1 Introduction

Laser-induced breakdown spectroscopy (LIBS) is a laser-based technique that has been used to determine the elemental composition of solids, liquids, aerosol, and gases [1,2,3,4,5,6,7,8,9]. In addition to nonbiological samples, LIBS has been successfully demonstrated for forensic, biological, and environmental applications and, specifically, as a high-throughput technique to quantify natural elements for phenotyping wood species [10] and plant material [11,12,13,14] and to detect and quantify soil carbon [11, 15,16,17,18]. LIBS has been used to detect metal pollutants in soils (soils analysis) [19, 20] and in the elemental distribution in plant leaves (phytoremediation) [12]. Research on LIBS continues to grow, and field-portable instruments capable of in situ and real-time analysis of samples have been developed in recent years with the availability of more reliable, smaller, and less costly laser systems along with sensitive optical detectors, such as the intensified charge-coupled device (ICCD) [21,22,23,24,25,26,27,28,29,30].

A field-deployable analyzer for the determination of metals in soils was developed by Cremers et al. based on the LIBS technique [31, 32]. This instrument consisted of a sampling probe connected to the main analysis unit by electrical and optical cabling. The handheld probe contained a small laser to generate laser-induced plasma on a surface and a fiber-optic cable to collect the plasma emission. LIBS was demonstrated for environmental lead contamination using a portable LIBS backpack unit to detect lead in paint at military installations in California and Colorado in 2001 [21]. An earlier version of a HH LIBS unit developed in Europe in 2008 integrated the pump laser, focusing optics, light collection optics, spectrometer, and power source into a handheld unit [22]. In 2011, the LIBS instrument was proposed to be applied to many different types of material, such as ice, dust, or rocks, specifically for elemental analysis of Europa’s specific surface materials [23]. Cremers et al. [25] demonstrated the development of a field-deployable instrument to monitor radiological, nuclear, and explosive threats. A review detailing the development of portable LIBS systems and their applications gives in-depth descriptions of laser sources (diode-pumped solid-state, microchip, and fiber lasers), compact spectrometers and detectors, and design and configurations to reduce the footprint of the LIBS instruments [26].

The greatest advantage of LIBS is its capability for remote chemical analysis of samples with minimal handling and in some cases with little or no sample preparation, which minimizes waste generation to the microgram per pulse of ablated material. The instrumentation and operation of a LIBS system is simpler than some of the more sensitive techniques, making it more amenable for real-time qualitative analysis of chemical processes. Although calibration standards are required for quantitative analysis, the generation of a single calibration curve will suffice for analysis of samples in a similar matrix. One coauthor has expended tremendous effort in undertaking the quantitative analysis of the elemental content of chemically treated wood and natural wood products [10]. This technique also has been used to determine the presence of minor elements, such as, C, N, Ca, Al, Fe, Ti, Si, Mg, Mn, and Na, in natural wood using the LIBS spectral data sets, which are large and complex representations of the samples.

Recently, a number of instrument companies have developed compact LIBS devices that have a small footprint and can be used in situ to evaluate environment-sensitive materials. In this manuscript, the authors evaluate the applicability of a class-1 laser, within a HH LIBS instrument for the following objectives in air and in a glovebox environment:

  1. 1.

    Generate calibration curves of Cu and Al as major elements present in their respective alloys.

  2. 2.

    Generate calibration curves for minor elements, such as Zn, Ni, Mn, Fe, and Mg, present in these Cu and Al alloys.

  3. 3.

    Surface damage by laser ablation. The ablation craters formed per laser pulse for the different alloys were measured and the diameter and height are reported.

2 Experimental setup, samples, and measurement protocol

The compositions of four Cu and four Al alloys are presented in Tables 1 and 2, respectively. Test samples were bulk solid samples. Measurements were taken using a HH LIBS Z500 from SciAps, Inc. [33, 34] that contains a class-1 solid-state laser. It has an incident wavelength of 1532 nm, a pre-set laser energy of 5 mJ/pulse, and a pulse duration of 5 ns. The laser variability is 6–7% shot to shot laser energy per pulse producing an energy variation between 4.7 mJ to 5.3 mJ/pulse. The HH LIBS system requires the sample to be held flush to the sampling window during data acquisition. In this sampling window, the laser beam is focused on the sample surface, ablation of the surface material occurs, and the plasma plume emitted from the sample surface is collected via collection optics and delivered to spectrometers and a CCD to be recorded as emission lines for elements present in the sample.

Table 1 Composition of four Cu alloys in weight percent (wt%) using reported values from SciAps [33]
Full size table
Table 2 Composition of four Al alloys in wt% using reported values from SciAps [33]
Full size table

The LIBS investigation was carried out on the as-machined surface without additional sample preparations. The measurements were performed on top of each specimen at three different locations, areas 1–3, as marked in Fig. 1a in a 12-point-grid pattern. The 12-point grid employed allows the data to be collected in a specific order, as shown in Fig. 1b, which involves starting at location 1 with two cleaning shots followed by three data shots before the laser moves automatically to location 2. This approach was undertaken for all 12 locations; therefore, 36 spectra were collected for each area and then averaged automatically into one spectrum per area. Therefore, there are three spectra available for each alloy.

Fig. 1
figure 1

Schematic of a sample with the three measurement locations on the surface showing the 12-grid pattern and b close-up of the 12-grid pattern showing the order of each data collection (not to scale). The step size between locations is 400 µm, and each spot is 40 µm in diameter

Full size image

Table 3 shows a summary of the setup parameters for the HH LIBS instrument used in this study. The cleaning process consists of flowing argon gas to the surface of the sample just before applying two cleaning pulses of the laser per location. This procedure is done to ensure that spectra are devoid of external impurities and moisture interference from the surrounding air. The argon-gas preflush is purged through the sample window for 3 s before laser ablation and data acquisition. The spot size of the laser ablation on the surface of the material is expected to be about 40 µm in diameter. The step size between each location in the 12-point grid is set to 400 µm. In the HH LIBS instrument, the unitless numbers of 0 and 19, given by the manufacturer, are the equivalent of a 250 nsFootnote 1 integration period and a 650 ns integration delay, respectively. The integration period is the time from laser initiation to the starting of data collection by the spectrometer (250 ns). The integration delay is the total length of time the data are collected via spectrometer (650 ns).

Table 3 Summary of parameters set in the HH LIBS instrument employed for the data collection
Full size table

A similar approach was performed in an argon-atmosphere glovebox, and the argon preflush from the instrument was kept for consistency, although not necessary.

A color 3D laser confocal microscope VK-9700 (Keyence, USA) was used to evaluate the surface damage created by the HH LIBS laser and the samples’ surface roughness.

3 Results and discussion

Figure 2 shows examples of broadband spectra for (a) one Cu alloy, CDA 836, and (b) one Al alloy, Al 3003, covering a large wavelength range from 180 to 900 nm. Each spectrum shown is an average of three spectra collected on areas 1 through 3, as depicted in Fig. 1. The major peaks for Cu and Al metal considered for calibration of these elements in their respective matrixes are highlighted in Fig. 2c, d. These emission lines are at 396.152 nm for Al and at 521.820 nm for Cu. Although there were two Cu peaks identified, Fig. 2a, for brevity only the analysis on 521.820 nm is included. Since the sampling was done under argon-gas atmosphere, very distinct peaks for argon gas in each of the spectra were detected. These peaks are at 696.543, 706.722, 738.398, 750.387, 751.465, 763.511, 772.376, 794.818, 800.616, and 801.479 nm. The full spectra for Cu and Al alloys show that the HH LIBS instrument can be used to detect not only Cu and Al but also minor elements such as Zn, Ni, Fe, Mg, and Mn, marked in Tables 1 and 2. Overall intensities for the Cu peaks follow a consistent trend with the known elemental concentration for Cu. Although there is a 4.49 wt% Cu composition difference the CDA 715 and CDA 863 alloys show only a slight variation in intensity. In the case of intensity for Al in Al alloys, the most intense peak is for Al 3003 alloy with 98.07 wt% Al while Al 6061 has the highest Al content at 98.91 wt% Al. The intensity for Al 6061 overlaps with Al 7050 with 89.69 wt% Al. The Al peaks are significantly sharper than the Cu peaks; however, there is a noticeable peak shift for the two lowest Al concentrations. Another observation in Fig. 2a, b is at 421 nm there is a significant jump in the spectral background that was observed on all spectra collected and it is believed to be specific to this instrument. Therefore, for each element of interest the spectral data were further analyzed to correct the peak intensity by subtracting the background.

Fig. 2
figure 2

Examples of spectra obtained by averaging three tests (areas 1–3 in Fig. 1) into one spectrum per alloy, a CDA 836 alloy and b Al 3003 alloy. Zoomed in to a narrow wavelength range for all alloys to highlight, c Cu peak at 521.921 nm for the four Cu alloys and d Al peak at 396.152 nm for the four Al alloys

Full size image

The corrected intensities are plotted as a function of the element’s composition from Tables 1 or 2. A linear fit was applied to produce the calibration curve for that element in its matrix. For example, the regression coefficient for the major element Cu in the Cu matrix was determined from the intensity variation of Cu peak at 521.921 nm in all four Cu alloys, Fig. 3a. The value for R2 was constructed to be 0.8541. Similarly, the regression coefficient was calculated for elemental Al in the Al alloys using the emission peak at 396.152 nm, Fig. 3b, and was found to exhibit a non-linear trend. It is known that LIBS technique could have poor repeatability and its performance depends on the experimental conditions and setup, such as laser wavelength, pulse energy, time interval observation and optics alignment for example [35,36,37,38].

Fig. 3
figure 3

Corrected intensities for the matrix elements are plotted against the reported weight percent, a Cu peak at 521.921 nm in Cu alloys using the reported Cu compositions from SciAps and b Al peak at 396.152 nm in Al alloys using the Al compositions from SciAps, SEM–WDS and corrected ZAF SEM–WDS data. The linear fit provides the calibration curves for Cu and Al as the major elements

Full size image

Because of this, the reported compositions of the Al alloys were further analyzed to confirm that the observed LIBS results are indeed due to the detection capability of HH LIBS instrument. Therefore, using wavelength-dispersive X-ray spectroscopy (WDS) in the scanning electron microscope (SEM) the composition of Al alloys was examined. Quantitative elemental analysis was performed using a JEOL JXA-8200X electron microprobe analyzer (EPMA) instrument equipped with five crystal-focusing spectrometers for WDS. Quantitative data acquisition with the JXA-8200X for this study utilized a 15 kV accelerating voltage, an electron beam current of 200 nA and a beam diameter of 25 microns. The WDS technique works by employing traceable elemental standards with known standard intensities that are used to compare to the experimentally acquired X-ray intensities for each element in the Al alloys. Table 4 shows the composition reported as raw WDS (as collected) and corrected ZAF WDS data in comparison to the SciAps composition for elemental Al. The ZAF X-ray intensity correction must be done to account for the effects from atomic number (Z), the absorption (A) and the fluorescence (F) excitation. Furthermore, the corrected LIBS peak intensity for elemental Al is plotted as a function of all three elemental compositions. The R2 was determined to be 0.33488, Fig. 3b from the linear fit of all three compositions concatenated. The error bars for Cu intensity, Fig. 3a, and for the rest of experimental results below represent the standard deviation of data from the three areas measured. In the case of elemental Al, Fig. 3b, the error bars were not included to prevent overloading the graph.

Table 4 Composition of Al in Al alloys in wt% as reported by SciAps [33] and data obtained from SEM–WDS analysis
Full size table

Figure 4 identifies three minor elements present in Cu alloys in low subpercent amounts as Zn with emission peak at 481.051 nm, Ni at 221.658 nm, and Fe at 238.076 nm. The coefficient of regression for Zn and Ni in the Cu alloys was calculated to be 0.997 and for Fe it was 0.928, as shown in Fig. 4a–c, respectively. This result is an important aspect of the calibration study using the HH LIBS technique as minor elements have concentrations in the subpercent levels. The obtained regression coefficients indicate a confident fingerprint of composition, demonstrating that the HH LIBS instrument can be used in determining the semi-quantification of elements dispersed in a complex matrix at low concentrations. The error bar increases with the increase in elemental concentration, suggesting a better consistency in detection (better instrument performance) for elements in subpercent level in the Cu alloys.

Fig. 4
figure 4

Calibration curves for three minor constituents within the Cu alloy samples a Zn using the 481.051 nm Zn peak, b Ni using the 221.658 nm Ni peak, and c Fe using the 238.076 nm Fe peak

Full size image

Similarly, in the case of elemental Al, the calibration curves for three minor constituents within the Al alloys were obtained by plotting the corrected intensity as a function of the elemental composition from SciAps, SEM–WDS and ZAF-corrected SEM–WDS. Shown in Fig. 5 are the calibration curves for (a) Mg (peak at 518.360 nm) with R2 value of 0.780, (b) Mn (peak at 478.342 nm) with 0.920 as the regression coefficient and (c) Zn (peak at 481.051 nm) with R2 of 0.730.

Fig. 5
figure 5

Calibration curves for three minor constituents within the Al alloys which are a Mg using the 518.360 nm Mg peak, b Mn using the 478.342 nm Mn peak, and c Zn using the 481.051 nm Zn peak

Full size image

Again, the Al alloys present a challenge and the composition determined for the minor elements do not follow a very linear trend, with the exception of Mn. It is noteworthy that Zn in Cu alloy was identified with a higher confidence than Zn in the Al alloys. As mentioned before, the experimental setup and instrument parameters have potential to produce significant signal variation. Laser energy variability was reported to have a small error contribution; however, when associated with a sample that is chemically inhomogeneous, it adds additional spectral fluctuations [36]. In this case, the 12-point grid was employed to overcome the possible chemical inhomogeneity. Additionally, the signal-to-noise ratio (S/N) was examined as another potential error source. The detector’s response at a given wavelength in a spectrum is the sum of the analyte signal (S), the signal due to plasma continuum background (B) and the background introduced by the detector, which in general can be considered negligible [36, 38]. S/N increases with laser energy to 100 mJ where it plateaus. Weak spectral lines with a S/N ratio close to 1 led to low reproducibility. Better results were observed for S/N close to 2 and even greater performance in the case of strong lines with at S/N close to 20 [38]. For Cu alloys S/N ratio or S/B was calculated using the peak intensity and the background value in the flattest region near the peak in the same way data for Figs. 3, 4 and 5 were obtained. For the Cu alloys the S/B was calculated to be 5.4 for Cu, 2.5 for Zn, 2.1 for Ni and 1.8 for Fe. For the Al alloys, the S/B average was 56 for Al, 1.1 for Mn, 1.9 for Mg and 1.4 for Zn. For Cu alloys the S/B values are consistent with the literature having close to two or higher values and good results. Al has very strong spectral lines and low background resulting in a large S/B ratio and poor results, while Mn in the Al alloys yields a linear calibration curve despite the S/N being close to 1. Finally, the significant inconsistencies for the Al alloys are due to the laser wavelength of 1532 nm that does not efficiently couple with the material under test. In addition, the amount of energy is inadequate to generate a stable plasma. Thus, the calibration curves do not show the best R2 values for the major elements and this has been carried over to certain minor elements. Three minor elements (Ni, Zn, and Fe) in the Cu matrix show excellent R2 values since the ionization potential for those elements is low under the test conditions. In the case of the minor elements present in the Al alloys, only Mn is easily ionized from the matrix into the plasma but Zn and Mg are not efficiently ionized.

The disagreement observed on the elemental compositions between WDS measurements and those reported from SciAps are due to the as-machined surface finish. The WDS analysis in general requires the sample to be polished to mirror finish; however, the authors wanted to probe the application of this class 1 HH LIBS on the as-received sample surface without additional sample preparation. Figure 6 shows images at magnifications of 50× and 150× of one indentation formed during the HH LIBS measurement, corresponding to one point in Fig. 1, taken on Al 6061 alloy Fig. 6a, b, and CDA 864 alloy Fig. 6c, d. These images were composed by overlapping the microscope laser intensity to the optical image. The roughness was calculated applying the B060:1994 filter with a wavelength cut-off of 0.8 mm. The roughness area (Ra) around the indent is more than two times larger for Al 6061, which explains the inconsistency in the SEM results. The laser ablation produced a crater in the shape of a cone consistent with the literature [39]. The higher magnification images, Fig. 6b, d, allowed a precise measurement of the cone’s diameter at the surface of the sample (base of the cone) which is 55.94 µm for Al alloy and 52.99 µm for the Cu alloy. However, the HH LIBS ablated depth (cone’s height) per laser pulse is calculated to be 9.151 µm for the Al alloy and 5.133 µm for the Cu alloy. As expected, Al alloys with density of 2.5–2.7 g/cm3 and melting temperatures up to 660 °C, which is lower than the Cu alloys (980–1100 °C) and density of 7.8–8.9 g/cm3, underwent larger size craters because it is easier for the laser to ablate the sample. The depth of ablation per pulse depends on the material being measured, laser energy per pulse, laser wavelength, beam diameter before the focusing lens, and diameter of the focal spot on the sample’s surface.

Fig. 6
figure 6

Images showing one indent equivalent to one point in Fig. 1 for a, b Al 6061 alloy; c, d CDA 863 alloy taken at ×50 and ×150, respectively. Ra represents the calculated roughness area

Full size image

The objective of this research was to evaluate a HH LIBS instrument on Cu and Al alloys as a starting point for undertaking future complex material characterization work. These alloys consist of major elements such as Cu in the 60–85 wt% range and Al in the range of 89–100 wt%. Minor elements such as Fe, Ni, and Zn in the Cu alloys were detected, and also Mg, Mn, and Zn were detected in the Al alloys. The evaluation was based on the advantages that the HH instrument is small in size, easy for in-field sampling, and that it can clean and scan surfaces using a few laser shots before spectra are acquired. Yet certain disadvantages were identified during this study related to the incident laser wavelength and laser energy per pulse available for this instrument. Considering the laser wavelength of 1532 nm is an infrared wavelength and laser energy of only 5 mJ/pulse was used, these results are directly correlated to the instruments performance and capabilities [36,37,38]. It is well known that the mechanism of excitation/breakdown for infrared wavelengths of the sample surface is a thermal-breakdown process rather than a multiphoton-ionization process that occurs when ultraviolet-laser wavelengths have been used as the excitation sources [38, 40]. This process for infrared excitation is described and calculated in detail in Ref. [40]. Literature indicates that at 193 and 266 nm the plasma initiation is purely due to the onset of multiphoton-absorption processes, whereas for 1064 and 1532 nm wavelengths the plasma formation mechanism is solely a thermal avalanche process [41, 42]. This effect has been confirmed by computing the threshold intensities using different wavelengths, indicating that plasma initiation at the visible wavelength, 532 nm, is a combination of multiphoton and thermal avalanche breakdown effects, whereas it is a pure multiphoton effect at 266 nm and a thermal-breakdown effect at 1064 nm. In general, when a thermal event is launched, the laser energy needed to initiate the thermal-breakdown effect is much higher than the multiphoton-breakdown mechanism, which explains why some of the regression coefficients obtained for certain minor elements (Mn, Mg and Zn in the Al matrix) are not as good as for others (Zn, Fe, and Ni in the Cu matrix). Specifically, for the HH LIBS instrument, the excitation energy is only 5 mJ/pulse which is not high enough to break bonds for all the minor elements and get a consistent representation of the excitation of the minor elements that are excited into the plasma created by the laser. It is noted that the laser energy per pulse in case of the HH instrument is much lower than that used in table-top experimental setups needed to generate stable plasma, which is in the range of 25–120 mJ/pulse [40, 43].

Finally, other factors affecting the results obtained here could be the Stark effect and loss of the elemental stoichiometry. The Stark effect in plasmas is due to the collisions of the emitting atoms with electrons and ions, resulting in spectral peak broadening and even more evident in a shift of the line wavelength, since peak broadening can also be an effect of self-absorption of plasma [37]. The crater aspect ratio, formed upon laser ablation, was reported to be a key factor in the degree of fractionation [39]. While spectral line broadening and shifting in peak wavelength, along with ablation craters of different sizes for the two groups of alloys were observed here, these phenomena were not further studied.

Similar experiments were carried out in an argon environment glovebox. The obtained results were very similar to those in air. Therefore, it is concluded that there is no measurable effect from the glovebox on the instrument performance or the results obtained.

4 Conclusions

This fundamental research demonstrated the calibration of Cu and Al concentrations in the Cu and Al alloys, respectively, using a HH LIBS instrument. The instrument used was a class-1 (eye safe) laser model Z-500 acquired from SciAps, Inc. In addition to the major matrix elements Cu and Al, the calibrations of minor elements present in the alloys, such as Zn, Ni, Fe, Mg, and Mn, also were determined. The commercially available alloys used in this study were tested in air and in a glovebox to identify the constituents within the materials under different environmental conditions; however, there was no measurable effect from the glovebox on the instrument’s performance or resulting data. The major element calibration curves of Cu exhibited a linear trend; however, the regression coefficient for elemental Al was 0.3348. Similarly, calibration curves for the minor elements Zn, Ni and Fe in the Cu alloys follow a linear R2 of 0.9 while in Al alloys only Mn has the regression coefficient of 0.9 and Mg and Zn have R2 value of 0.7. The results for the Al alloys are due to a number of factors related to the instrumental and experimental parameters along with a rougher sample surface. It is believed that the significant contribution to the observed results came from the combination of the laser energy of only 5 mJ and infrared wavelength of 1532 nm producing the thermal avalanche breakdown effect leading to a larger error in elemental quantification. In conclusion, it has been observed that the effect of laser–material interaction with the surface of the sample is a major contributor to the elemental detection within the sample for this class-1 laser handheld instrument. Nevertheless, the gentle damage to the surface produced by the laser (5 µm for the Cu alloys) makes this instrument and technique preferential in many applications. It is reinforced that while the semi-quantification was successful only for certain elements and alloys, all the elements known to be present were successfully identified.