Firstly, historical and spatial aspects of research into elemental analysis in aqueous solutions using LIBS are discussed. Sample types and sample preparation techniques to overcome the physical issues are then elaborated on. It is highlighted which sample preparations have given surpassing results and which are not recommended. The focus is less on a comprehensive discussion of the various methods of sample preparation, but rather on which methods are more commonly used and give better results for a specific application. The different types of instruments and their main advantages and disadvantages are also presented. For signal enhancement, different experimental setups are used and their main advantages and disadvantages are discussed. In addition, the most important acquisition settings are presented, as they strongly influence the analysis results. The analysis of the frequency of the different acquisition settings and the different sample preparation techniques used may help scientists who want to use LIBS for their research to make an informed choice for a particular application.
2. Sample Preparation Techniques
It has been shown that there is no standard approach to aqueous LIBS analysis in terms of sample preparation, LIBS experimental setup, or acquisition settings. Several very different techniques are used, each with its own advantages and disadvantages. This is due to the variety of possible applications, ranging from basic research in laboratory analysis, typically using standoff or self-arranged LIBS
[20][66], to remote analysis for hazardous substances, using telescopic systems
[21][32], and to in situ analysis for environmental concerns using online
[22][67] or portable instruments
[11]. Therefore, both the LIBS setup and the sample preparation technique should be selected according to the desired application.
However, one of the sample preparation methods can be excluded in the choice of sample preparation. Surface analysis of bulk liquids has many limitations, such as high laser energy requirements and low sensitivity, which have been extensively described in the literature (e.g.,
[23][93]), and is therefore not recommended. Splashing and liquid evaporation compensation methods should be used instead. Analysis inside the liquid is a bit more promising but is subject to shot-to-shot variability
[24][36].
Furthermore, in situ applications require sample preparation techniques that are easy to use in the field. Therefore, methods such as LB soaked on FP, SE LSC, and LSC FP are recommended. However, when selecting a sample preparation technique, it should be noted that several sample preparations also pre-concentrate the sample solution and thus can lower the LoDs significantly (e.g., most LSC techniques). This can help to achieve LoDs significantly below 1 mg/L. As portable instruments are typically less sensitive than laboratory devices using higher energies, sample preparation should include preconcentration steps. For online or laboratory analysis, more complex LIBS setups with more complicated sample preparation techniques can be chosen (e.g., LJ, LAC). Here, preconcentration may be helpful but not imperative. Some LJ analysis already reached sensitivities comparable to commercial ICP-AES devices without any preconcentration
[7] and could also be used in online and real-time analysis. For Na, cylindrical jets showed lower sensitivity compared to sheet jets
[7]. The special form of an LJ capillary effectively reduces splashing
[25][40] but brings along a more complicated experimental setup and costly consumables.
When only hydride-generating elements (e.g., As or Se) are to be analysed, HG is a good choice as it significantly improves sensitivity by eliminating spectral and chemical interference
[26][63]. However, this method is limited to a small group of elements, which can form hydrides (e.g., As, Sn, Sb, Se, Ge, Pb, Bi, Te), and it requires a complicated measurement setup.
Coupling LIBS with ion exchange techniques can also be a good option to improve sensitivity and reduce matrix effects in LIBS liquid analysis
[27][47]. However, this also requires a complicated setup, which reduces the advantage of simple and fast sample preparation over other methods such as IC. The same is true for the isolated droplet technique. An easy way to avoid splashing is the LSC technique of freezing the sample
[28][29][60,61]. This sample preparation does not require much expertise or a complicated setup and could also be performed in situ using liquid nitrogen
[29][61]. However, no preconcentration is achieved this way.
ESD achieves preconcentration, but many chemicals are required, as well as additional parts and experimental knowledge. ED also pre-concentrates the samples and therefore has improved sensitivity
[30][31][58,59]. The main drawback is the comparatively long time required for analysis due to the deposition process (>>10 min). Ad/ab on wood chips
[32][49] or graphene oxide
[33][50] is also quite time consuming but can be realised cost-effectively and improves sensitivity as well. Chelating (e.g.,
[34][51]) is even more time-consuming, requires more chemicals, and is usually not possible for multi-element analysis. However, it can be more efficient in enrichment than FP
[34][51]. MeGe, as described by
[35][53], is very similar to chelating, but requires even more time in sample preparation due to the long drying process. This makes the method less favourable in comparison to the other presented methods.
Analysis in ID, either with an isolated droplet generator (IDG) or by acoustic levitation
[36][38], adds complexity in the experimental setup, which minimises its advantages over conventional analytical techniques. NPs have the potential to improve the sensitivity as well
[37][38][54,55] and are easier to introduce in sample preparation.
When using LSC of a droplet on a surface with no adsorption/absorption step, some form of SE is recommended to improve the homogeneous distribution and shape of the evaporation residue (EvR). Without any SE, an effect typically referred to as the coffee ring effect (CRE) occurs, leading to lowered repeatability due to inhomogeneous distribution of the EvR
[39][40][44,72]. SE can be implemented by making the surface more hydrophobic to improve the distribution of the EvR
[39][44] to introduce a geometric constraint (GC) technique
[41][42][94,95] or to combine both
[38][55].
3. Instrument Types, Experimental Setups, and Acquisition Settings
Furthermore, it is also possible to improve the various sample preparations with a special LIBS setup. These include, e.g., double-pulse (DP), laser-induced fluorescence (LIF), resonance-enhanced (RE), resonant (R), and microwave enhancement (MW) LIBS. Depending on the design of the LIBS, such setups may not be possible or lead to complexity that diminishes the main advantage of LIBS over other laboratory methods in being simple in sample preparation
[43][96]. The simplest method to use is DP, which typically enhances the signal ten times and for some elements up to 50 times
[43][96]. The other experimental setups are more complicated, require more experienced staff, and do not all allow multi-elemental detection, but can improve sensitivity.
The acquisition settings also affect sensitivity and should likewise be adapted to the method. For example, the gate delay should be long enough to minimise the influence of continuum radiation, which typically occurs at the very beginning of the breakdown and provides no useful information, in addition to degrading the signal-to-noise ratio
[44][26]. However, with overly long gate delays, the signal becomes weaker. Although it is possible to set the optimum delay time for each element individually, using different delay times for different elements prevents simultaneous detection. For this reason, when analysing multiple elements, a gate delay should be selected that is suitable for all the elements to be detected.
Another enhancement possibility, which is often easy to implement, is to use a purge gas such as Ar, which also amplifies the signal intensity significantly
[45][75]. Even some portable LIBS units are already equipped with Ar cartridges for signal amplification
[39][44].
A change in pulse duration, pulse energy, wavelength, or repetition rate essentially influences the energy that is supplied to the sample and is thus available for the formation of the plasma. Moreover, increasing the repetition rate reduces analysis time and improves averaging
[13]. Typical lasers used in LIBS liquid analysis are Nd:YAG lasers with a fundamental wavelength of 1064 nm. Using different wavelengths can improve detection. For example, the CO
2 wavelength of 10,600 nm is particularly suitable for the analysis of water, as it absorbs in the infrared range
[13]. However, the wavelengths used were between 266 and 1064 nm. The wavelengths 532, 355, (352), and 266 nm are the second to fourth harmonics of the Nd:YAG laser, with a fundamental wavelength at 1064 nm.
According to
[13], typical pulse energies for LIBS are between 10 and 500 mJ. In the literature reviewed, pulse energies between 0.2 and 800 mJ were found, with most in the range 0.2 to 100 mJ. Energies below 10 and around 100 mJ are the most common. The pulse energy must be high enough to ablate enough material to produce a strong signal. However, a low energy pulse combined with a shorter pulse width and higher repetition rate will also result in a higher power density
[13]. Handheld devices generally operate with lower pulse energies. For example, the SciAps Z300 uses a pulse energy of 5–6 mJ, a frequency of 50 Hz (10 Hz), and a pulse width of 1 ns
[46][97].
4. Elements and Their Reported Detection Limits
Looking at the elements analysed, several areas of interest stand out. The frequent testing of potentially toxic elements
[47][98] such as Cr (N = 87) or Pb (N = 76) is particularly striking. There is a very strong interest in the detection of these elements because they are problematic elements in drinking water
[48][99]. Also, Cu and Cd are analysed very frequently, which are also potentially toxic elements
[48][99]. Furthermore, there is a second group of elements that are often analysed: Na, (Mn), Ca, Mg, Sr, K, and Li. These are typical cations in natural waters and therefore of interest in natural water analysis.
It is striking that arsenic (As) is less often analysed then Cr, Pb, Cu, Cd, and Zn, although it is a very problematic element in drinking water
[49][100]. This may be because As is not as well analysed by LIBS as other elements such as Li and Cr. According to Cremers and Radziemski, the most common and useful spectral range for LIBS analysis is 200–900 nm
[13]. Unfortunately, there are few strong lines for As in this range that can be used for quantification. Therefore, for As quantification, it is recommended to use pre-concentration methods or to choose HG.
S, N, F, Br, and P (and to a lesser extent Cl) are also relatively rarely analysed, although they are typically present as anions in natural water. This is also due to the fact that these elements are not analysed as well by LIBS as other elements such as Li and Cr. These six elements have among the highest ionisation energies of the 56 elements for which detection limits have been found. The ionisation energy plays a major role for these elements, since the lines typically chosen for quantification lie in or near to the infrared range (except for P) and are difficult to excite due to the high ionisation energy
[50][101]. The intensities are therefore usually so weak that low concentrations cannot be detected
[50][51][101,102]. For example, the strongest lines for F and Cl are below 200 nm, below the range of typical CCD detectors and in a range where a vacuum is required in the light path
[52][103]. Therefore, a detour via indirect analysis is usually used (for example, molecular emission, as compounds such as CaF and CaCl show higher intensities)
[14]. Another option is to analyse the excess of a more easily analysable element (e.g., Ag or Ba) that reacts with the anions by precipitation
[50][101].
It is also interesting to note that some atypical elements such as Tc were analysed in water. The reason for this is the possibility of remote LIBS analysis. Samek et al. analysed Tc with a telescopic LIBS system over a distance of 3 m
[21][32]. Also, the catastrophe of the Fukushima nuclear disaster has led to development of LIBS aqueous analysis for elements like Cs, Sr, and Zr
[53][54][55][56][43,104,105,106].