Laser-induced fluorescence of filament-produced plasmas

The ability to detect elements and isotopes at standoff distances is a highly desirable capability for safely identifying hazardous materials, in addition to space, atmospheric, and environmental monitoring.^1–4 Ultrafast laser filaments is a promising mode for extended delivery of the laser pulses at large distances as an excitation source for sensing applications.⁵ Filaments are generated when an intense femtosecond (fs) laser pulse is propagated in a transparent medium (e.g., air). The beam self-focuses as a result of the nonlinear Kerr effect and is defocused due to the negative refractive index of the plasma medium. The dynamic balance between the self-focusing and defocusing phenomena leads to the generation of filaments in the laser beam path that become self-guided and can be propagated to hundreds and even thousands of meters.^5,6 The distance of filament formation can be manipulated by adjusting the laser chirp at the laser frontend.

Filament ablation combined with optical emission spectroscopy or filament-induced breakdown spectroscopy (FIBS) is an emerging tool for standoff and remote detection applications.^3,7–13 Although ultrafast filaments have several attractive properties, such as the ability to be self-guided and the production of standoff plasma, further research efforts are needed to improve laser–target coupling and, therefore, plasma emission signal strength. Filaments carry only a fraction of the energy of the ultrafast laser beam. The power required for filament generation (called the critical power, $Pcr$⁠) is $≈3$–10 GW for an 800 nm laser beam, and the remaining laser energy is stored in the energy reservoir surrounding the central beam. Once the laser reaches the critical power for filamentation, intensity clamping occurs, which defines the upper limit for the laser intensity at the self-focus point. Increasing the laser energy of the propagating beam can lead to the generation of multiple filaments,¹⁴ which can affect the geometry of the energy reservoir, density, and temperature of the ablation plume and, therefore, the resultant emission intensity. Changes in plasma properties and emission signal strength can be attributed to variability in laser–target coupling due to the stochastic nature of filament generation.^15–17 Thus, significant research efforts are still needed to overcome such challenges and allow for the practical application of filaments for standoff material detection, including in-depth understanding of the filament ablation mechanisms, changes in the ablation properties due to multiple filament formation, and control of the stochastic nature of filament generation and its effects on plume emission properties.³ 

One of the challenges associated with laser-induced breakdown spectroscopy (LIBS) is its analytical inferiority compared to other laboratory-based methods (e.g., mass spectrometry). On the other hand, LIBS is a superior analytical tool for standoff detection.^1,18 Many methods have been suggested for improving the emission intensity of LIBS, including double pulsing, cavity confinement, microwave-induced enhancement, nanoparticle enhancement, ambient or wavefront control, and beam shaping.^19–21 Other laser ablation (LA) hyphenated methods such as LA laser absorption spectroscopy (LA-LAS) and LA laser-induced fluorescence (LA-LIF) have shown improved analytical capabilities compared to LIBS such as reduced matrix effects, enhanced precision, and improved signal-to-noise ratio, detection limits, and spectral resolution.^22–27 However, given that LIF and LAS are active sensing tools, a second laser with strict requirements for the optical transition of interest is required. LAS also has specific geometry requirements, where the second (probe) laser beam must pass through the plasma parallel to the target to allow for absorbance to be measured, making LAS difficult to use for standoff detection. Since there are no such geometrical constraints for LIF, it is well suited for standoff analyses.²⁸ Demonstrated applications of LIF for LA plumes include isotopic analysis of uranium²⁹ and Li;³⁰ trace element detection in metallic materials (e.g., boron in nickel-based super alloys²⁴), lead in aqueous solutions,²⁶ and heavy metals in soils;²⁵ and studying plume dynamics.³¹ Other advancements in the application of fluorescence spectroscopy to studying LA plumes include two-dimensional fluorescence spectroscopy (2D-FS), which provides simultaneous excitation and emission spectra.^32,33

Several studies highlighted the differences in physical properties of filament-generated plasmas compared to focused fs or ns laser produced plasmas (LPPs).^11,13,34 Recently, LIF was used in combination with LIBS and LAS to characterize and compare the physical properties of focused fs and ns LPPs over the entire plasma lifecycle.³⁵ LIBS was used to monitor early times in plasma evolution through tracking the excited state populations of Al I species, while LIF and LAS measurements were used to characterize the plasma at lower temperatures and later times because these methods use lower-state populations. The temperatures and temporal histories of plasma species are influenced by environmental conditions (i.e., pressure, reactivity of the ambient gas) and can affect LIBS, LIF, and LAS spectral features. In atmospheric pressure and oxygen-containing environments such as air, both emission and LIF signal intensity can be influenced by plasma chemistry.^36,37 Filament-generated plasmas in air are cooler than those produced from focused fs-laser pulsing because of the reduced laser–target coupling. Therefore, the molecules were generated earlier in filament-generated plasmas compared to focused fs-laser ablation plumes.^38,39 For reactive plasma species (e.g., Al, U), atoms will be consumed in the formation of diatoms or higher oxide molecules, affecting both excited and lower-state populations and thus emission and LIF signals.³⁸ The impact of the reactivity of plasma species with the ambient environment and the role of background gas in confinement are all critical considerations in the application of LIF using any ablation mechanism, including filaments.

Recently, standoff LIF analysis of ns laser-produced plasmas was demonstrated.²⁸ However, the diffraction of the laser beam is the bottleneck for propagating the ns laser beam to larger standoff distances. The filaments generated during fs-laser propagation can be generated at large standoff distances by negating the diffraction effects. However, for any standoff application, improving the SNR of the collected signal is very important. In this article, we report the use of LIF for signal enhancement of standoff FIBS, which has not yet been reported. We perform complementary LIF and FIBS measurements by focusing the incident ultrafast laser at a distance up to 10 m. Both temporally integrated and resolved measurements were performed in order to compare changes in spectral profiles and signal intensities for LIF and FIBS methods. Our results demonstrate that LIF of filament-produced plasmas at standoff distances effectively boosts the FIBS emission signal and reduces self-reversal features. 2D-FS was also performed to evaluate optical saturation effects and to characterize plasma properties at varying times after the onset of the filament-produced plasma.

A chirped-pulse, Ti:Sapphire amplifier laser system (Coherent Astrella) with a center wavelength of $≈800$ nm and a pulse duration of $≈35$ fs full width half maximum (FWHM) was used for filament generation and subsequent ablation at various standoff distances. The laser was operated at a pulse energy of $≈5$ mJ and a frequency of 5 Hz. For generating filaments in the laboratory, the fs-laser beam was loosely focused using f = 1, 2, 5, and 10 m plano-convex lenses in the air at atmospheric pressure (⁠$≈760$ Torr). The target used was an Al alloy (Al6061) that contains Mg, Si, Fe, Cu, Cr, Zn, Mn, and Ti as minor alloying elements.

For both LIF and FIBS measurements, light emitted by the plasma was collected in a spatially integrated manner by positioning the collection optics $≈25$ cm from the target. We note that the signal intensity will vary with distance (r) relative to the light source (here, LPP) by a 1/r$2$ dependence. A 1,500 $μ$m fiber was used for light transport to a 0.5 m Czerny–Turner spectrograph (Acton SpectraPro SP-2500) that contained an intensified charged coupled device (ICCD; Princeton Instruments PIMAX4) as the detector. The system provided a maximum spectral resolution of $≈40$ pm at $≈400$ nm.

For performing LIF, a continuous-wave (CW) tunable Ti:Sapphire laser (⁠$M2$ SolsTis) in conjunction with a resonant cavity frequency doubling unit was used. The probe laser provided a narrow linewidth output (⁠$≈$100 kHz) tunable over a spectral range of 700–1000 nm and 350–500 nm (second harmonic), with power up to $≈200$ mW at 394.4 nm. Both the filament ablation pulse and LIF laser were propagated to the target at an angle of $≤10°$⁠. The LIF laser beam size at the target was $≥$3 mm, which encompassed the entire plasma plume. The probe laser was tuned to excite the Al 394.40 nm (0–25 347 $cm−1$ ) transition, which shares the same upper energy level with Al 396.15 nm (112–25 347 $cm−1$⁠). Directly coupled LIF emission was collected at Al I 396.15 nm to avoid the spurious background signal caused by LIF laser scattering. 2D-FS was performed by stepping the CW laser across the selected transition. 2D-FS was also performed by overfilling the plasma, similar to LIF and FIBS measurements previously described, and hence is a spatially integrated measurement. The experimental setup, a photograph of filament ablation at 10 m, and the energy level diagram of the selected transitions for filament LIF spectroscopy are given in Fig. 1.

Emission signal intensity and spectral profiles are important considerations for elemental or isotopic detection using optical spectroscopy of LPPs. In this study, the FIBS and LIF emission features are collected from filament ablation plasmas generated at various standoff distances. Typical FIBS and LIF signals collected from a standoff distance of 10 m are given in Fig. 2, collected using a gate delay of 2 $μ$s and an integration time (gate width) of 15 $μ$s. These results show significant emission enhancement for LIF of filament-generated plumes compared to thermally excited emission. Here, the FIBS signal (fs-laser energy of 4.6 mJ) is boosted by $≈$22 times for a LIF laser power of $≈$90 mW. It is important to note that the LIF pump beam overfilled the filament ablation plume, and hence, it is not optimized. The typical diameter of a single filament is $≈$100 $μ$m, and the incident LIF beam had a diameter of $≥$3 mm. Hence, the LIF beam was large enough so that even in the case of multiple filamentation (⁠$≈$500 $μ$m diameter ablation crater), or any beam wandering, the entire plume would still be encompassed by the LIF beam. This experimental configuration is preferable for standoff detection to account for the atmospheric wandering effects of both beams.

Understanding filament-target interactions is important for FIBS, where signal intensity depends strongly on the filament coupling with the target. Typically, LIBS intensity scales with the energy of the incident laser used for plasma generation.⁴⁰ However, this trend may not be true for filament-generated plasmas because higher laser energies lead to the formation of multiple filaments. For evaluating the changes in FIBS and LIF signal intensities, the fs-laser energy was varied from 2.3 to 5 mJ. Considering that the critical power, $Pcr$⁠, for filament generation at 800 nm is $≈$3 GW, this energy range corresponds to 19–41 $×Pcr$⁠.

Changes in FIBS and LIF signal intensity with increasing fs-laser energy are given in Fig. 3 for a 10 m standoff distance. Below $≈$2 mJ, the emission signal was found to be negligible. The measurements were carried out with a gate delay/width of 2 $μ$s/15 $μ$s. The LIF signal enhancement is increased from $≈$20–87 times when the laser energy increased from 2.5 to 5 mJ. At a laser energy of 3.4 mJ, signal enhancement is maximum (87$×$⁠). The laser energy range used in the present experiment may belong to the transformation from single to multiple filaments.⁴¹ However, multiple filamentation does not profoundly affect the emission signal for the laser energy range studied given that the signal intensity increases approximately linearly with the laser energy, as shown in Fig. 3.

In addition to the filament laser energy, the LIF laser power will also affect the LIF signal intensity. Figure 4 reports the changes in LIF signal intensity with LIF laser pump power. The filaments were generated at a standoff distance of 10 m, with a constant fs-laser energy of 5 mJ. As shown in Fig. 4, when the probe beam power is increased, the LIF intensity follows an approximately linear relationship with laser power. This relationship suggests that LIF depletion of the ground state (or bleaching) is negligible. If bleaching effects were present, the LIF signal intensity would plateau at higher laser powers, indicating that any additional pumping does not result in excitation and a corresponding increase in the signal intensity. This plateau corresponding to depletion of the ground state population has previously been reported in LIF studies using a pulsed probe beam.²⁶ However, in the present experiment, a CW laser was used for LIF pumping, and such optical saturation effects are not expected. It should be noted that given the stochastic nature of filaments, a fluctuation in the LIF signal can be expected and is captured by the standard deviation from duplicate measurements, given as the y-error bar in Fig. 4. In addition, reabsorption of the LIF photons may also generate nonlinear signal behavior.⁴² The LIF power dependence study also highlights that even lower power levels of LIF pumping boost signal intensity. Also, note that the LIF pump beam size was significantly larger than the filament-generated plume size, and hence, it was not optimized.

The temporal evolution of FIBS and LIF emissions was monitored by delaying the detection gate width (0.5 $μ$s) with respect to plasma onset, and the results are given in Fig. 5 for a 10 m standoff distance. FIBS uses thermal excitation, and its emission signal, therefore, peaks at early times of plasma evolution when the temperature of the LPP system is higher. However, the thermally excited signal is also found to decrease rapidly with time. The temporal history of the LIF signal showed a significantly different temporal history. The LIF emission intensity showed a gradual decay and persisted for long times compared to the FIBS signal. The slower decay in the LIF signal can be correlated with an increase in the lower-state population of Al atoms because of the reduction in the plasma temperature with time. The rapid decrease of the FIBS signal after filament plasma onset can also be related to the rapid decay in the plasma temperature, where higher temperatures are required to have a large excited state population. Thus, LIF can effectively extend the signal lifetimes in LPPs and characterize the plasma at cooler temperatures when line broadening and number densities are lower. Here, the use of a CW tunable laser for LIF excitation provides constant re-excitation of the lower-state population during the entire lifetime of the plasma plume. The CW excitation also provides simplicity in timing and provision for boosting the LIF signal by increasing the LIF power in comparison with pulsed LIF excitation.

A change in the standoff distance may affect the FIBS and LIF signal intensities due to beam wandering effects,¹⁸ changes in the properties of the filament energy reservoir that may undergo diffraction,³ and LIF beam coupling with filament ablation plumes. Therefore, we carried out a study by generating filaments at various standoff distances. Here, for generating filaments at various standoff distances, the fs-laser pulses were loosely focused with f = 1, 2, 3, 5, and 10 m, and results are given in Fig. 6. The laser energy used was 5 mJ, and LIF pump power was 90 mW. Signal intensities were determined from the averaging three spectra, where each spectrum was averaged over 10 shots. Both LIF and FIBS signals are within the expected 20% variability, and no clear trend with filament generation distance is observed. In addition, LIF signals reported in Fig. 6 vary by a factor of $≈$1.45, whereas the fluctuations in FIBS signal are more significant and vary by a factor of $≈$4. Typically, femtosecond laser filaments are generated in the laboratory by loosely focusing the laser beam using a very long focal length lens similar to the present experiment. However, for long standoff distances, the filaments can be generated by adjusting the laser chirp where self-focusing is the only mechanism driving the self-collapse (i.e., lens-free filaments). A previous study compared the emission features of plasmas generated using loosely focused fs filaments or lens-free filaments, and results showed that the filament generation conditions appreciably influenced the plasma properties and emission features.³⁴ Changes in ultrafast laser–target coupling have previously been linked to ablation efficiency and changes in plasma properties, where a loosely focused beam led to increased persistence for atoms and molecules in LPPs, in comparison with freely propagating filaments.³⁴ 

The time-resolved FIBS and LIF signals highlight that LIF effectively boosts emission signals over the lifetime of the plasma (Fig. 5). At a given gate delay/width of 2 $μ$s/15 $μ$s, standoff LIF analysis of filament-generated plasmas also clearly shows that LIF boosts the emission signal for standoff distances ranging from 1 to 10 m (Fig. 6). For an LPP, the spectral features change significantly with time due to changes in various broadening mechanisms. Therefore, temporally resolved spectral features were collected to illustrate the important differences in the line shape between LIF and FIBS, with results given in Figs. 7(a) and 7(b), respectively. The thermally excited Al spectral profiles show the presence of an apparent dip at all times of its evolution, indicating self-reversal. It is well known that the self-absorption processes in the plasma may lead to the broadening of an emission line because the absorption is the strongest at the line center and the weakest at the wings of the spectral profile. For inhomogeneous plasmas such as LPPs, a central dip in the line profile or self-reversal is also expected for optically thick lines due to the absorption in the cooler outer layers of the plasma during radiation transport.⁴³ These processes can lead to an increase in measured linewidths and a decrease in the peak height and is an undesirable effect in emission spectroscopy measurements. However, as shown in Fig. 7(b), the spectral dip seen in the FIBS spectral features from 0.1 to 1 $μ$s is not present in the LIF boosted emission signals at times $≥$ 0.5 $μ$s after the plasma onset.

The LIF boosted emission spectrum at very early times of plasma evolution (e.g., 500 ns) also shows a central dip, which is due to the presence of thermally excited emission. As time evolves, a sharp LIF emission peak compensates for the spectral dip seen in the FIBS emission. At later times of plasma evolution, the LIF emission features show a narrow peak with no indication of self-reversal effects. Hence, our results highlight that LIF addresses some of the challenges associated with thermal emission spectral analysis (i.e., self-absorption and/or self-reversal). Considering the collinear LIF pumping scheme and that the LIF beam encompasses the entire plasma plume, it can be anticipated that the LIF beam preferentially interacts with and excites the colder outer layers of the plasma plume, which may reduce the self-reversal effects. While the central dip in Fig. 7(b) spectra suggests that opacity issues of the plasma are prominent in FIBS, the lack of a central dip in the LIF spectra does not necessarily preclude optical saturation effects.

All measurements given above correspond to LIF boosted emission signals, and hence, the spectral features contain signatures from both thermally excited and LIF emission. In addition, the spectral resolution is constrained by the spectrograph resolution. The optical saturation effects can be better understood by evaluating the LIF excitation spectral features. Thus, to further examine absorption, emission features, and plasma properties, time-resolved 2D-FS was performed, with results reported in Fig. 8. For 2D-FS, the LIF signals were collected when the excitation laser wavelength was stepped across the selected transition. The key parameters for 2D-FS are excitation wavelength, emission wavelength, and gating (gate delay/width).⁴⁴ 2D-FS reported here were taken at delay times from 0.5 to 10 $μ$s at progressively increasing gate widths (2–10 $μ$s) to account for the reduction in the LIF intensity at later times. The recorded 2D-FS at an early time of plasma evolution (0.5 $μ$s) shows self-reversal at off-resonance positions, as evidenced by a dip at the peak center [Fig. 8(a)].

The excitation spectra obtained at various times from the 2D-FS contours are given in Fig. 9. The smooth curves in Fig. 9 are Voigt fits. The measured FWHMs from the Voigt fit of excitation spectra are given in Fig. 10 for times ranging from a gate delay/width of 0.5/2 to 15/15 $μ$s. With increasing time after plasma onset, the excitation spectra linewidths decrease. For example, the measured FWHM decreases from $≈$5.8 pm at 0.5 $μ$s to $≈$5.6 pm at 2 $μ$s to $≈$4.5 pm at 15 $μ$s. Emission spectra have significantly higher linewidths of $≈$300 pm at 0.5 $μ$s, and $≈$100 pm at later times of 2 $μ$s–15 $μ$s.

Because an Al-rich target material is used for the present experiment, the optical saturation effects due to high absorbance cannot be ruled out in the LIF excitation spectra. Since a CW laser is used in the present experiment for LIF excitation, the signal saturation effects due to the depletion of lower levels are expected to be minimal. However, under conditions of high absorbance, the LIF intensity may vary nonlinearly with the atomic number density. The relationship between the detected LIF signal intensity [$SF(λex,t)$], incident laser intensity (⁠$I0$⁠), and absorbance [$A(λex,t)$] can be described as $SF(λex,t)=S0×I0×[1−e−A(λex,t)]$⁠, where $S0$ is a constant that accounts for collection area/efficiency and fluorescence quantum yield, $λex$ is the wavelength of the LIF excitation laser, and $t$ is the time.⁴⁵ For a spatially uniform plasma, the absorbance is defined as $A(λex,t)=σ12(λex)×N1(t)×L$⁠, where $σ12(λex)$ is the absorption cross section, $N1(t)$ is the atomic population in the lower state, and $L$ is the length. If optical saturation effects are absent, the $SF(λex,t)$ should vary linearly with $I0$⁠. However, the above relation may not be valid in high absorbance conditions. Hence, the high atomic number densities in the plume may distort the shape of the excitation spectrum in a manner similar to the self-absorption effects observed in emission spectroscopy. In addition, 2D-FS measurements were all performed in a co-linear geometry in which the LIF laser beam overfilled the LPP and hence are spatially integrated measurements. Considering the heterogeneous nature of the filament-produced plasma, this spatial integration could also contribute to distortions in excitation spectra line shapes.

Although an approximate linear dependence of the LIF signal with respect to the LIF laser power is noticed in Fig. 4, the time evolution of the FWHMs given in Fig. 10 does not show a monotonic decrease with time. In the absence of any optical saturation effects, the FWHM of the excitation spectra would show a different trend than is given in Fig. 10; there would be a rapid decay of FWHM due to the reduction in the plume temperature and the number density with time. The measured Gaussian components of excitation spectra at 10 and 15 $μ$s are 3.75 and 3.67 pm, respectively. These Gaussian widths correspond to Doppler temperatures of $≈$4750 K (10 $μ$s) and $≈$4560 K (15 $μ$s). Based on prior work, excitation temperature of a filament-produced plasma from an Al target in atmospheric air at 10–15 $μ$s after plasma onset is significantly lower than measured here.³⁹ These results highlight that the optical saturation effects may be present in the recorded LIF spectral profiles due to the presence of high atomic density medium. This finding is also supported by the fact that the measured FWHM given in Fig. 10 is not decreasing monotonically with time as the FIBS plume expands into an ambient medium.

In addition, the measured spectral profiles of Al I (396.15 nm) showed significant broadening at times $≈$500 ns and 1 $μ$s (Fig. 7). The major mechanisms that contribute to line broadening in plasmas generated by lasers are Stark and Doppler. In addition to these, van der Waals can contribute, and the Al I transition used in the present study may undergo resonance broadening. The spectral lines may also be broadened due to the presence of the hyperfine structure. The reported van der Waals pressure broadening coefficient of the selected Al I transition in Ar is $γ=21.9±0.4$ MHz/Torr (FWHM) at 300 K.⁴⁶ However, the van der Waals broadening coefficient in air is not reported. The estimated broadening contribution by the Doppler effect is negligible. Electron density was calculated using Stark broadening of Al II transition at 281.62 nm, using the impact parameter obtained from Konjević et al.⁴⁷ The measured electron density is found to decay from $2.75×1017cm−3$ (50 ns after plasma onset) to $4.25×1016cm−3$ (250 ns after plasma onset). At later times, the measured linewidth approached the instrumental limit of the spectrograph. The Stark broadening of the Al I line at 396.15 nm is expected to be lower than the instrumental linewidth at times $≥$500 ns by considering the reported impact parameter.⁴⁸ We verified this argument by measuring the line broadening of Al I (396.15 nm) in filament-induced plasmas generated from a target with a minor Al alloying addition (Inconel target with only 0.4% Al by wt.) and found that the linewidth of the transition reached the instrumental limit by 500 ns. Hence, we attribute the broadening seen in Fig. 7 for Al I spectral profiles at later times (500 ns and 1 $μ$s) to distortions from self-absorption/reversal due to high Al atom density.

In this article, we demonstrate the application of LIF for boosting signal levels of standoff FIBS. Ultrafast laser filament-produced plasmas were generated from an Al6061 target at varying standoff distances from 1 to 10 m. For all distances, self-similar LIF and FIBS intensities were measured. Both FIBS and LIF signals were found to have an approximately linear relationship with fs-laser energy, and the LIF intensity was found to increase approximately linearly with CW LIF laser power. The LIF signal enhancement increased from $≈20$–87 times when the fs-laser energy increased from 2.5 to 5 mJ. A sharp central dip was observed in thermally excited emission spectra, but was absent from LIF spectra, suggesting that LIF is useful for overcoming self-reversal effects in emission spectroscopy. Given the collinear LIF pumping scheme and the LIF beam encompassing the entire plasma plume, it is hypothesized that the LIF beam preferentially interacts with and excites the colder outer regions of the plasma plume, which may reduce re-absorption effects. 2D-FS was performed to measure nonlinear effects in LIF emissions due to high Al atomic densities, and optical saturation effects were found to be non-negligible.

Although LIF clearly reduces self-reversal effects present in FIBS spectra and boosts the emission signal, self-absorption effects may still be present in the LIF spectra. It has to be mentioned that an Al alloy target (Al mass $≈$97%) is used in the present experiment and hence optical saturation effects are not unexpected; however, such a scenario is not anticipated for trace elements. Advanced data analysis methods, such as CSigma or curves of growth methods,^49,50 developed to overcome self-absorption effects in emission spectroscopy may also be useful for addressing optical saturation effects in LIF, particularly with the analysis of an analyte with high concentrations in the target material.

This work was supported by the Department of Energy (DOE)/NNSA Office of Defense Nuclear Nonproliferation Research and Development (DNN R&D). Pacific Northwest National Laboratory is a multi-program national laboratory operated by Battelle for the U.S. Department of Energy under Contract No. DE-AC05-76RL01830.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

The authors have no conflicts of interest to disclose.