3.4. Two-dimensional Spatial ResolutionIn order to evaluate a two-dimensional spatial resolution of the spectrometer system, we prepared a sample that a square-shaped copper chip having a dimension of 1.0 × 1.0 mm, which was made from the same copper tape described in Section 2, was stuck on a pure nickel plate. The thickness of the copper tape was 0.35 mm.
Figure 5 shows a variation of the emission intensity of the Cu II 224.7 nm line in depth direction, together with several emission images. The gate width of the ICCD detector was set to be 1 μm, and the delay time was set to be 60 μs. The emission intensity is obtained by averaging over a central area of 32 × 32 pixels in the emission image. The emission intensities kept unchanged during sputtering for the first 30 min and then almost disappeared after about 40 min. During this depth profiling, there appeared little change in the HWHM values along the horizontal axis (not shown as a figure) which were similar to the result indicated in Section 3.3. However, the lateral resolution in vertical axis was much larger than the real size (not shown as a figure). The reason for this was that the image of the spectrometer system, employed in the study, was enlarged along the slit direction, which is in the direction of vertical axis. Therefore, this distortion should be corrected to obtain the emission image more accurately.
Variation of the emission intensity of Cu II 224.7 nm line in the depth direction, together with several emission images. In this measurement, the square waveform of the applied voltage had a peak voltage of −800 V at a duty ratio of 50%.
Accordingly, it is possible to carry out a 3D measurement through sputtering of a thin and chip-like specimen, indicating that an in-depth profile, the sputtering rate of which was ca. 12 μm/min, could be conducted with kept the lateral resolution to be ca 1.5 mm.
In spatially-resolved GD-OES, the emission images of a copper chip which was prepared as a test specimen, were observed to estimate the lateral resolution when the plasma was generated by loading a pulsed direct-current voltage. Under the optimized conditions in this experiment, the best lateral resolution was obtained to be 1.43 mm when the gate width was 1 μs and the delay time was 60 μs. However, it was larger than the actual sample size, because the emitting zone would extend slightly in the plasma over the size of the copper chip. By reducing the gate width, it was possible to improve the lateral resolution. However, the detected emission intensity became very weak when the gate width was further reduced, and it was thus difficult to take an emission image for calculating the lateral resolution. In order to obtain an emission image to resemble the specimen more closely, we will have an ICCD detector with higher sensitivity, which is one of our future works. The gate width should be reduced as small as possible to detect the emission signal of analyte atoms from the plasma just above their original positions, immediately after emitting.
