Expansion dynamics of ultrafast laser produced plasmas in the presence of ambient argon

Laser ablation (LA) plumes or laser-produced plasmas (LPP) have numerous applications. Some of the key applications are pulsed laser deposition (PLD),¹ nanoparticle, nanostructure and cluster formation,^2,3 laser induced breakdown spectroscopy (LIBS),^4,5 LA inductively coupled plasma mass spectrometry (LA-ICP-MS)⁶ and light sources for lithography,⁷ and microscopy.⁸ Understanding plasma plume dynamics is important for optimizing LPPs for all of these applications. The evolution of the plasma plume is highly transient in nature and is influenced by laser (wavelength, pulse duration, intensity, and spot size) and target (density, thermal properties, and geometry) parameters. Apart from this, the expansion features of LA plumes are influenced by the nature and pressure of the background ambient gas. It has to be pointed out that most of the present day applications of LPPs are performed in the presence of an ambient gas though the roles of background gas are different for various applications. For example, in LIBS an ambient medium is used for plasma excitation and confinement, in LA-ICP-MS the purpose of the ambient gas is for particle condensation and flushing the particles from ablation zone to ICP torch, in PLD the ambient gas acts as a moderator for particles and/or generating reactive species, in nanoparticle or cluster production the presence of ambient gas is for rapid cooling of the plasma and nucleation. However, the presence of an ambient gas during the plume expansion makes its evolution very complex by influencing plume deceleration, thermalization of ablated species, diffusion, recombination, and formation of external and internal shock waves.^9–12 Ambient gases may influence the mass ablation rate, and this affects the LPP emission spectra, electron temperature and electron density, thereby influencing the analytical capabilities of the LPP using LIBS or LA-ICP-MS.^13,14

Until recently, nanosecond laser pulses have been regularly used for LA analytical and other applications. However, with the advent of short and ultrashort laser pulses, there has been a growing interest in the applications of femtosecond LA.¹³ Ablation mechanisms differ significantly for fs and ns laser excitation due to the vast difference in pulse durations.^15–17 For ns LA, laser absorption, ionization, and evaporation occur during the duration of the laser pulse itself, thereby influencing the plume hydrodynamics. However, in fs LA the laser pulse duration is shorter than the characteristic relaxation times resulting in minimal or negligible laser-plume interactions. The ablation occurs in the ps time frame resulting in an explosion-like removal of electrons from the sample followed by ion removal. There are several advantages in using ultrashort pulses for laser ablation instead of short ns pulses in many applications, which results in reduced elemental fractionation in LA-ICP-MS,¹⁸ less droplets in PLD,¹⁹ precise micromachining,²⁰ and reduced heat affected zone (HAZ) and continuum emission in LIBS,²¹ etc.

Ambient gas and environmental conditions play a critical role in the laser ablation process and influence the plume hydrodynamics for both ns and fs laser produced plasmas.^22–24 In recent times, there have been numerous studies that focus on the nanosecond plume dynamics in different background ambient gases, typically air and Ar.^25–27 However, such studies are limited for the case of an ultrafast or femtosecond LPP.^9,28–30 In ns LA, the ambient gas influences the laser-target and laser-plasma coupling along with the plasma hydrodynamic expansion features.¹⁴ However, such laser-plasma coupling is absent in fs LA. Since the duration over which the ablation and plume evolution occurs is much longer than the pulse duration, the influence of the fs laser pulse on plume hydrodynamics is expected to be minimal, aside from influencing the initial conditions. However, it was shown that the ambient gas influences the expansion of a fs laser plume more than a ns laser plume.³¹ According to their study, a fs laser plume expands into a cold ambient gas, and hence the ambient gas properties can play an important role in plasma dynamics as well as emission lifetime. For ns LPPs, the plume expands into a heated ambient environment with longer plasma persistence resulting in a lesser influence of the ambient gas for ns LPP.³¹ Recent computational fluid dynamics (CFD) study of ns and fs laser ablation plumes at atmospheric pressure levels^32,33 showed oscillatory plume structures for ns LA while internal shock waves were absent in the case of fs LA.

In this study, the expansion dynamics and internal plume structures of fs laser ablated brass plasma in Ar at different pressures ranging from vacuum to atmospheric pressure levels were probed using time and spectrally resolved two-dimensional fast gated imaging. Hydrodynamic features of the fs LA plume were collected from spectrally integrated images for obtaining overall plume morphology at various times and pressure levels, while monochromatic line filters were used for obtaining emission distribution and expansion dynamics of two species (Cu I and Zn I) in the brass plume. The expansion features of the plume were also compared with various existing models and found to be generally in good agreement.

Experiments were conducted inside a vacuum chamber in which the pressure could be varied from ∼ 10⁻⁵Torr to atmospheric. A brass target consisting of roughly 70% copper and 30% zinc was used for the plasma generation. Plasma creation was accomplished by focusing pulses from a Ti: Sapphire laser system which uses a mode-locked Ti:Sapphire oscillator (Synergy, Femtolaser, Inc.) in conjunction with a chirped pulse amplifier (Pulsar, Amplitude Technology, Inc.) to produce 40 fs laser pulses, measured using an autocorrelator, at a wavelength of ∼ 800 nm. The laser energy used was 5 mJ with a spot radius of 300 μm. All the experiments were conducted in single shot mode. Fast photography was performed using an intensified CCD (ICCD, Model: PI MAX) camera placed orthogonally to the direction of plume expansion. An objective lens (Nikon Macrolens, f = 70–200 mm) was used to collect the self-emission from the plume, and visible radiation in the wavelength range of 350–900 nm was spectrally integrated. For imaging spectrally resolved emissions from copper and zinc species, monochromatic line filters at 510 nm (Cu I) and 481 nm (Zn I) were used, each with a 10 nm bandwidth. A 2 ns gate width was used to image the plasma at early times (<100 ns), and at later times a gate width of 10% of the sequential delay time was used to compensate for the reduction in intensity. A programmable timing delay generator was used to control the delay between the laser pulse and the ICCD with a time resolution of 1 ns.

Fast photography employing short time-gated ICCD provides 2D snapshots of the 3D plume expansion and is one of the simplest diagnostic tools for studying the plume's hydrodynamic expansion features. We selected brass for the present study since it has been shown in numerous studies that a brass target is subjected to elemental fractionation during LA owing to huge differences in the thermal properties of Cu and Zn which affects LIBS and LA-ICP-MS quantitative measurements.^34,35 Fig. 1 shows a representative plume morphology and expansion from spectrally integrated and spectrally filtered emission at a delay of 300 ns after the onset of plasma for various Ar pressures. For obtaining the plume morphology of all species in the plume, the self-emission from the plume was spectrally integrated in the wavelength range 350–900 nm. For imaging species-resolved emission, narrow-band filters were used for collecting emission from Cu I at 510 nm (3d¹⁰4p-3d⁹ 4s²) and Zn I at 481 nm (4s5s-4s4p). The duration of integration for obtaining the images given in the figure was 30 ns. To account for the large differences in absolute emission intensities as the pressure is varied, each image was normalized to the maximum emission intensity recorded within that image. It should be asserted that these spectrally filtered images provide information about a particular emission line for a specific excitation energy, and therefore, the images are not representing the distribution of the entire Cu or Zn species in the plume. However, these images provide a good tool for studying the hydrodynamics of LPP plume and species dependent collisional excitation.

From Fig. 1, it can be clearly seen that the ambient pressure significantly influences the plume morphology, plume excitation and expansion. The fs laser plume expands adiabatically and freely in vacuum. The forward centric nature or cylindrical geometry of the fs LA plume in vacuum was reported previously and explained as pressure confinement due to strong overheating in the laser impact zone.^21,36 The interaction between the plasma and ambient gas species is found to be similar to the vacuum case until the background gas reaches 10s of mTorr pressure levels. The images given at 100 mTorr clearly show the collisional effect indicated by plume confinement in the target normal direction. It can also be seen that at moderate pressure levels, ∼0.1–1 Torr, the plume morphology is more spherical. At still higher pressures, ∼10 Torr, the plume is found to be confined both in the radial and lateral direction (with a torpedo-like shape). Compared to ns LA, in fs LA most of the ions, as well as ablated mass, are ejected from the target surface in a narrow cone angle with respect to the target normal.³⁷ This results in cylindrical geometry for fs LA plumes at higher pressures due to less confinement in the plume expansion direction in comparison to the axial direction. At higher pressures, >100 Torr, the pressure exerted by the ambient gas is very high so that the expansion of the plume is limited to a few mm.

Because of the highly dynamic behavior of LPP expansion, the interaction of the plume with an ambient gas is best characterized in a combined time- and space-resolved manner. Hence, space-time contours were generated from the ICCD images obtained at various pressure levels, viz., vacuum, 0.1 Torr, 0.5 Torr, 1 Torr, 10 Torr, 100 Torr, and atm Ar pressures, respectively, and results are given in Figs. 2–8. These space-time contour images were obtained from the ICCD image sequences by averaging 120 μm emission zone across the plume expansion direction from the laser impact zone. These contour images provide a comprehensive picture of the evolution history of the expanding brass plume and the various species (Cu and Zn) in the plasma from the onset of plasma formation to 1 μs.

Fig. 2 shows the space-time contour plots of spectrally integrated (brass) and spectrally resolved (Cu and Zn) emission in vacuum, and it can be seen that plume undergoes a nearly free-expansion in these conditions. Two components of the plume can be clearly observed: a fast moving component moving away from the target surface along with a slow moving component staying closer to the target even at later times. Such two-component plume structures have also been reported previously by others^3,9 for fs LA and the fast moving component essentially consists of atomic and ionic emission while nanoparticles composed of Cu and Zn contribute the emission features observed very close to the target. As seen in Fig. 2, in vacuum the emission from excited Zn neutral species are more pronounced than the emission from excited Cu species although the Zn I at 481 nm (E_j = 6.65 eV) originated from a significantly higher excited energy level compared to Cu I at 510 nm (E_j = 3.81 eV). Plume velocity was estimated by measuring the plume front displacement as a function of time from the time resolved images. Plume front was defined as the plume region where emission intensity decreased to 10% of maximum intensity. Two-point displacement method was used to estimate the average velocity from 0–1000 ns duration at an interval of 50 ns. The estimated average velocities of brass plume, Zn I and Cu I from the contour plots are 1 × 10⁶cm/s, 1 × 10⁶cm/s and 0.7 × 10⁶cm/s, respectively. The slower component seen near to the target (nanoparticle plume) is found to propagate with a much slower velocity ∼2.3 × 10⁴cm/s.

Plume morphological changes were found to be insignificant when the ambient pressure increased from vacuum to 10s of mTorr levels. This is caused by low collision probability of plume species with ambient species at these pressure levels due to lower density (larger mean free path) of the ambient gas. Above these pressure levels, the plume morphology changes significantly, along with an enhancement of emission intensity due to enhanced ambient Ar collisions with plume species. For example, the space-time contour plots obtained at 100 mTorr and 500 mTorr, given in Figs. 3 and 4, showed significant enhancement in plume emission at the extended regions. Plume deceleration is also evident in the space-time plots obtained at these pressure levels, as shown by the curvature of the plume emission distribution in the contour plots, especially at later times (>500 ns). The plume shape, expansion distance, and the aspect ratio start to change due to an increase in collisions with ambient Ar (see Fig. 1). At these moderate pressure levels, the plume shape is found to be more spherical. It can also be observed that Zn appears to propagate faster than Cu, especially at earlier times. Additionally, Zn appears to be predominantly present in a narrow front region of the plume while Cu is uniformly distributed especially observed during later times of plume evolution. However, it is important to note that the emission intensity is also strongly affected by the plasma temperature, which also has a spatial and temporal dependence. Because the Zn emission line observed has a higher excited state energy than the Cu emission line, the Zn emission is expected to show a stronger dependence on the temperature field (I ∼ exp(−E₂/kT)), which may explain the observed localization of the Zn emission near the plume front. Additional measurements of the spatial- and temporal-dependence of the plasma plume temperature would be required to understand the Cu and Zn distributions fully and is planned for future study.

When the pressure increases to 1 Torr, the plume deceleration is more apparent due to the resistance from Ar. Fig. 5 shows the plume dynamics at 1 Torr Ar, and it is very apparent that plume deceleration is rapid especially at times>200 ns. At early times, the propagation of the plume is less affected by the ambient gas due to significantly higher plume pressure compared to background gas pressure. However, the plasma density decays rapidly with time³⁸ and the background gas deceleration is more rapid at later times. As before, the Zn emission was more localized to the front of the propagating plume in comparison with the Cu emission, indicating possible fractionation of Cu and Zn species in brass plasma.

Drastic changes in the plume morphology can be seen at 10 Torr where a torpedo-like shape results from a pinching of the plasma by the ambient Ar (see Fig. 1). At these pressure levels, the most intense emission appeared in a very localized region at the plume front. This peculiar effect is clearly seen as intense emission spots in the space-time contour plot shown in Fig. 6. Such plume sharpening behavior suggests that higher kinetic energy particles are emitted closer to the target surface normal. It has been shown in a recent study that the ion emission from a femtosecond laser ablation is restricted to a narrow cone angular distribution which implies that the effect of ambient gas on fs LPPs will not be uniform.³⁷ Therefore, at these pressure levels, the squeezing of the plume is more apparent from the side directions in comparison with plume propagation direction (normal to the target). Bulgakov and Bulgakova²² used an analogy of an under-expanded jet to explain the focusing and narrowing of a laser ablated plume in ambient gas. They showed that for plumes with forward-directed initial expansion, a focusing of the plume can occur. This effect is more prominent at pressure levels ∼10 Torr in which the focusing distance shifts closer to the target which could explain the pinching or focusing nature of the plume in this study. The molecular weight and the specific heat ratio of the ambient gas also play an important role in the narrowing or focusing of the beam, and for Ar with a higher molecular weight and high specific heat ratio, the focusing effects are expected to be more prominent.²² 

At higher pressures (∼>100 Torr), the plasma dynamics are mostly dominated by collisions from ambient gases, and instability in the plume morphology results in candle-like structures, as can be clearly seen in Fig. 1. Figs. 7 and 8 provide space-time contours obtained at pressure levels of 100 Torr and 760 Torr Ar, respectively. The images of the plasma plume at early times at atmospheric pressure levels were not obtained to avoid detector saturation. At these pressure levels, plume deceleration is apparent even at the earliest times of plasma evolution. Previous studies employing ns lasers showed instabilities in the plume-ambient interface at moderate to high pressure levels.^39,40 According to Braranov et al.³⁹ at these pressure levels, the expanding plume fronts tend to become unstable due to Rayleigh-Taylor instability and the interface between the plume and ambient gas becomes perturbed. Such instabilities in the ns LA plumes were manifested by intense intensity spots at the plume-ambient interface. However, though we have seen candle like structures at ∼100 Torr pressure levels, it is apparent that such structure appeared in a more forward centric manner and the intensity spot appeared approximately near normal to the target which could be caused by forward centric ion emission from the fs laser ablation plumes. At these higher pressures, expansion features of Zn and Cu are similar which is mostly due to plasma confinement and the collision-dominated nature of the plume.

At lower pressures, the Ar ambient gas interacts mainly in the outer boundary of the plume while the core of the plume is intact, but at higher pressures, the plume interacts with ambient media even at the earliest times, indicated by strong shockwave formation.³² At these pressure levels, the plume is found to be decelerated even at very earlier times of its evolution which leads to collisional excitation and confinement and eventually the plume length is limited to a few mm. By comparing spectrally integrated images with spectrally filtered images, it can be inferred that spectrally integrated images follow the features and trends of emission from Cu I and Zn I species. At atmospheric pressure levels, the oscillatory behavior at the plume interface noticed at 10–100 Torr pressure levels is found to be absent. Recent CFD modeling results also confirmed that the oscillatory internal shock waves routinely seen in ns laser ablation plumes were absent in the case of fs LA plumes at atmospheric pressure levels.³² 

ICCD images obtained at different pressure levels showed significant changes in the plume's morphology and hence the aspect ratios (Fig. 1). Fig. 9 shows the aspect ratio of the plume (plume length/ plume width) for different pressures. Ambient pressure plays an important role in plume propagation as well as in the aspect ratio. There is a rapid increase in the aspect ratio of the plume at early times which shows that plume expansion is preferentially biased forward at early times. At later times, as the plume's kinetic energy decreases, the plume stabilizes and reaches a steady aspect ratio. In vacuum, the expansion is preferentially in a forward direction, while at moderate pressure levels (0.1 and 1 Torr), the plume initially starts expanding in a forward direction. Then, at later times, due to the presence of a higher pressure ambient gas, forward motion is decelerated and eventually expands sideways, too, resulting in what is close to spherical expansion as seen in Fig. 1. Similar aspect ratio features have been observed using KrF excimer laser ablation plumes.⁴⁵ At ∼10 Torr, the plume aspect ratio is always higher than 1 and retains its asymmetric expansion for longer delays. This asymmetric expansion can be explained due to a non-uniform pressure effect on the plume. As mentioned earlier, angular distribution of ions in fs LPPs is very narrow, and so at higher pressures, a squeezing effect is experienced by the plume from sideways as compared to the axial direction leading to an increase in aspect ratio. At atmospheric pressure, the plasma plume attains a spherical aspect ratio due to confinement as a result of the ambient pressure from all the directions.

In order to get a better understanding of plasma expansion and evolution of Cu and Zn species at varying ambient pressure levels, position-time (R-t) plots were obtained from the spectrally resolved ICCD images (Zn I and Cu I), and the results are given in Fig. 10. The symbols in Fig. 10 represent experimental data points and the curves represent fits using different expansion models. We used classical point blast wave and drag models for comparing our experimental results. The position front with time in a point blast model varies as⁴¹ 

where ε is a constant dependent on specific heat capacity ratio, E is the energy released during explosion, ρ is the background gas density, and t is the delay time following ignition. The parameter n depends on the shape of the propagating shock wave; n = 3, 2, or 1 for spherical, cylindrical, or plane shock waves, respectively. The plume expansion is also described by a drag model, which is given by

where R₀ is the stopping distance of the plume and β is the slowing coefficient (R₀β = v₀). The drag model predicts that the plume will eventually come to a stop due to resistance from collisions with the background gas.

In the present study, the plume front position was estimated based on a 90% reduction in the maximum intensity. The R-t plot clearly shows that ambient Ar slows down the plume species' expansion with more confinement observed at higher pressures. At 100 Torr Ar, the plume is extremely confined with a stopping distance of ∼3.5 mm. Comparing the plume front boundary with the spectrally filtered time-resolved images, it appears that Zn species move faster than Cu, especially at lower pressure levels. This could be attributed to differences in physical properties of Zn as compared to Cu and similar observations were made previously with ns LA.^42,43 The species expansion trends were the same for both Zn and Cu at various pressures. Since Zn always moves faster than Cu and is present in the plume front, Zn data was used as representative data for brass plume expansion for fitting drag or blast models. For different pressure levels given in Fig. 10, the R-t plots are showing good agreement with point blast model at lower pressures (0.1 and 1 Torr) while the drag model is a good fit for higher pressures (greater than 10 Torr).

We estimated the plume stopping distance from the fast gated images and compared with the adiabatic expansion and the Predtechensky and Mayorov (PM) models. According to the adiabatic expansion model, the ablated species push against the background gas and the plume expansion stops when the ambient gas pressure and plasma pressure equilibrate. The stopping distance, L, is then given by

where γ is the specific heat ratio for Ar (1.67), E is the energy of laser pulse, P is the ambient pressure, V is the initial plasma volume, C is the geometric factor depending on the shape of laser spot on the target which is given by C = (1 + 1/tanθ). [(3 tan θ)/(π + 2π tan θ)]^1/3, and θ is the expansion angle of the plume. The estimated plume expansion angle was ∼22° from the ICCD images in vacuum. The PM model⁴⁴ provides another simplistic gas-dynamical model for describing various processes in plume expansion in the presence of an ambient gas. In the PM model, all the plume material is located in a thin expanding shell which sweeps up the background gas on a thin shell present on the outer surface of the plume front. Both the shells expand against background pressure, and the increasing mass of the swept up gas and plume mass leads to a deceleration of the plume front. The momentum equation for a thin hemispherical gas layer at distance r(t) is given by⁴⁴ 

where m_p is the plume mass, ρ_g is ambient gas density, u is the radial velocity of plume front, and p_g is gas pressure. This equation can be analytically solved to estimate the position of the plume front as well as the stopping distance. By applying the energy balance and momentum equation as described by Amoruso et al.,⁹ the stopping distance can be estimated, and the results obtained are shown in Fig. 11 along with the measured values. The stopping distance was calculated from the ICCD images at later times in plasma (∼5 μs; images not shown) where the plume front stopped propagation. Plume mass was estimated by calculating ablated mass as¹³ 

where A is the absorption coefficient of material, F is the fluence of laser pulse, n_a is the atomic density of material, ε_b is the binding energy, ε_esc is the work function,S_foc is the focal spot area, and M is the atomic mass number. By assuming properties of copper, the confined plume mass was estimated to be ∼0.8 × 10⁻¹⁰kg which scales with estimated values for the fs LPP of iron as reported by Amoruso et al.⁹ The experimental stopping distance values of the plume obtained from the fast gated imaging were compared with adiabatic, PM models and drag models and results are given in Fig. 11. As can be seen in Figure 11, the PM model matches reasonably well with the experimental values especially at moderate pressure ranges and deviation from the experimental values are apparent at pressures < 500 mTorr and at higher pressures. At lower pressure regimes, (<500 mTorr), the plume is less collisional with the ambient gas and the energy balance is not valid. Similarly, the adiabatic expansion model agrees well at lower pressure and a deviation from the experimental values is more apparent at higher pressure levels. Such discrepancy has also been noticed for ns LA plumes previously.²⁵ Stopping distance estimated using the drag model (shown in Figure 10) also matches very well with measured values from moderate to high pressure ranges.

Expansion dynamics and plume structures of an ultrafast laser ablated brass plasma in an Ar ambient gas at different pressures ranging from 10⁻⁵Torr to atmospheric pressure were systematically studied using spectrally integrated and monochromatic images of various species in the plume. Fast-gated brass images showed that the pressure of the ambient gas plays an important role in fs plume dynamics as well as elemental emission which can influence various fs LPP based applications including LIBS and LA-ICP-MS. The plume showed free and adiabatic expansion with a forward-centric bias in vacuum. At 10 s of mTorr pressure levels, the ambient gas started having an effect on the plume hydrodynamics which was more dominant at higher pressures. ICCD images showed that at pressures >0.1 Torr, collisions with the ambient gas at the plume interface lead to enhancement in emission intensity as well as plume deceleration. The plume morphology tends to be spherical in the pressure range of 0.1–1 Torr due to plasma confinement in the lateral direction as a result of collisional effects from the ambient Ar. At even higher pressures (∼10 Torr), the plume confinement occurs in both the radial as well as in the lateral plume expansion directions resulting in a torpedo-like plume morphology. Plume intensity was also observed to increase significantly in this intermediate pressure range level. At further higher pressures (∼>100 Torr), ambient gas penetrates the plume leading to the formation of a candle-like plume. The plume was more confined at atmospheric pressure resulting in plume dimensions of ∼1 mm. A secondary emission very close to target was observed and the velocity of the slower component was estimated to be ∼20 times lower than the faster moving component. Emission from the slower moving component was shown to depend on ambient gas pressure; at higher pressures above 1 Torr, the emission from the slower moving component ceases to exist and further studies are needed to understand the mechanisms and processes involved.

Spectrally filtered images and R-t plots showed that Zn species always traveled faster than Cu species and the effect was more prominent at lower pressures. Zn was also observed to be present predominantly in the plume front region. This could be attributed to differences in physical properties of Zn as compared to Cu, however, further studies are needed. Plume expansion was modeled using point blast and drag models and R-t plots showed good agreement with point blast model at lower pressures studied while the drag model of plume expansion showed good agreement at moderate to higher pressures. The plume stopping distance was also modeled using the adiabatic model and the Predtechensky and Mayorov (P-M) model, and the results showed that the adiabatic model predicted the stopping distance well in the low pressure levels while the P-M showed good agreement at moderate pressure range. Ambient pressure was shown to have a significant effect on the plume aspect ratio. In general, the fs LPP plume was observed to have a forward bias at earlier times, which was stabilized in time to a constant aspect ratio. In vacuum, the plume retains a higher aspect ratio at later times, but for higher pressures the plume tends to attain a spherical shape due to plasma confinement.

This work was partially supported by the US DOE, Office of National Nuclear Security Administration (NNSA Award No. DE-NA0001174) and US National Science Foundation. Pacific Northwest National Laboratory is operated for the US Department of Energy by the Battelle Memorial Institute under Contract No. DE-AC05-76RLO1830.