Plume Composition and Evolution in Multicomponent Ices Using

Jul 2, 2014 - neutral plume is ejected by an IR laser pulse and ionized by a UV laser pulse for ... studied, no group has yet studied the plume evolut...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/JPCA

Plume Composition and Evolution in Multicomponent Ices Using Resonant Two-Step Laser Ablation and Ionization Mass Spectrometry Bryana L. Henderson and Murthy S. Gudipati* Science Division, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California 91109, United States S Supporting Information *

ABSTRACT: The composition and evolution of plumes generated in a resonant infrared (IR) laser desorption of low-temperature ices is investigated via a recently developed twostep laser desorption and ionization mass spectrometry (2S-LAIMS) technique where a neutral plume is ejected by an IR laser pulse and ionized by a UV laser pulse for analysis via time-of-flight mass spectrometry. By varying the delay between the lasers, we can construct a complete time-resolved model of the ejected plume components. We found that water ices containing mixtures of polar and nonpolar analytes displayed complex mass spectral profiles that varied as the plume evolved. In these samples, the low-volatility polar analytes and clusters were restricted to the early part of the plume, whereas volatile or nonpolar analytes were spread throughout the plume. The distributions of low-volatility polar species, clusters, and impurities from the copper substrate were well-represented by single Maxwell-like distributions centered at high velocities (600−800 m s−1), while nonpolar, volatile species contained two distinct components, indicating both ablation and thermal desorption processes. Characterization of plume distributions can therefore provide new insight into an analyte’s chemical identity and can aid in assignment of otherwise ambiguous signals in the mass spectra.

1. INTRODUCTION Combination laser ablation and mass spectrometry techniques (i.e., matrix-assisted laser desorption and ionization (MALDI)) have proven indispensable for precise analysis of fragile condensed-phase samples containing large molecules.1 The laser ablation methods involved in these techniques have been developed and extensively studied theoretically and experimentally by several groups2−12 and typically use a single highenergy laser pulse to achieve both desorption and ionization of molecules from solid or liquid matrices. Laser pulse width, pulse energy (fluence impinging on the target), and wavelength are the key parameters that can be varied to meet the analytical requirements for a given sample using these methods. However, due to the fact that both neutrals and charged species are generated and ejected from the target at the same time and that the target is biased at high positive (or negative voltage), intraplume collisions become significant in single-laser ablation/ionization processes, and they contribute to collisional reactions within the plume. Here, we investigate plume formation and evolution in a recent two-step method where ablation and ionization are initiated by separate lasers. This method, called two-step laser ablation and ionization mass spectrometry (2S-LAIMS), was developed in our lab13,14 following techniques already reported in the literature by Focsa and others.15−17 In resonant 2SLAIMS, the majority of the analytes and the matrix medium are desorbed from the substrate as neutrals with an infrared laser through resonant vibrational excitation of the matrix material, and then are subjected to resonance-enhanced multiphoton © XXXX American Chemical Society

ionization by a secondary ultraviolet laser. By separating the ablation and ionization processes and optimizing them independently, we: (1) minimize the amount of energy directed at the sample, which reduces the possibility of analyte damage or chemistry during the ablation process itself (especially important when studying reaction mechanisms or unstable molecules), and (2) control ionization timing, which yields additional information on the chemical composition of the solid under investigation. With resonant 2S-LAIMS, we can track the time-dependence of analytes, matrix molecules, and other impurities in ices as they are ejected. Our earlier work13,14 demonstrated that the 2S-LAIMS does not cause measurable damage to the solute species in cryogenic ices, confirming the first advantage mentioned above. In this Article, we demonstrate the second advantage, that additional information on the chemical composition can be obtained by separating ablation and ionization steps. To investigate how these molecular properties change the appearance of the mass spectra in different parts of the plume, we undertook a time-resolved resonant 2S-LAIMS study of irradiated toluene doped water ices, which contain polar and nonpolar species of all sizes and volatilities.13 While the effect of laser energy on bulk ejection of molecules has been extensively studied, no group has yet studied the plume evolution of ice mixtures containing a mixture of polar and nonpolar analytes. Received: March 28, 2014 Revised: June 26, 2014

A

dx.doi.org/10.1021/jp503111k | J. Phys. Chem. A XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry A

Article

Laser wavelength dependence on resonant ablation efficiency of PAH (polycyclic aromatic hydrocarbon)18 as well as inorganic compounds19 mixed in water ice films has been previously demonstrated. We examine ablation and desorption trends in detail and find that a thorough understanding of plume evolution over time is essential for accurate interpretation of ice composition and optimization of mass spectra. The goal of our research is to develop resonant laser ablation and laser ionization methods that are suitable for in situ study of reaction pathways in condensed media, particularly cryogenic solar system and interstellar ice analogues, causing the least possible damage to the analytes while opening another dimension in analyte separation and characterization methods.

2. EXPERIMENTAL METHODS All resonant 2S-LAIMS experiments described here were conducted inside a custom-made stainless steel time-of-flight mass spectrometer (TOF-MS) chamber (Jordan TOF Systems) connected to an ice chamber with rotatable cryostat. Toluene (CHROMASOLV, Sigma-Aldrich) and water (ULTREX II Ultrapure, J. T. Baker) were added to glass flasks in desired ratios (typically 5% toluene and 95% water by volume), and four freeze/pump/thaw cycles were completed with liquid nitrogen at 77 K to fully remove O2 and N2 from each sample. The flasks were then attached to the vacuum chamber, and the vapor was directed through stainless steel tubing toward one face of a conduction-cooled 5 K copper block mounted on the closed-cycle helium cryostat head, located inside the stainless steel chamber, maintaining a background pressure of 1 × 10−6 mbar during deposition. The ice (considered to be amorphous solid water at these temperatures) was allowed to accumulate for 1 h for a final thickness of ∼5 μm, and then the copper block was rotated for analysis so that the icy surface is oriented toward the time-of-flight tube. All lasers and the TOF-MS were in the same horizontal experimental plane, while the cryostat was in the vertical plane. Background pressure during the ablation studies was typically in low 10−9 mbar. For analysis, an infrared laser (Vibrant IR OPO, Opotek, Inc.) was tuned to the absorption of water-ice (λ = 2948 nm; ∼5 ns pulse width, 10 Hz repetition rate) and focused through a quartz lens (150 mm focal length; ∼280 μm 1/e2 beam width) onto the surface of the ice inside the chamber. As the IR pulse strikes the ice surface, it ejects a plume of material containing both matrix and analyte species through resonant excitation of the water molecules in the ice matrix. In our experiments, the wavelength is adjusted to match the peak absorption of water; if needed, this can be adjusted to other wavelengths for desorption of other species.18 After firing the IR pulse, a UV laser (Quantel Brio Nd:YAG, Quantel Laser; λ = 266 or 355 nm depending on the type of multiphoton ionization desired; ∼5 ns pulse, 10 Hz) was fired with controlled variable delay. This UV pulse, which was focused to a ∼240 μm beam waist approximately 3.5 mm from the grounded substrate and about the same distance from the ion-extraction optics of the TOF-MS, intersects and ionizes a cross-sectional slice of the plume. Using ion extraction optics, the positive ions generated are immediately subjected to acceleration toward the TOF-MS detector, focused using X,Ydeflectors and an Einzel lens, and passed through a 100 cm flight tube. Inside the flight tube is a sheet metal liner floating at a voltage of −2.5 keV (variable up to −4 keV) to provide a uniform electric field for the accelerated and focused ions in the flight tube. Figure 1 shows a photograph and a sketch of the 2S-

Figure 1. Experimental setup. Top: A photograph of the instrument including IR and UV laser paths. Bottom: Instrument layout. Further details are given in the text.

LAIMS instrumentation. A Stanford Research Systems DG645 digital pulse generator was used to trigger the flashlamp and Qswitch for both IR and UV lasers, and a photodiode was used to trigger data collection by a 500 MHz oscilloscope (TDS5054BNV, Tektronix, Inc.). The ice surface was raster-scanned by changing IR lens position with a digital three-dimensional stage (ThorLabs, Inc., position-controllable to 0.1 μm, scanned at a velocity of 10 μm s−1 or 1 μm per pulse), and 100 pulses were typically averaged to produce a single spectrum. The digital stage, oscilloscope, and pulse generator were controlled via homemade LabVIEW software, enabling precise instrument coordination and data collection. Depending on the laser fluence, the initial laser impact ejects material in one or both of two stages: (1) a rapid volume ejection/ablation and/or (2) a slow thermal desorption. If the energy imparted into a material is sufficiently high, an analyte can be ejected in a dual-component distribution.15,16 Rapid heating of a small volume of sample by an infrared laser pulse causes a buildup in pressure, which if confined to a small volume can lead to a violent mass ejection of material. Any discussion of ablation (transient) versus thermal desorption (equilibrated) processes must also consider the effects of stress confinement and thermal confinement during the laser pulse impact.20 We have previously determined that our IR laser pulses, with energies and pulse widths similar to those used here, lead to thermal confinement of the plume.14 On the other hand, the parameters for stress confinement are not as clear as for the thermal confinement due to several unknown parameters as discussed below. Stress confinement conditions are met when the laser pulse is shorter or comparable to the B

dx.doi.org/10.1021/jp503111k | J. Phys. Chem. A XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry A

Article

time it takes for the built-up pressure to diffuse, Lp/cice, where cice is the speed of sound in water ice. While the precise speed of sound in porous, low-pressure direct-deposition amorphous water ice at 5 K is yet unknown, an estimate of ∼2 km s−1 transverse velocity observed in 70−180 K high-pressure amorphous ices21 gives that Lp/cice = ∼3 × 10−10 s (0.3 ns). However, direct vapor deposition leads to low-pressure lowdensity amorphous ice under our experimental conditions. If the speed of sound in this amorphous ice were to be significantly lower (∼few hundred meters per second), then our laser ablation (pulse width ∼5 ns) could be a borderline case for stress confinement. In either case, relatively low laser energies can still produce an ablative phase explosion of materials.

spectrum. The expected toluene isotope mass ratios were seen clearly from 89 to 93 m/z, and the smaller fragments observed correlated well with those reported in NIST electron ionization spectra,22 although there appeared to be slightly more fragmentation. We observe a direct correlation between UV pulse-energy and fragmentation of parent molecules. With higher UV energy, we observed higher fragmentation. No peaks were observed past 93 m/z, indicating that complex chemistry is not induced by the IR desorption and UV ionization process or in the flight tube itself, in agreement with our previous observations.13,14 The resonant 2S-LAI mass spectrum of hydrogen flowdischarge VUV light (predominantly Lyα at 121.6 nm) irradiated toluene/water ice is plotted in panels b (0−400 m/ z) and c (100−700 m/z) of Figure 2. The toluene parent ion was again the most intense signal in the spectrum, but many new peaks were seen past 93 m/z. As reported previously,13 the Lyα irradiation initiates ionization and hydroxylation of toluene, forming cresol (seen at 108 m/z), which exhibits increased hydrogen bonding with nearby water molecules. Formation of hydrated mixed clusters containing water and inorganic compounds or organic species has been reported and discussed elsewhere in the literature.23−25 Similarly, in our studies, clusters of water containing cresol C7H7OH(H2O)n were seen up to n = 20. These photoprocessed toluene ices were found to contain species that varied in polarity and volatility, making them an excellent source material for investigation of desorbed plume composition and dynamics. 3.2. Plume Dynamics. To track changes in plume composition, we obtained spectra at several IR−UV time delays to generate a series of virtual cross sections of the plume. A set of four spectra from a plume formed by laser desorption of photoprocessed toluene/water ice is plotted in Figure 3 (spectra of the 0−90 m/z region are plotted in Supporting Information Figure S1). Toluene, cresol, and hydrated cresol are shown at 3.6, 5.6, 7.6, and 17.6 μs (each trace normalized to 92 m/z for clarity). The maximum intensity of toluene correlated with a delay of approximately 4.6 μs. Comparison of the traces reveals that the later parts of the plume contain a much larger proportion of toluene relative to the polar cresol and cresol−water clusters. 3.3. 2S-LAI Mass Spectra of Pure Water Ice. To better understand the plume evolution of the H2O matrix with no impurities added, pure H2O ices were evaluated independently. However, experiments using the 266 nm UV laser gave spectra that contained neither water monomers nor clusters. Although water−cresol clusters (ionization energy, IE, of cresol = 8.3 eV) are easily ionized with 1 + 1 multiphoton ionization (MPI) from 266 nm UV irradiation (4.66 eV/photon, 0.5 mJ/pulse), the ionization of H2O (IE = 12.6 eV in the gas phase26) was not easily observed in our experiments. Mihesan et al., however, have observed the ionization of water and clustering using 266 nm laser with pulse energies of the order of 30−60 mJ.15,25 This discrepancy is likely due to higher power densities, that is, 38 GW cm−2 used by Mihesan et al., which enable higher order photon ionization processes such as a 2 + 1 MPI. The higher laser fluence also results in ionization of atoms, molecules, and clusters of very low abundances in the plume, as observed by Mihesan et al. (mentioned above). In contrast, our experiments used 266 nm laser at only 0.22 GW cm−2 fluences. However, we were successful in ionization of water and clusters using a 355 nm UV laser (3.49 eV/photon, ∼2−3 mJ/pulse, ∼5−7 J cm−2 or ∼1−1.4 GW cm−2), which enabled a clear viewing of

3. RESULTS 3.1. 2S-LAIMS of Water and Toluene Ice. A resonant 2SLAI mass spectrum of 5% (v/v) toluene in water ice at 5 K, taken using a 2948 nm IR laser pulse for ablation/desorption and 266 nm UV pulse for ionization, is shown in Figure 2a (0− 150 m/z). Because toluene’s UV absorption is strong at 266 nm, resonance enhanced multiphoton ionization (REMPI) was efficient, and toluene was the most intense signal in the

Figure 2. Mass spectrum of (a) unirradiated and (b,c) VUV-irradiated 5% toluene/95% H2O ice at 5 K. A 266 nm UV laser was used for ionization. Irradiated ices contain the photoproduct cresol and a large number of cresol−water clusters (panel c). C

dx.doi.org/10.1021/jp503111k | J. Phys. Chem. A XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry A

Article

Figure 3. Close-up of toluene (92 m/z), cresol (108 m/z), and cresol−water (126 m/z) at UV delays of 3.6, 5.6, 7.6, and 17.6 μs from the thermal desorption region, each normalized relative to toluene’s 92 m/z peak. As the plume evolves, the polar cresol and cresol−water molecules diminish relative to the nonpolar toluene peaks. Full, nonnormalized spectra of these ices in the range of 3−8 μs are available in Supporting Information Figure S1.

Figure 4. Comparison of 2S-LAIMS spectra of pure water at 4.6 μs (red) and 6.6 μs (blue) IR−UV delays (2948 nm fluence = 1.6 J cm−2). Water monomers were most prevalent at long delays, whereas water clusters dominated at shorter delays, appearing early in the plume.

the H2O signal and allowed for examination of plume dynamics of the water matrix material itself. The pure water ice resonant 2S-LAI mass spectra (2948 nm, 1.6 J cm−2 fluence) are shown in Figure 4. Water (H2O+) and its dissociation products (OH+ and O+) were easily seen in all spectra, and the signal at 18 m/z saturated the detector at most settings. Protonated water clusters ((H3O+)(H2O)1−12) were observed to a lesser extent. Figure 4 highlights the difference in plume position between H2O monomers and water clusters. At an infrared laser fluence of 1.6 J cm−2, the water monomers remained abundant at later time delays (∼6.6 μs), whereas the large water clusters were more abundant at shorter delays (∼4.6 μs). We have compiled detailed time- and fluence-dependendent spectra of water and its large clusters (up to n = 15). In agreement with the observations of Focsa et al.,16 we also find that water clusters have a smaller detection window than the H2O monomer in the plume. Although the IR wavelengths are slightly different (3100 nm vs 2948 nm in our setup), the IR laser fluences used in our studies and in the laboratory of Focsa and co-workers are similar (∼2−3 mJ/pulse, 5 ns pulse width), suggesting that the ice plumes generated in both of these laboratories should have similar composition and physical characteristics. Much higher UV fluences used by Focsa et al. enabled them to detect larger water clusters (beyond n = 100). In addition to lower UV laser fluences used in our experiments (as discussed above), the distance between the ice surface and ionization (UV) laser focus is an order of magnitude shorter at ∼3−4 mm in our experimental setup as compared with Focsa’s instrumentation at ∼30−35 mm. The fast component of the plume appears at approximately 5 μs in our experiments versus about 50 μs with the longer distance traveled. The long plume flight could

potentially allow for large clusters with excess internal energy to rearrange and fragment into stable medium-sized configurations prior to ionization. Conversely, a short ionization distance (as employed in our experiments) would ionize clusters before they are able to reach stability, leading to an exponential decrease in higher cluster abundance. As large clusters travel toward the ionization laser, fragmentation and rearrangement causes them to lose kinetic energy, which may explain why Mihesan et al.15,25 observed relatively larger (30−40 water molecules) clusters at longer time-delays. This comparison clearly demonstrates that, although the ablation conditions are similar, the fluences of the ionization laser and its placement from the plume source make a significant difference in the detection of clusters. 3.4. Effect of Incident IR Desorption Fluence on Cluster Dynamics. To test the effect of laser fluence on desorption in these samples, the IR laser fluence was varied (1.6, 1.2, and 0.6 J cm−2; see Supporting Information Figure S2), while all other parameters were held constant. H2O monomer signals were found to peak at UV−IR delays of 7−8 μs in each case, but the water cluster signals were found to be dependent on incident IR fluence. This will be discussed in more detail in section 4.6.

4. DISCUSSION 4.1. Toluene/Water Ice Plume Dynamics. To investigate molecular transport in the plume, a variety of fragment signals in the mass spectra from the Lyα irradiated toluene/water ice were selected for analysis of plume evolution. Toluene, cresol, D

dx.doi.org/10.1021/jp503111k | J. Phys. Chem. A XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry A

Article

molecular dynamics simulations by Zhigilei et al.30 When applied to our results, this provides further evidence for segregation of polar and nonpolar species in the plume based on molecular properties. Nonpolar species can desorb over a wide range in the plume, while polar species (which are hydrogen bonded with the water matrix) are restricted to the high-fluence laser ablation regime (during the phase-explosion time scale). Observation15,25 of similar behavior with formaldehyde in water ice indicates that thermal desorption applies to any volatile species that can easily be evaporated with transient heating process of the matrix. 4.2. Phase Explosion versus Thermal Desorption. As seen in Figure 5b, each profile contained a fast component that arrived early, at Δt = ∼5 μs. While the fast components of toluene, cresol, and hydrated cresol were comparable in terms of maximum delay time and shape of distribution, toluene had an additional secondary component found near t = 15 μs. The components were also plotted in terms of velocity (see Figure 5c) for additional insight. The absorption of a large amount of IR energy from a 10 μs for n = 2−4 clusters were too small to be reliably measured.

clusters that would result in varied velocities of the H2O monomer before reaching the UV-ionization region. This physical breakup would manifest as a delayed increase in monomer concentration following the initial fast ejection step. Correspondingly, larger clusters prone to this type of fragmentation would be skewed toward short IR−UV delays and would artificially appear to have higher velocities (which may explain the higher stream velocities seen with n = 2−4 in H+(H2O)n). Similar conclusions have been drawn by other authors studying laser desorption of clusters.33 The observed spectral variation seen here underscores the need for thorough analysis of laboratory conditions in resonant 2S-LAIMS experiments prior to data interpretation. The signal F

dx.doi.org/10.1021/jp503111k | J. Phys. Chem. A XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry A

Article

vibrational emission/relaxation. Rotational states would also work in a similar manner. Another radiative process that may slow the atoms and molecules is through dissipation of internal energy involving electronic transitions (such as fluorescence/ phosphorescence). However, electronic processes involve much higher energies, and they are always in competition with molecular dissociation processes. Thus, as compared to vibrational and rotational excitations that are only a fraction of an eV for Cu2, electronic transitions need to compete with dissociation processes. On the basis of this, we expect atomic clusters to travel faster in the plume than the molecular clusters. Another result observed in the plume profiles is that copper dimers arrived at the UV beam slightly before copper monomers. Cluster fragmentation always results in lower kinetic energy and hence lowers velocity of the monomer. The data in Figure 7 indicate that contribution of cluster dissociation to Cu peaks was significant. While studying nonclustering molecules is straightforward, multicomponent clusters can dissociate as they travel away from the substrate, complicating analysis. For example, if a neutral Cu−Cu dimer falls apart inside our plume profile region, for example, at 5 μs, it will contribute to the dimer signal intensity in the 0−5 μs region but to the monomer intensity at longer delay times. This effect artificially shifts the copper dimer (and trimer, etc.) distributions toward higher velocities, as seen in Figure 7. This effect may also justify the high apparent velocities of water clusters of n = 2, 3, and 4 as compared to water monomers. Small clusters of copper have significant binding energy as exemplified by Cu2 (181 kJ/mol36). To dissociate these clusters, close to 2 eV energy is needed. It seems clear from our studies that a significant part of Cu dimers possess such high internal energy so that they dissociate during their flight soon after ablation. In contrast to all other species observed (Cu, Cu2, water clusters, toluene, cresol, and cresol clusters), water monomers displayed a widespread velocity distribution (see Figure 6; full non-normalized profiles of all species studied can be seen in Supporting Information Figure S3). The dissociation energy of water dimers (22.8 kJ/mol binding energy found experimentally by Curtiss et al.37) is lower than that of cresol−water dimers (p-cresol−water dimers are calculated to be 32 kJ/ mol38) and is much lower than found in copper dimers (and 181 kJ/mol36). Cluster dissociation prior to ionization (possibly due to a high residual internal cluster energy from the laser impact39 or due to collisional fragmentation) is likely responsible for the wide H2O+ velocity distribution. Fournier et al.40 have hypothesized that desolvation may result at the onset of a cluster’s acceleration due to an applied electric field. Regardless of the initiating mechanism, fragmentation must be occurring prior to ionization with the UV laser under our experimental conditions; otherwise, the fragmentation of larger already-ionized clusters would noticeably affect resolution in the time-of-flight spectra. Because copper substrate impurities are undesirable and interfere with sample signals, knowledge of their distributions in the plume can aid in optimization of the spectra. By avoiding very short UV−IR delays (or by increasing the thickness of the sample layer), the contribution from substrate impurities can be minimized in the spectra. 4.6. IR Fluence Dependence. Many groups have investigated the influence of the desorption laser fluence on the velocities of ejected molecules in traditional MALDI (where the plume and its ions are produced by a single laser pulse),

impurities were occasionally found in the mass spectra. Most notably, thinner ice samples (taken from substrate surfaces that were not directly exposed to the incoming sample stream and were estimated to be ∼500 nm thick) contained significant amounts of copper and its clusters (i.e., Cu, Cu2, and Cu3) incorporated into the plume via IR pulses that penetrated all of the way to the substrate surface. Plume studies of these thinner samples revealed a similar plume dynamics pattern, with toluene remaining in the plume slightly longer than cresol, indicating that little ice was left around the ablation spot that could undergo thermal desorption and the plume is dominated by shock-wave jets. Unfortunately, the reduced water ice thickness caused an increase in copper contaminant peaks at 126−130 m/z, which overlapped with the region of interest for hydrated cresol (126 m/z) in this sample, making quantitative study of these organic impurities in thinner ice films difficult. The substrate contaminant peaks are plotted along with the toluene and cresol traces in Figure 7. Interestingly, copper

Figure 7. Evolution of plume composition ablated from thin ices as a function of IR−UV delay time, normalized for comparison. Ablation of the substrate resulting in contamination of the plume by the substrate atoms such as Cu and Cu2 is predominant in the studies involving thin ice films (