Robust Remote Sensing of Trace‐Level Heavy‐Metal Contaminants in Water Using Laser Filaments

Abstract Water is the major natural resource that enables life on our planet. Rapid detection of water pollution that occurs due to both human activity and natural cataclysms is imperative for environmental protection. Analytical chemistry–based techniques are generally not suitable for rapid monitoring because they involve collection of water samples and analysis in a laboratory. Laser‐based approaches such as laser‐induced breakdown spectroscopy (LIBS) may offer a powerful alternative, yet conventional LIBS relies on the use of tightly focused laser beams, requiring a stable air–water interface in a controlled environment. Reported here is a proof‐of‐principle, quantitative, simultaneous measurement of several representative heavy‐metal contaminants in water, at ppm‐level concentrations, using ultraintense femtosecond laser pulses propagating in air in the filamentation regime. This approach is straightforwardly extendable to kilometer‐scale standoff distances, under adverse atmospheric conditions and is insensitive to the movements of the water surface due to the topography and water waves.


Temporal gating of the signal collection
In Figure S1, we show the emission spectra obtained with different ICCD gate delays. The gate delay is measured relative to the moment t=0 when the laser pulse arrives at the target. All spectra are measured with a fixed gate width of 1 microsecond. The Al concentration is 500 ppm. It can be seen from Figure S1 that because plasma excited through the interaction with an ultrashort laser pulse is cold, the useful Al spectral signal can be cleanly separated from the thermal emission background.
Supplementary Figure S1. Optimization of the gate delay. The ICCD gate width is set to 1 μs, and the gate delay varies from 60 ns to 200 ns. When gate delay is set to 200 ns, thermal background is no longer observed.
In Figure S2, we show the spectral signals for 500 ppm concentrations of four metals in water as a function of the ICCD delay time. The gate width is fixed at 1 μs.
Based on these data, we conclude that the gate width of 1 μs is sufficient to collect all useful signals on the spectral lines of interest.
Supplementary Figure S2. Detected signal with a fixed ICCD gate width of 1 μs, vs.

Assignment of the observed spectral lines
The spectral lines shown in Fig. 1 of the manuscript are summarized in Table S1.
The lines are assigned to the transitions in Al I, Cr I, Cu I, and Pb I according to the NIST database of atomic spectra (https://www.nist.gov/pml/atomic-spectra-database), as shown in Table S1.

Video demonstration of the insensitivity of the obtained spectral lines to water waves
Here we show videos of our experiment in the case when artificial waves are generated in the water cuvette, by a water-immersed mechanical shaker running at two different frequencies. It can be clearly seen from these videos that the extended light string in this approach overcomes the limitation imposed by the linear diffraction of the laser beam in the conventional nanosecond laser induced breakdown spectroscopy.
Vibrational Frequency: f = 43 min -1 Vibrational Frequency: f = 120 min -1 Supplement Video S1 (opens by double click). Videos of our experiment with undulating water surface. Artificial water waves are generated by a water-immersed mechanical shaker running at the vibrational frequency f = 43 min -1 (top) and f =120 min -1 (bottom).

Scalability of our approach to the realistic standoff distances
In the long-range implementation of the measurement of water pollutants, due to the phenomenon of intensity clamping in the laser filament, the level of the optical signal at the point of emission on the water surface will be similar to that in our laboratory-scale demonstration. To maintain the same level of the optical signal detected from a distance, the diameter of the signal-collection aperture needs to be scaled approximately in proportion to the distance. In our experiments, we detect the emission from a 1.1 meter distance, using a 5.08 centimeter-diameter collection lens.
Collection of the same order-of-magnitude signal from a 100 meter distance would require a meter-scale collection optic, which is within the feasibility range. Generation of multiple filaments would help improve the signal level. Further enhancements of the detection sensitivity could be achieved through the application of dual laser-pulse and dual-wavelength excitation, gating optimization and the optimization of spectrometer resolution for the detection of particular spectral signatures of interest.