LIBS Basics, Part III: Deriving the Analytical Answer — Calibrated Solutions with LIBS

Oct 01, 2014
Volume 29, Issue 10

This installment is the third in a three-part series focusing on major aspects of laser-induced breakdown spectroscopy (LIBS). The first installment set the stage with the basics of the LIBS measurement physics and standard applications. The second installment discussed the choices and trade-offs for LIBS hardware in detail, in particular focusing on lasers and spectrometers. This third installment discusses LIBS analysis in some depth, exploring the various ways to go from a LIBS spectrum to a solution.

One of the strengths of laser-induced breakdown spectroscopy (LIBS) is that it can be applied to multiple media. In various configurations, it is possible to measure solids, liquids, gases, and suspended aerosols. Measurements have been made on the ocean floor at depths of hundreds of meters, and on the surface of Mars in an atmosphere of 7 Torr, largely consisting of CO2. This flexibility is possible because the atomization and excitation source are combined in the LIBS laser pulse. This strength of LIBS is a complication, however, because the material ablation and plasma formation process profoundly impacts the analytical plasma, as we pointed out in the previous installments in this series (1,2). As a result, users must be careful when developing material classification and calibration methods for LIBS, in particular using "like" materials for calibrations and choosing methods consistent with the type and range of expected data. While it is impossible to cover the gamut of considerations in a short column installment, here I will attempt to cover some major features of the calibration and classification landscape.

Basic Calibration

Any discussion of calibration should start with univariate analysis of spectral peaks, the primary method of calibration of spectroscopic output since such analysis has been attempted. Spectral peak analysis in the context of LIBS spectra has a number of considerations. First, astute readers will note that the LIBS plasma is a time-dependent event. The acquisition period of the spectra relative to the plasma initiation becomes an important parameter in the quantification. This collection period is also a function of the laser energy density on the plasma, because the plasma lifetime scales with the deposited energy in the plasma, and the ablation event is important (primarily for solids) not only for exciting analytes, but also in determining the amount of laser energy channeled into the plasma. Hence, any calibrated LIBS system should take care to ensure consistent laser fluence (energy/area) on the samples.


Figure 1: Spectra of chromium aerosol collected at different delay/gate combinations, indicated by delay/gate in microseconds to the left of each spectrum. Spectra are offset for clarity, but are otherwise the same scale. Adapted from reference 3 with permission.
The influence of the data collection timing has been shown often, for example in an early paper by Fisher from Hahn's group (3). Figure 1 shows the chromium emission lines from particles in an airborne aerosol at three different settings of delay (after the plasma initiation) and gate width (shutter open time), collected with a Czerny-Turner spectrometer and an intensified charge-coupled device (ICCD) camera. At 2 μs delay and gate, the chromium triplet is not visible at all. At later times and longer collection periods, the signal-to-noise ratio (S/N) of the triplet is optimized. Fisher's paper goes on to point out that the optimum detector timing to maximize S/N is element-dependent under otherwise constant experimental conditions.


Figure 2: Example of peak integration and baseline (background) definitions.
After the data collection (laser energy, spot size, and collection timing) is fixed, data can be collected from known samples to form a calibration curve. Typically, a particular peak is selected for analysis of an element of interest. As shown in Figure 2, the peak (blue, solid fill) is defined and integrated, with the subtraction of the background, which can be defined by the area under the peak, defined by the peak edges (green, grid fill), or by nearby areas with no spectral interferences (orange, diagonal fill). Sometimes the peak area alone can be used for calibration, but in other cases the peak is divided by the background, sometimes called the "baseline," to form the "peak/base" ratio. If the spectrum is collected at proximate enough time to the plasma that there is a measurable continuum, this peak/base calibration tends to work well, because the baseline is indicative of the energy absorbed in the plasma. In other cases, normalization of the peak by the total plasma emission (Figure 3a) can be useful if the total plasma emission is fairly constant (for example, measurement of a low concentration alloying element in a steel matrix). If there is a known, fairly constant element, another similar approach is to use a line of the constant element to standardize the peak of the element of interest (Figure 3b). For groups of samples with a consistent or relatively simple background, univariate methods are often very successful calibration schemes for LIBS.

Extensions to Basic Calibration — Line Selection


Figure 3: Univariate calibration of the (a) chromium 397.67-nm line divided by the total light intensity from the plasma, and the (b) chromium 397.67-nm line divided by the iron 404.58-nm line, in a range of low- to high-alloy steels. Adapted from reference 4 with permission.
Individual emission lines each have a particular range of applicability. Lines with low excitation energy and ending at or near the ground state tend to have a nonlinear response at higher concentrations because of self-absorption, but these lines are the most sensitive for measurements at low concentrations. Harder-to-excite lines, particularly those not ending on the ground state, are more useful for higher-concentration measurements. Lines not ending on the ground state are not as susceptible to self-absorption as lines that transition to the ground state (there is negligible population to absorb in upper states). The result of these factors is that calibration lines are selected based on the concentration range being measured, and often the strongest lines in the spectrum of a particular element are not used for higher-concentration measurements. Similar to inductively coupled plasma (ICP) methods, calibration over a large range of analyte concentrations may use a succession of lines, some for lower concentrations and some for higher concentrations.


native1_300x100
lorem ipsum