1 Introduction

Humidity and moisture are the basic and enduring problems which can be seen in most buildings. In some cases, ignoring this issue causes irreparable damages to structures and buildings, and this is whereas the solution is insulation. For this purpose, one special type of materials produced by various factories is the bituminous waterproofing membrane, which is known as isogam very often. As we know it, the use of this product is very common and yet essential in the construction of the dams, pools, passovers, tunnels, roofings, etc.

Isogams are generally made up from the tissue layer, polyester, sap (special additives of the product) and polyethylene. Although the quality of all these constituents is very important in producing this type of product, the quality of sap plays the most important role. According to the type of applied sap, isogams can be categorized in two groups of blown and polymer. Sap in blown isogam consists of tar (which is extracted from oil and includes the components such as annular and aromatic molecules, CnH2n, CnH2n+2, and talc powder, Mg3Si4O10(OH)2), while sap in polymer isogam consists of tar, talc powder and polyethylene layers [1]. Talc powder helps better absorption of humidity and plays a complementary role. Sometimes, for different reasons mainly due to high prices of raw materials, some factories, instead of talc powder, add impurities such as limestone powder whose main component is calcium carbonate (CaCO3). This increases the volume of the material and causes a decrease in the tar quality. Hence, the extent of limestone powder is much determinant in the quality of isogams [2, 3]. On the other hand, the standard methods for specifying the amount of this substance are mostly intricate and time consuming, and they normally require a chemistry lab.

One rapid and appropriate way to determine the quality of isogams is the use of laser-induced breakdown spectroscopy (LIBS). LIBS is an in situ and real-time analytical technique based on the emission of a plasma arising from the laser-matter interaction. In this method, all the elements of the periodic table can be detected with a detection limit close to the ppm (parts per million), regardless of the nature of sample: solid, liquid or gas. LIBS can conduct elemental and even molecular analysis, which the latter ability makes it a trustworthy technique for the identification of organic materials [4,5,6,7,8].

In recent years, the excellent capability of the concurrent use of molecular bands and atomic lines to analyze materials through LIBS has attracted the attention of many researchers. Particularly, CN violet bands and C2 Swan system are two famous bands which have been used along with atomic lines of carbon, hydrogen, oxygen, and nitrogen to identify and classify organic materials especially polymers [9,10,11,12,13,14,15,16,17].

The goal of this research is to specify the quality of isogams by the use of plasma emission spectra obtained from LIBS process as a simple, fast, and low-cost method. The concentration of calcium existed in isogams due to the presence of limestone relative to that of carbon and hydrogen is taken into account as a criterion of the quality. For this purpose, the intensity ratios of \({H}_{\alpha }\) line of hydrogen and molecular emission bands of cyanide (CN) and diatomic carbon (C2) to the emission line of calcium (Ca) is scrutinized for several different brands of isogams and considered as an index for the quality of them. The correctness of the results is also confirmed through FTIR spectroscopy and EDX analysis. Furthermore, to prove the ability of LIBS in discriminating different samples, LDA as a statistical technique is used for classifying a set of observations into mutually exclusive groups on the basis of a set of independent variables [18,19,20,21,22].

2 Experimental

The schematic diagram of the setup used in our experiment for measuring the plasma emission spectrum is shown in Fig. 1. The Q-switched, Nd:YAG pulsed laser (with the wavelength of 1064 nm, the maximum pulse energy of 360 mJ and the repetition rate up to 10 Hz from Quantel Co.) was adjusted to generate a laser pulse with the duration of 5 ns, the energy of 126 ± 5 mJ and the repetition rate of 1 Hz. A convex lens is used to focus the exciting laser light on the surface of samples. All measurements were performed at ambient pressure. To adjust and change the position of samples, they were connected to an adjustable XYZ manual micro-positioner. For two reasons, the position of laser light on the sample must be successively changed after about every seven laser shots. First, to avoid the formation of deep craters, which in turn establishes two problems: the plasma shielding effect and the non-stoichiometric ablation [23, 24]. The former causes an amount of exciting laser energy to be absorbed by the plasma created through the previous pulses, and the latter can invalidate conventional equations governing LIBS process. And the second is to take into account the effect of probable inhomogeneity existing in each isogam. In practice, to achieve the main layers of isogams, first, one piece of each isogam is cut from its thickness cross section. Prior to each measurement, to eliminate possible superficial contamination which can be established while cutting, in addition to cleaning the samples, each point of isogams’ layer is irradiated by three laser shots. Then, five cleaned points of the surface of each isogam are irradiated by four laser shots, and all \(5\times 4=20\) plasma spectra, recorded by the spectrometer, are exploited to analyze the isogam under experiment. The plasma emission spectrum is collected by an optical fiber (UV 600/660 type with SMA-905 connector and 1 m length) using a quartz collimating lens. The fiber output is coupled to the entrance slit of a compact wide range spectrometer (SpectraStar S150 from Solar Laser Systems Co.) with the spectral range 182–826 nm and the wavelength resolution 0.44 nm. A charge-coupled device (CCD) detector array (Toshiba TCD 1304AP with 3648 pixels) is used to detect the dispersed light. The CCD camera is triggered ~ 1 μs after the onset of laser shot using a suitable delay generator to reduce the continuum originated from bremsstrahlung radiation and ion–electron recombination.

Fig. 1
figure 1

Schematic setup used in the experiment of LIBS

3 Results and discussion

Five different types of isogam were used to investigate in this experiment. The samples were provided from different manufacturers which are henceforth named A, B, C, D, E. The mean and standard deviation of 20 successive spectra of each sample were statistically taken into account to evaluate the relative concentrations and discriminate samples.

As we know, 90% of the tar consists of hydrogen and carbon. On the other hand, due to the existence of talc and some dopants like CaCO3 in the isogam, it is expected that in addition to emission lines of some elements such as C, H, Mg, and Ca, molecular emission bands of CN and C2 can be also observed in the plasma plume. Anyway, in the case of carbon, since the excitation energy of atomic carbon (7.685 eV) is greater than that of excited CN (3.2 eV) and C2 (2.4 eV), except in highly sensitive spectrometers having an intensified charge-coupled device (ICCD), the signals of different species of carbon are hardly observed [25].

Figure 2 shows several atomic lines and molecular bands emitted by the laser-generated plasma related to isogam B (as the representative of all samples). The dominant characteristic peaks with appropriate signal-to-background ratios are related to the molecular vibrational bands of C2 Swan system at 516.52 nm and highly sensitive CN violet bands at 386.19, 387.14, and 388.34 nm as well as the strongest spectral lines of Ca (315.89, 317.93, 370.60, 373.69, 393.37, 396.85, 422.67, 445.48 nm), Mg (279.80, 280.27, 517.27 nm), and \({H}_{\alpha }\) (656.28 nm).

Fig. 2
figure 2

Plasma emission spectrum of isogam B (as the representative of all samples)

Since excitation energies of different species of carbon, as mentioned earlier, are significantly large, > 7.685 eV, regarding the usual exciting laser energy used in the experiment, the intensity of spectral lines of C, even its well-known line at the wavelength of 247 nm is very low compared with the line intensities of the other elements. Therefore, the observation and measurement of these spectral lines on the spectrum recorded by spectrometers with a moderate sensitivity such as S 150, which are affordable for industrial applications, are not easily possible.

In our experiment, the samples were irradiated by the laser beam in the presence of ambient air. Therefore, the formation of C–N compound in the laser-generated plasma plume can be primarily due to recombining excited carbon atoms belonging to isogam with excited nitrogen atoms of the ambient air. To achieve more accurate results, it would be better that the experiment is implemented in an inert atmosphere, for example, helium or argon [7, 12]. However, it should be noted that our goal is to evaluate and compare different isogams under normal industrial laboratory conditions where a helium or argon atmosphere is not readily accessible [26, 27].

To investigate the quality of isogams, among all atomic lines and molecular bands observed in the plasma spectra, \({H}_{\alpha }\) line of hydrogen (656.28 nm), the line of once-ionized calcium (317.93 nm), the (0, 0) band of CN (388.34 nm), and the (0, 0) band of C2 (516.52 nm) were considered. The reason to opt the above atomic lines is that they have fairly low self-absorption ability, while in the case of the bands the strongest ones are important to select. Here, it should be emphasized that the intensity of the Swan bands is proportional to the concentration of the carbon dimer in the excited state, which directly depends on the concentration of carbon in the sample; while the CN bands’ emission can be mainly due to the combination of the ambient air nitrogen and carbon of the sample.

As we know, there is normally a significant fluctuation in the intensity of each spectral line which is due to statistical variations of the ablated mass and plasma characteristics including the plasma volume, temperature, and electron density [28]. This disadvantage can be corrected to a considerable extent using the ratio of lines' intensity rather than the intensity of single lines [29, 30]. Hence, in this research work, the ratios of \({\stackrel{-}{I}}_{\mathrm{C}\mathrm{N}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), \({\stackrel{-}{I}}_{{\mathrm{C}}_{2}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), and \({\stackrel{-}{I}}_{\mathrm{H}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\) were calculated and used to evaluate the quality of isogams of A, B, C, D, and E (Table 1), where \({\stackrel{-}{I}}_{\mathrm{H}}\), \({\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), \({\stackrel{-}{I}}_{{\mathrm{C}}_{2}}\), and \({\stackrel{-}{I}}_{\mathrm{C}\mathrm{N}}\) are the mean intensities of the above specified lines and bands, which have been calculated from 20 measured spectra for each sample. As seen from Table 1, although the amounts of each of the three ratios can be singly exploited to assess the quality of isogams, the sum of them, \({\stackrel{-}{I}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\equiv \left({\stackrel{-}{I}}_{\mathrm{H}}+{\stackrel{-}{I}}_{{\mathrm{C}}_{2}}+{\stackrel{-}{I}}_{\mathrm{C}\mathrm{N}}\right)/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), owing to having better resolution, is an more suitable marker for this purpose. Furthermore, to indicate the accuracy limit of measurements, the standard error of the mean (SEM) has been computed for each ratio in the form of both absolute and percent errors and inserted in Table 1 beside each value. Since the sap in isogams is usually formed from a combination of tar (involving C and H) and CaCO3, in an ambient with low humidity, an increase in the ratio of \({\stackrel{-}{I}}_{\mathrm{H}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\) can be indicative of a high tar/CaCO3 ratio in the isogam. On the other hand, although carbon exists in both tar and CaCO3, an increase in the percent of tar of the isogam could lead to an enhancement of the ratios of \({\stackrel{-}{I}}_{\mathrm{C}\mathrm{N}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\) and \({\stackrel{-}{I}}_{{\mathrm{C}}_{2}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\). Accordingly, one can assert that the stronger the signals of H, CN, and C2, and of course, the weaker the calcium signal, the better the quality of isogam will be. Figure 3 shows the bar charts of the ratios of \({\stackrel{-}{I}}_{\mathrm{H}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), \({\stackrel{-}{I}}_{\mathrm{C}\mathrm{N}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), \({\stackrel{-}{I}}_{{\mathrm{C}}_{2}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), and \({\stackrel{-}{I}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\) for the five samples. As seen, all ratios, as expected, behave in a similar way in all samples. They concurrently experience a decrease process in order of the samples as B, E, D, C, and A, so that the maximum and minimum ratios belong to samples B and A, respectively. In other words, the more the values of these ratios in an isogam, the better the quality is expected to be for that. Accordingly, to realize the quality level of an unknown isogam (allocation process), it is sufficient to measure the marker \({\stackrel{-}{I}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\) and compare it with the one represented in Fig. 3 for each sample. This means that depending on whether the unknown sample marker is closer to the marker of which sample, one can assert that the unknown sample is qualitatively closer to that sample.

Table 1 The ratios of \({\stackrel{-}{I}}_{\mathrm{C}\mathrm{N}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), \({\stackrel{-}{I}}_{{\mathrm{C}}_{2}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), \({\stackrel{-}{I}}_{\mathrm{H}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), and \({\stackrel{-}{I}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\) for the five samples along with their standard errors of the mean in the form of both absolute and percent errors
Fig. 3
figure 3

Bar charts of the relative intensities \({\stackrel{-}{I}}_{\mathrm{C}\mathrm{N}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), \({\stackrel{-}{I}}_{{\mathrm{C}}_{2}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), \({\stackrel{-}{I}}_{\mathrm{H}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), and \({\stackrel{-}{I}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\equiv \left({\stackrel{-}{I}}_{\mathrm{H}}+{\stackrel{-}{I}}_{{\mathrm{C}}_{2}}+{\stackrel{-}{I}}_{\mathrm{C}\mathrm{N}}\right)/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\) for the five isogams

Up to now, by the use of the three relative intensities, the quality of five isogams relative to each other has been determined through a comparative standardization method in LIBS. Although this approach to a great extent is reliable, it can be also confirmed by some other standard analytical techniques such as FTIR spectroscopy and EDX analysis.

Since the largest spectral range of FTIR spectrometers, when used a broadband KBr beam splitter, is normally between 400 and 10,000 cm−1, these spectrometers are generally utilized to detect molecular vibrational spectra. Figure 4 simultaneously shows the FTIR spectra of the five samples. The measurements were carried out by a FTIR spectrometer (Tensor 27 FT-IR, from Bruker Co.) with the spectral range 400–4000 cm−1 and the resolution better than 1 cm−1. As seen, two stronger transitions of 2849 cm−1 and 2917 cm−1 are clearly related to the symmetric and asymmetric stretching vibrations of C–H, respectively [31]. To more accurate study, once again these two vibrational bands have been depicted with more magnification in Fig. 5. As identified in Fig. 5, the intensity of each of the symmetric and asymmetric stretching bands changes from one sample to the other. For both spectral bands, as shown in Table 2, the reduction of intensity is in order of the samples as B, E, D, C, and A, so that the maximum and minimum intensities belong to samples B and A, respectively. On the other hand, the intensity reduction of C–H stretching bands is indicative of a decrease in carbon and hydrogen concentrations of the sample, which in turn demonstrates an isogam with a low quality. Although the upper part of the FTIR spectrum gives us valuable information about the samples, the region between 700 and 1500 cm−1 is not straightforwardly usable for the following reasons. In this region, there are three marked absorption bands at the peaks of 712, 876, and 1430 cm−1 related to pure calcium [32]; two absorption bands at the peaks of 1374 and 1459 cm−1 which, respectively, arise from CH3 symmetric bending vibration and CH2 asymmetric bending vibration of methyl or methylene groups existing in the tar [33]; one band at 872 cm−1 which is related to the out-of-plain bending vibration of CH in phenyl [34]; and some other weak bands which are not so effective in the analysis. The absorption bands at 712 and 876 are significantly narrow, while the one at 1430 cm−1 is very broad. Due to too much width, the band related to the peak of 1430 cm−1 of calcium totally or partially overlaps with the bands corresponding to the peaks of 1374 and 1459 cm−1. In addition, the band at 872 cm−1 can overlap with the band centered at 876 cm−1 of calcium. Therefore, it is obvious that these spectral interferences make it incapable and unreliable to use these signals for the analysis. Concerning the band centered at 712 cm−1 of calcium, since the calcium absorption in this region is so trivial, that also cannot be appropriate to analyze. Therefore, as seen from Fig. 5, one can readily observe that the results obtained from FTIR analysis, related to two stronger transitions of 2849 cm−1 and 2917 cm−1, are in full agreement with those derived from LIBS (Fig. 3). As demonstrated, FTIR spectra are only able to qualitatively confirm the LIBS results. To quantitatively confirm, EDX analysis as a more accurate technique could be very useful.

Fig. 4
figure 4

FTIR spectra of the five different brands of isogam which are concurrently recorded by the spectrometer Tensor 27 FT-IR, from Bruker Co

Fig. 5
figure 5

Magnified FTIR spectra of the five isogams in the wavelength range of two stronger transitions of 2849 cm−1 and 2917 cm−1 related to the symmetric and asymmetric stretching vibrations of C–H

Table 2 Peak transmittance of symmetric (2849 cm−1) and asymmetric (2917 cm−1) stretching vibrational bands of C–H for the five samples

EDX spectroscopy is a micro-analytical technique conventionally used in scanning electron microscopy (SEM) for the local determination of chemical elements in solid samples [35, 36]. In this technique, all elements from atomic number 4 (Be) to 92 (U) can be detected in principle, though the detection of light elements (Z < 10) is not facile very often. Figure 6 indicates the EDX spectrum of sample B (as the representative of all samples), measured by a versatile scanning electron microscope fully integrated with a selected EDX microanalyser (VEGA3 SEM, from TESCAN Co.). The mass concentrations of carbon (\({\mathrm{C}}_{\mathrm{C}}\)) and calcium (\({\mathrm{C}}_{\mathrm{C}\mathrm{a}}\)) and also their ratio (\({\mathrm{C}}_{\mathrm{C}}/{\mathrm{C}}_{\mathrm{C}\mathrm{a}}\)) were calculated by the use of the results obtained from the EDX spectra for all the five samples (Table 3). As seen, the ratio \({\mathrm{C}}_{\mathrm{C}}/{\mathrm{C}}_{\mathrm{C}\mathrm{a}}\) reduces in order of the samples as B, E, D, C, and A, which, like FTIR outcomes, completely confirms the correctness of the results yielded from LIBS process (Table 1 and Fig. 3).

Fig. 6
figure 6

EDX spectrum of isogam B (as the representative of all samples)

Table 3 The mass concentrations of carbon and calcium and the ratio of them for the five different isogams obtained from EDX analysis

Finally, to comprehend the ability of LIBS method in determining the degree of quality difference of isogams, LDA as a computerized chemometric technique was applied to the predictors (i.e., statistical independent variables) obtained from the emission plasma spectra of the samples. The intensities of 12 atomic lines of Ca (315.89, 317.93, 370.60, 373.69, 396.85, 422.67, 430.25, 442.54, 445.59, and 612.22 nm), O (777.19 nm), and H (656.28 nm) and four molecular bands of CN (386.19, 387.14, and 388.34 nm) and C2 (516.52 nm) were considered as the predictors. To minimize the effect of uncontrollable systematic fluctuations related to the stimulating laser light intensity, the amount of ablated material and the direction of the axis of plasma plume relative to the axis of the light collecting optic, the intensity of each spectral line was normalized to the sum of the intensities, whose impact is to approach a normal (Gaussian) distribution for predictors. In this case, considering the 16 predictors and the five samples, four linear discriminant functions (LDFs) corresponding to four eigenvalues (\(\lambda\)) with different percent of variances are derived (Table 4); the discriminant space is four dimensional (it should be noted that the number of functions possible is generally either \(g-1\) where \(g=\) number of groups (a group is defined by the data set measured for a sample), or \(p\) (the number of predictors), whichever is smaller). Each function maximizes the differences between groups on that function, and meanwhile none of them is correlated with the other. All calculations were carried out by the commercial program SPSS 24. Figure 7 illustrates a scatter plot of the individual observations in terms of the first two more effective canonical discriminant functions with the variance cumulative percentage of 81.4% along with their group means (centroids) and class-specific 95% confidence ellipses. As can be seen from Table 4, percent of variances of the first two LDAs are 55.7% and 25.8%, respectively, which implies that the first canonical variable (discriminant function 1) has more the discriminatory power than the second one; a thing that can be observed from Fig. 7. Furthermore, as the confidence ellipses indicate, the groups related to the five isogams have fairly well been separated and classified; the between-group scatter is noteworthy, while the within-group scatter is comparatively insignificant.

Table 4 Eigenvalues (\(\lambda\)), percent of variances, cumulative percentages, and canonical correlations related to the four LDFs
Fig. 7
figure 7

A scatter plot of the first two canonical discriminant functions along with the mean groups and their class-specific 95% confidence ellipses related to the individual observations of the five isogams

As a final point, it should be emphasized that although some spectral lines of Ca are affected by the self-absorption effect, for several reasons, one can assert that this phenomenon is not seriously effective in our results. First, the concentration of Ca compared to the other elements of isogams, as seen from the EDX results, is much low; lower than 0.8 wt%. As we know, the less the concentration, the lower the self-absorption will be. The second is that, here, the relative concentrations and not absolute ones are taken into account using relative intensities, implying that the self-absorption effects of the lines placed at a ratio can partly neutralize each other. And the final reason is that the concentrations of calcium in all samples are much near to each other, and in other words they are of the same order of magnitude. This means that the self-absorption coefficients related to each line of calcium can approximately be considered equal to a constant for all samples. The correctness of these argumentations can also be confirmed by Fig. 7 because it well indicates the ratio of the between-group scatter to the within-group one is quite large; an outcome that will never be achieved if there are serious self-absorption effects.

4 Conclusion

In this research work, a rapid way to determine the quality of isogams was presented through LIBS process and using the comparative standardization method. For this purpose, the five different pre-known isogam brands were set under LIBS experiment. After analyzing plasma emission spectra obtained from the five isogams, a number of important elements being in the samples, including C, H, Mg (related to the sap part), and Ca (associated with the impurity part, CaCO3) were identified. Since the element Ca is undesired in isogams, the intensity ratios of \({\stackrel{-}{I}}_{\mathrm{H}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), \({\stackrel{-}{I}}_{\mathrm{C}\mathrm{N}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), \({\stackrel{-}{I}}_{{\mathrm{C}}_{2}}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\), and \({\stackrel{-}{I}}_{tot}/{\stackrel{-}{I}}_{\mathrm{C}\mathrm{a}}\) for the five samples were considered as principal criteria to determine the degree of the isogams’ quality. Although the results derived from this approach, as expected, were in full correspondence with our primary awareness of the brands, to further confirm the results, FTIR spectroscopy and EDX analysis were also performed for the five samples. In FTIR spectroscopy, the results obtained from two stronger transitions of 2849 cm−1 and 2917 cm−1 related to the symmetric and asymmetric stretching vibrations of C–H are clearly in full agreement with the results of LIBS. Furthermore, in EDX analysis, mass concentrations of carbon (\({\mathrm{C}}_{\mathrm{C}}\)) and calcium (\({\mathrm{C}}_{\mathrm{C}\mathrm{a}}\)) and also the ratio of \({\mathrm{C}}_{\mathrm{C}}/{\mathrm{C}}_{\mathrm{C}\mathrm{a}}\) entirely confirm the outcomes of LIBS. In the end, the linear discriminant analysis indicated that the five isogams, to a significant extent, are qualitatively different from each other.