2019 Volume 59 Issue 10 Pages 1838-1845
This paper described how the lateral resolution of an elemental mapping was estimated in laser-induced breakdown optical emission spectrometry (LIBS), when the focus point of a high-frequency Q-switched Nd:YAG laser was moved on a sample surface, along with measuring the emission signal from the resultant plasma. Several measuring parameters were optimized to improve the lateral resolution; namely, they were an averaged laser power of 1 mJ/pulse, a laser repetition frequency of 1 kHz, a scanning rate of the laser beam of 0.5 mm/s, and an atmospheric gas pressure of He 1000 Pa. Using these optimal parameters, a lateral resolution was obtained to be ca. 20 µm in the one-dimensional direction of laser scan. Furthermore, two model samples, in which regularly-aligned copper circles were deposited on a nickel plate, were irradiated by a scanning laser beam to determine actual resolving abilities both in a line direction along travelling the laser and in a two-dimensional direction over a certain sample area. The sample having an interval of 85 µm between the copper circles could give an emission image which was appropriately resolved in the two-dimensional as well as the one-dimensional direction; however, in the other sample having the 25-µm interval, the two-dimensional resolution became degraded compared to the resolution of the line scan, probably because the ablation grooves, which were left on the sample surface, had a width of more than 100 µm and were overlapped with each other in the observed area.
For process control in the steel-making process, various kinds of analytical values need be fed back to the production site precisely and rapidly. Non-metallic inclusion particles embedded in the iron matrix, such as alumina, should especially be noticed in steel analysis, because they generally reduce the fatigue strength of structural steels and sometimes leave any scratches and creases on the surface of the final product.1,2) Therefore, we concern an analytical method to obtain the inhomogeneous and spatial distribution of them in the metallurgical structure of steels, and further to apply it to the on-site/in-line analysis.3) Conventional analytical methods for estimating such a distribution are based on the structural/chemical observation by using an optical microscope, a scanning electron microscope equipped with an X-ray detector, or various X-ray and electron spectroscopies.4,5) A major disadvantage of these methods is that time-consuming and complicated procedures have to be required for the measurement; therefore, it is difficult to apply them to an on-site/in-line analysis in the actual production process.
Laser-induced breakdown spectroscopy (LIBS) has been recognized as a direct analytical method for various solid samples, because the sample preparation is very simple and the measurement time is very short.6,7,8,9) In addition, a remote-sensing measurement is possible due to the probe and detection of radiation in visible/ultraviolet wavelength regions. Therefore, it is employed as an analytical method for industrial usages, such as the quality control of steel materials or separation of scrapped materials.10) In detection of oxide inclusion particles in steel materials, several papers reported on the distribution analysis of them by using LIBS.11,12,13,14,15,16) Boue-Bigne suggested a LIBS method for obtaining the chemical composition as well as the size of inclusion particles in a steel sample, in order to estimate the cleanness of the metallurgical structure.11) Kuss et al. published a paper which discussed a method for determining the position of inclusion particles from variations in the emission signals under a statistical analysis.12) Nakahata et al. reported a quantitative map of alumina inclusions in commercial ferritic stainless steels not only on the surface but in the depth direction by using LIBS in a single-shot mode.16)
Scanning of the laser beam in LIBS can give the distribution of constituent elements in a sample area by measuring their emission intensities. For this purpose, a low-frequency/high-power laser or a high-frequency/low-power laser can be employed, as schematically illustrated in Fig. 1. The former can give the emission intensity enough to be quantitatively estimated each for a laser shot; therefore, the individual crater left by each laser shot would determine the lateral resolution of the elemental map intrinsically. On the other hand, the latter generally leaves successive craters, which are observed as an ablated groove, due to a number of the laser shots; in this case, a better lateral resolution is expected when the emission signal can be obtained from an edge portion of the groove appropriately, but it would depend on several laser parameters complicatedly. The objective of this study is to optimize the measuring conditions for better lateral resolution in the beam-scanning LIBS when using a high-frequency/low-energy Q-switched Nd:YAG laser, and then to apply it to the measurement of a pattern of copper circles deposited on a nickel substrate which is prepared as a model sample, for determination of the actual resolution under the optimized condition.
Schematic explanation for scanning of a laser bean using a high-power/low-energy laser (a) and a low-power/high-frequency laser (b). The former generally leaves individual ablated craters and the latter an ablation groove. (Online version in color.)
Figure 2 illustrates a schematic diagram of the experimental system. The laser used in the experiment was a Q-switched Nd:YAG laser (NL204-1K-SH, EKSPLA, Lithuania). The laser wavelength was selected at 532 nm, which was the second harmonic of the fundamental frequency. The laser beam was focused onto the sample surface using a planoconvex lens with a 150-mm focal length, the irradiation energy of which was measured with a thermal sensor (OPHIR, 15(50)A-PF-DIF-18). The irradiation energy of a laser shot could be prefixed with a guaranteed stability of less than 2.5%. The laser had an ability to be operated at frequencies up to 1 kHz with a pulse width of less than 8 ns. The emission signal from the plasma was detected on a spectrometer system, comprising a grating monochromator (MC-30N, Ritu Ouyo Kougaku Co. Ltd., Japan) and a photomultiplier (R-955, Hamamatsu Photonics Co. Ltd, Japan), through a reflector, a biconvex lens and an optical fiber. The effective spectral band-path of this spectrometer system was ca. 0.1 nm, depending on the wavelength. The emission intensity was detected and averaged with a current/voltage converter (LI-76, NF Corp., Japan), a pre-amplifier and a low-pass filter circuit(Model 3611, NF Corp., Japan), and it was finally recorded through an analog-to-digital converter on a personal computer. A nickel plate on which copper thin film was deposited (see section 2.3) was set to a plasma chamber, so that the slit-pattern or circle-pattern of copper could be placed at the laser irradiation. The chamber was evacuated down to 1 Pa by an oil rotary pump, and then pure Ar (99.9999% purity) or pure He (99.9999% purity) was introduced in it while keeping a predetermined chamber pressure. The chamber itself could be precisely moved to the laser beam in three-dimensional directions, by using an X-Y-Z stage which was controlled by a system controller (GSC-02, SIGMA Koki, Japan). The moving speed of the stage was an important parameter to determine the lateral resolution in the LIBS measurement.
Setup of the experimental apparatuses and a block diagram of electronics for the signal detection (right figure). (Online version in color.)
An optical microscope was employed to observe the shape and the dimension of copper patterns on the sample surface before the laser irradiation. Craters formed by the laser irradiation were observed with a digital microscope (Keyence, VHX-700) to determine the diameter as well as the ablated depth of craters with the progress of laser shots.
In order to optimize the measuring parameters in beam-scanning LIBS analysis, we prepared a specimen of a nickel substrate, on which four rectangular pieces of copper thin film were deposited in parallel by using an electron-beam vaporization method. It is called a slit-pattern sample in this paper. A pure nickel plate (more than 99% purity) was polished with abrasive papers and then finished in ethanol solution with a ultrasonic cleaner. The nickel plate was covered with a grid sheet for TEM measurement (Okenshoji Co., Ltd, Japan), which was normally used to reduce the charge-up effect, and then copper vapor was evaporated on it through electron bombardment against a pure copper target in a vacuum chamber at an argon atmosphere of about 3 Pa. The deposition rate was kept at about 0.1 nm/s, and the thickness of the copper slit-pattern was controlled to be about 1 μm. Figure 3 illustrates a photo of the slit-pattern sample taken with an optical microscope. The width and inter-distance of the rectangular pieces were about 150 μm and 100 μm, respectively.
Photograph of a slit-pattern sample taken with an optical microscope.
In order to determine the lateral resolution for actual samples, we prepared two kinds of specimens, in which circles of copper thin film were deposited successively and regularly on a nickel substrate, by using the electron-beam vaporization method. They are called circle-pattern samples in this paper. For their preparation, we employed 100-mesh and 150-mesh grid sheets for TEM measurement (Okenshoji Co., Ltd, Japan), which the nickel plate was covered with and left circle patterns of copper on it after the same procedure of evaporation as the slit-pattern sample. Figure 4 presents photographs of circle-pattern samples for the 100-mesh (A-specimen) and 150-mesh (B-specimen) grid sheets, taken with an optical microscope. The diameter and inter-distance of the copper circles were about 185 μm and 85 μm for A-specimen, and they were about 135 μm and 25 μm for B-specimen, respectively. The thickness of the copper circles was controlled to be about 1 μm.
Photographs of circle-pattern samples prepared by electro-beam deposition with the 100-mesh (a) and 150-mesh (b) grid sheets.
A resonance atomic line of copper, Cu I 324.75 nm,17) was selected for two-dimensional mapping of copper patterns on the nickel substrate, because it had a strong intensity and was free from spectral overlapping with emission lines of nickel, argon, and helium. Another resonance atomic line, Cu I 327.49 nm,17) was also measured to check the detection of copper pieces on the sample surface.
In the slit-pattern sample, the X-Y-Z stage was adjusted in each direction until the irradiation position of laser had been focused just on the sample surface, and then the laser beam was moved in the lateral direction in order that the laser beam could go across the rectangular pieces of copper perpendicularly at a constant scanning rate, where the frequency of laser and the scanning rate were experimental parameters which should be optimized. In the circle-pattern samples, the X-Y-Z stage was adjusted in each direction to obtain the focal position of laser on the nickel substrate, and then the laser beam was moved along a row of the circles at a constant scanning rate, which was repeated at a certain interval. In this experiment, the distance between the laser traces was also an experimental parameter in addition to the frequency of laser and the scanning rate of laser beam. Furthermore, we examined an effect of the pressures of atmospheric gas (argon or helium) on the lateral resolution. Figure 5 shows a typical three-dimensional image (a) and the cross section (b) after a trace of the laser beam, which was observed with the digital microscope, when the laser was irradiated at a pulse energy of 1 mJ/pulse, a laser frequency of 1 kHz, and a scanning rate of 0.5 mm/s. The laser irradiation did not leave individual ablation craters but did an ablation groove with a width of ca. 150 μm including both the edges and a depth of 4 μm, implying that a number of craters were overlapped with each other due to the high repetition rate. We directly observed the shape of the resulting grooves with the digital microscope, and roughly estimated the ablated amount of samples from their dimensions, as described later.
A typical three-dimensional image (a) and the cross section (b) after a trace of the laser beam, taken with a digital microscope. (Online version in color.)
It is expected to improve the spatial resolution when the resulting size of the groove is reduced as the irradiation energy of laser decreases; however, it also leads to a decrease in the emission intensity with larger variances. As explained in the following section, the irradiation energy of laser was selected to be 1 mJ/pulse in this experiment, so that the emission signal could have enough intensity to be accurately estimated for a certain integration time, prior to the improvement of the resolution.
The ablated amount of a sample is dependent on an irradiation energy of each laser shot, a repetition rate of laser which is determined by the laser frequency, and a scanning speed of the laser beam moved on the sample surface. There might be many possible combinations among these laser parameters; in this measurement, when the laser irradiation energy was fixed at 1 mJ/pulse, the other parameters were varied for their optimization. Grooves left by the laser ablation were observed with a digital microscope, such as the photograph in Fig. 5. Their widths and depths were estimated on the average from five portions for each groove, which was produced at several laser frequencies and scanning rates of the laser beam. Figure 6 shows variations in the groove depth and width with an increase in the laser frequency at a constant moving rate of the laser beam, where the average values and their standard deviations (error bar) were estimated from five randomly-selected portions of two grooves each for the laser conditions. Both the width and depth were gradually extended with increasing the laser frequency, simply because larger numbers of the laser shot were irradiated on the sample surface at higher laser frequencies. It was also found in this plot that the data points included large variances at lower laser frequencies, probably implying that the ablation would be fluctuated due to smaller input power of the laser. On the other hand, the ablation seemed to occur more stable at a laser frequency of 1 kHz, which was a reason for selecting this frequency as the experimental condition in the following experiments. Figure 7 shows variations in the depth and width of the resulting grooves when the moving rate of the laser beam is changed at a laser frequency of 1 kHz. The depth became smaller at higher moving rates whereas the width was almost unchanged. Smaller laser energies were supplied as the moving rate was faster (the ablation occurred on a certain position of the surface for shorter time), thus reducing the depth of the grooves; on the other hand, the width would also be affected by fluctuations of the laser focus. The scanning rate of the laser beam needs to be further optimized because it principally determines the lateral resolution in scanning LIBS, which would be varied also with the laser frequency as well as the pressure of atmospheric gas in the plasma, as discussed in the next section.
Variations in the depth (square) and the width (circle) of resultant grooves as a function of the laser frequency. The moving rate of the laser beam was fixed at 0.5 mm/s, and the irradiation energy was 1 mJ/pulse. (Online version in color.)
Variations in the depth (square) and the width (circle) of resulting grooves as a function of the moving rate of the laser beam. The laser frequency was fixed at 1 kHz, and the irradiation energy was 1 mJ/pulse. (Online version in color.)
In order to obtain better lateral resolution in scanning LIBS, we should consider to optimize not only the laser parameters, such as the laser frequency and the scanning rate, but time-constant parameters for the signal detection. A cut-off frequency of the low-pass filter needs to be determined on an average/smoothing of the emission intensity with the efficient precision. For instance, a cut-off frequency of 1 Hz corresponds to a time constant of 1 s, which is an integration period for signal detection nominally. On the other hand, the cut-off frequency should be determined for a response time of the emission signal, whether the detection can follow its temporal variation; in other words, longer time constants might not be suited for measuring a variation of the emission intensity along with the laser irradiation in the scanning LIBS, whereas they could contribute to the emission measurement with better precision. In this study, the cut-off frequency was fixed at 20 Hz, corresponding to a time constant of 0.05 s, for which an averaging of the emission signal was nominally conducted. For instance, the emission intensities are integrated for 50 laser shots at a laser frequency of 1 kHz, which could be enough to obtain a reliable result for the emission detection.
Figure 8 shows a typical result of the variation of the Cu I intensity when scanning the laser on the slit-pattern sample, comprising four rectangular pieces of copper thin film, at a laser frequency of 1 kHz and the moving rate of 0.5 mm/s, under an atmosphere of argon 200 Pa. The pattern of copper pieces could well be recognized in this emission image. As indicated on the right hand of Fig. 8, a lateral resolution was defined using an interface distance between 20% and 80% as much as the maximum intensity of the copper signal to determine the measuring parameters appropriately.
An example for the variation pattern of the Cu I intensity when the laser beam goes across on the slit-pattern sample, and a lateral resolution, which is defined in this paper, is presented in the right-hand portion. (Online version in color.)
The lateral resolution as well as the Cu I intensity was obtained from six repeated measurements of the emission image, so that the average values and their standard deviations (error bar) could be estimated. At a constant moving rate of the laser beam, a dependence of the lateral resolution and the emission intensity on the laser repetition frequency was measured, as depicted in Fig. 9. Better lateral resolutions as well as larger emission intensities were observed as the frequency was higher. A reason for this was that the sampling of copper at the interface could occur more instantly at higher laser frequencies. The variation in the emission intensity could be attributed to the same reason as in Fig. 6. Therefore, the laser repetition frequency was determined to be 1 kHz for the following experiments. Figure 10 shows variations in the lateral resolution and the emission intensity when the moving rate of laser is varied in the 1-kHz laser. The resolution was gradually improved with a decrease in the moving rate; however, the resulting grooves reached the nickel substrate through the copper piece at moving rates slower than ca 0.5 mm/s, which were not suitable for determination of film-like samples having 1-μm-order thicknesses. Therefore, the moving rate was selected at 0.5 mm/s for the optimal condition.
Variations in the lateral resolution (circle) and the emission intensity (square) as a function of the laser frequency. The moving rate of the laser beam was fixed at 0.5 mm/s, and the irradiation energy was 1 mJ/pulse. (Online version in color.)
Variations in the lateral resolution (circle) and the emission intensity (square) of the Cu I line as a function of the moving rate of laser. The laser frequency was fixed at 1 kHz, and the irradiation energy was 1 mJ/pulse. (Online version in color.)
The lateral resolution as well as the emission intensity was clearly dependent on the atmospheric gas with which the laser-induced plasma was sustained; then, argon or helium was employed over certain ranges of the pressure in the plasma chamber. It has generally been reported that a laser-induced plasma is initiated by breakdown of the surrounding gas just on the irradiated point of laser and then expands towards the surrounding gas along with excitation/ionization of the gas species as well as the ablated atoms,18,19) from which their emission spectra are originated. Various collisions with gas species and electrons cause the excitation of sample atoms; therefore, it is an important issue how the surrounding gas and the pressure in the plasma chamber should be determined.20,21) Figure 11 shows variations in the lateral resolution and the emission intensity of the Cu I line as a function of the gas pressure of argon (a) and helium (b), where the other measuring parameters have been optimized. A maximum of the emission intensity was observed at an argon pressure of 100 Pa (a) or at a helium pressure of 1000 Pa, probably because the stopping power for collisions is different between argon and helium to receive the laser energy more efficiently during the plasma expansion, which is called a pinch effect; namely, the helium plasma is expanded more easily than the argon plasma. Kitaoka et al. explained the difference in the pressure dependence between argon and helium by observing two-dimensional images of the resultant plasma.22,23) Furthermore, it should be noted in Fig. 11 that the lateral resolution was generally better in the helium than in the argon atmosphere. This effect may be attributed to ablation and diffusion behaviors of copper atoms in argon and helium plasmas. Several previous studies pointed out that the ablated amounts of a target material in LIBS were largely varied depending on the kind of surrounding gases and their pressures,24,25,26) and further, Wen et al. indicated that LIBS caused the diffusion of the ablated atoms more rapidly under the plasma conditions of lower gas pressures.26) We also observed in this study that the ablated material was less re-deposited on the surface in the helium atmosphere compared to the argon atmosphere. The difference of the lateral resolution might relate to smaller pinch effect of copper atoms in the helium plasma compared to the argon plasma. As a result, a helium atmosphere of 1000 Pa was selected as an optimal condition, giving a lateral resolution of ca. 20 μm.
Variations in the lateral resolution (circle) and the emission intensity (square) of the Cu I line as a function of the gas pressure of argon (a) and helium (b). The laser frequency and the moving rate were fixed at 1 kHz and 0.5 mm/s, respectively, at an irradiation energy of 1 mJ/pulse. (Online version in color.)
The lateral distribution of copper circles in two kinds of samples (A-specimen and B-specimen) was investigated by using the scanning LIBS. The measuring conditions are summarized as follows: a laser irradiation power of 1 mJ/pulse, a laser frequency of 1 kHz, a moving rate of the laser of 0.5 mm/s, and helium atmosphere of 1000 Pa, and they could give an appropriate lateral resolution of about 20 μm, as discussed in the previous sections (see Fig. 11). This LIBS system was considered to have a resolution power of ca. 20 μm in the one-dimensional direction of laser scan. In addition, a distance between line scans of the laser beam (inter-distance of the laser grooves) needs to be pre-determined for the two-dimensional mapping, which records variations in the Cu I intensity along with lines of the laser irradiation. The line scan, having a length of max. 2400 μm, was repeated with a constant interval ranging from 50 to 150 μm, so that the observed area on the sample surface was ca. 1000 × 2400 μm.
The copper-circle distribution of A-specimen was measured when an inter-distance of the laser scan was fixed at 50 μm. A typical result of the two-dimensional map is illustrated in Fig. 12, where the color tone of red ranks an increment in the Cu I emission intensity from pale red to dark red, and the color of while expresses portions below the background intensity. As illustrated here, the intensity map generally corresponded to the actual positions of copper circles, in which 95% of the copper circles were recognized. On the other hand, the intensity map for B-specimen was obtained in Fig. 13, when the same experimental conditions were employed as those of A-specimen. (Note: there were no copper circles in the vertical portion of more than 20 mm due to an edge of the evaporated pattern.) Compared to the actual size and positions of copper circles, the pattern of the emission intensities appeared to be a little distorted, which included portions losing counts of the copper circle or overlapping with each other, in which 80% of the actual copper circles were recognized. Further, the copper circles were less recognized with increasing interval of the laser scan, giving a poor lateral resolution. These results are clearly because the inter-distance between copper circles is narrower in B-specimen (25 μm) than in A-specimen (85 μm). Since the width of the resultant ablation groove was 100 μm or more (see Fig. 5), it would determine the emission images actually. However, there appeared rows of the emission images corresponding to copper circles along several line scans of laser (one-dimensional), as found in Fig. 13, where each copper circle could be almost resolved, implying that the actual resolution power could exceed the 25-μm interval in these cases. Thus, the overall resolution of the two-dimensional map for B-specimen (see Fig. 13) would be principally restricted by overlapping between the laser scans. It can be mentioned that the distribution of such densely-arranged patterns can be mapped to a certain extent, when LIBS is conducted using a high-frequency laser and a continuous scan of the laser beam.
A typical two-dimensional image of the Cu I emission intensity for A-specimen at an inter-distance of the laser scan of 50 μm, obtained using the optimized experimental parameters.
A typical two-dimensional image of the Cu I emission intensity for B-specimen when the experimental conditions are the same as those for A-specimen.
This paper discussed the extent to which a scanning LIBS could be utilized for elemental mapping of distributed samples when a high-repetition Q-switched Nd:YAG laser was irradiated on them, while the resulting emission signal was simultaneously monitored during scan of the laser beam. For this purpose, two test specimens, in which lots of copper circles were closely and regularly aligned on a nickel substrate, were prepared by electron beam deposition. Several experimental parameters, such as an irradiation power of laser, a laser frequency, a scanning rate of the laser beam, and a pressure of the atmospheric gas, were optimized. Emphasis of this measurement was put on an elemental map in which the portion of copper circles could be recognized rapidly and easily. The distribution of copper circles was obtained from a two-dimensional map of the Cu I emission intensity precisely for a sample (A-specimen), whereas the distribution appeared to be somewhat distorted for another sample (B-specimen). This result was because the copper circles of B-specimen were arranged more densely than those of A-specimen; indeed, the groove width of the laser ablation was broader than the interval of copper circles in B-specimen. However, a better resolution was obtained along the travelling direction of each laser scan. These effects were probably because the ablated grooves having a width of more than 100 μm would be overlapped with each other, depending on the pattern dimensions of the copper circles. Our future work is to apply this LIBS method to local analysis of inclusion particles in an actual steel sample.
The authors sincerely thank S. Fujieda and S. Suzuki, Tohoku University, for the measurement with a digital microscope. This research was carried out under the support of a Grant-in-Aid from the Ministry of Education, Science, Sports and Culture of Japan. (No. 17H01903). A part of the experimental apparatuses was modified under the support of the 27th. ISIJ Research Promotion Grant (2018).