Nonequilibrium Vibrational Excitation of OH Radicals Generated

Article pubs.acs.org/JPCA

Nonequilibrium Vibrational Excitation of OH Radicals Generated During Multibubble Cavitation in Water Abdoul Aziz Ndiaye,† Rachel Pflieger,† Bertrand Siboulet,† John Molina,‡ Jean-François Dufrêche,† and Sergey I. Nikitenko*,† †

Institute for Separation Chemistry of Marcoule (ICSM), UMR 5257 − CEA-CNRS-UMII-ENSCM, BP 17171, 30207 Bagnols sur Cèze, Cedex, France ‡ UPMC-Université Paris 06, UMR 7195, PECSA, F-75005 Paris, France S Supporting Information *

ABSTRACT: The sonoluminescence (SL) spectra of OH(A2Σ+) excited state produced during the sonolysis of water sparged with argon were measured and analyzed at various ultrasonic frequencies (20, 204, 362, 609, and 1057 kHz) in order to determine the intrabubble conditions created by multibubble cavitation. The relative populations of the OH(A2Σ+) v′ = 1−4 vibrational states as well as the vibronic temperatures (Tv, Te) have been calculated after deconvolution of the SL spectra. The results of this study provide evidence for nonequilibrium plasma formation during sonolysis of water in the presence of argon. At low ultrasonic frequency (20 kHz), a weakly excited plasma with Brau vibrational distribution is formed (Te ∼ 0.7 eV and Tv ∼ 5000 K). By contrast, at highfrequency ultrasound, the plasma inside the collapsing bubbles exhibits Treanor behavior typical for strong vibrational excitation. The Te and Tv values increase with ultrasonic frequency, reaching Te ∼ 1 eV and Tv ∼ 9800 K at 1057 kHz.

■

multibubble cavitation in water.5 The spectroscopic analysis of OH(A2Σ+−X2Πi) emission lines appears to be a very useful tool to study the non-Boltzmann behavior of the plasma generated within the cavitation bubbles. However, the vibrational transitions in SL spectra are still poorly investigated and are mostly only used to identify excited species. The major difficulty in the quantification of SL spectral data is that OH(A2Σ+−X2Πi) emission yields dense overlapping vibrational structures. Rotational structures of these emission lines are not observed in SL most probably due to strong Doppler, collisional, or van der Waals broadening. This paper focuses on the development of an original approach where OH(A2Σ+− X 2 Π i ) emission lines in multibubble SL spectra are deconvoluted in order to probe the intrabubble conditions via the determination of the vibrational population distribution of the OH(A2Σ+) state for various ultrasonic frequencies.

INTRODUCTION The OH• radicals are important reaction intermediates in a large variety of advanced oxidation processes initiated by acoustic cavitation in aqueous solutions.1,2 These species are produced during the violent implosion of gas-filled microbubbles in liquids submitted to power ultrasound. In water saturated with noble gases, acoustic cavitation is accompanied not only by the generation of chemically reactive species but also by light emission, named sonoluminescence (SL).3 Spectroscopic analysis of the SL spectra helps to better understand the origin of the extreme conditions inside the cavitation bubbles. Despite numerous studies, the real nature of SL is still an open question. The multibubble SL spectra in water saturated with argon are composed of the emission lines of excited OH• radicals and a broad continuum ranging from UV to near-infrared (NIR) spectral ranges, which probably results from the superposition of several emission bands: H + OH• recombination, water molecule de-excitation, and OH(B2Σ+−A2Σ+) emission.4 Recently, Pflieger et al.5 reported SL from OH(A2Σ+) and OH(C2Σ+) excited states in water saturated with noble gases at various ultrasonic frequencies. These results clearly showed the strong effects of gas nature and ultrasound frequency on the relative intensities of OH(A2Σ+−X2Πi) (0−0) and (1−1) transitions. Moreover, the observation of OH(C2Σ+−A2Σ+) emission in the presence of Kr and Xe revealed nonthermal plasma formation during © 2012 American Chemical Society

■

EXPERIMENTAL SECTION The multifrequency ultrasonic device for SL measurements is shown in the Supporting Information (Figure 1SI). In brief, the thermostatted cylindrical reactor was mounted on top of a highfrequency transducer (25 cm2) providing power ultrasound at Received: February 29, 2012 Revised: May 3, 2012 Published: May 3, 2012 4860

dx.doi.org/10.1021/jp301989b | J. Phys. Chem. A 2012, 116, 4860−4867

The Journal of Physical Chemistry A

Article

Figure 1. Normalized SL spectra of water sparged with argon at 10−11 °C for different ultrasonic frequencies.

(3−2), and (4−3) of the OH(A−X) system can be identified using the Lifbase database.6 Figure 1 clearly demonstrates the dramatic effect of ultrasonic frequency on the relative populations of vibrational levels. In addition, for high-frequency ultrasound, OH(C2Σ+−A2Σ+) emission can be seen around 250 nm. According to the spectroscopic data reported by Felenbok,7 the emission in the spectral range of 230−260 nm is related to the (0−6), (0−7), (0−8), (0−9), (1−7), (1−8), and (1−9) vibrational transitions of the OH(C2Σ+−A2Σ+) system. Most emission bands of this transition are at λ < 230 nm and hence cannot be observed in our experiments due to the lack of sensitivity of the grating in this spectral range. As it was mentioned recently,5 the observation of OH(C2Σ+−A2Σ+) emission indicates the formation of a nonthermal plasma during multibubble cavitation in water. Our analysis of OH(A2Σ+− X2Πi) vibrational transitions described below supports this hypothesis. Determination of the Vibrational Population Distribution of the OH(A2Σ+) State. The vibrational population distribution of the excited level is needed in order to accurately estimate the thermodynamic state of the plasma generated inside the cavitation bubbles. For dense overlapping emission bands such as those observed in SL spectra, the deconvolution of molecular bands is required at the first step. We presumed a Gaussian distribution as is commonly done in the treatment of emission spectra:8

204/609 kHz or 362/1057 kHz (L3 Communications ELAC Nautik) powered by an LVG 60 RF-Generator. Ultrasonic irradiation with low-frequency ultrasound of 20 kHz was performed with a 1 cm2 titanium horn and piezoelectric transducer connected to a 750 W Vibra-Cell generator. The horn was placed reproducibly on top of the reactor opposite the high-frequency transducer using a tight Teflon ring. To collect the SL spectra generated by 20 kHz ultrasound at different distances from the ultrasonic horn, the position of the reactor relative to the detection area of the spectrometer was fixed stepwise along the z axis. Deionized water (18.2 MΩ·cm, 250 mL) was continuously sparged with Ar (100 mL·min−1, Air Liquid, 99.999%) 30 min before sonication and during the ultrasonic treatment. The temperature in the reactor during sonolysis was maintained at 10−11 °C with a Lauda RE210 cryostat and measured by means of a thermocouple inserted in the cell. The SL spectra were collected through a quartz window using parabolic Al-coated mirrors and recorded in the spectral range from 230 nm up to 450 nm using a SP 2356i Roper Scientific spectrometer (grating 300gr/mm blazed at 300 nm; focal length 300 mm; f/3.9 in aperture ratio; slit width 0.1 mm, resolution better than 1.5 nm) coupled to a charge-coupled device (CCD) camera (SPEC10-100BR with UV coating, Roper Scientific) cooled by liquid-nitrogen. Spectral calibration was performed using a Hg(Ar) pen-ray lamp (model LSP035, LOT-Oriel). The spectra acquisition was started after reaching the steady-state temperature. For each experiment, at least three 300 s spectra were averaged and corrected for background noise and for the quantum efficiencies of grating and CCD. The statistical error of emission intensity measurements did not exceed 20%.

Gi(λ) =

2 2 ai ·e−(λ − λi) /2σi σi 2π

(1)

where ai is the amplitude, σi is the bandwidth, and λi is the central wavelength of the considered vibrational band. In the framework of this numerical deconvolution approach, we first subtract a linear baseline to the measured spectrum in the wavelength interval 250−360 nm. Input parameters of the fit are the central wavelengths of the main vibrational transitions of the OH(A2Σ+−X2Πi) system: (0−0), (0−1), (1−0), (1−1), (2−1), (2−2), (3−2), and (4−3). The parameters ai and σi are incorporated in the model as fitted coefficients. The bandwidths σi are assumed to be constant. The remaining continuum part of the SL spectrum from λ = 325 nm up to λ = 400 nm is modeled as a ninth virtual Gaussian function with a free bandwidth. Note here that some wavelength data available

■

RESULTS AND DISCUSSION Identification of OH(A2Σ+−X2Πi) Vibrational Bands. Figure 1 shows the experimental SL spectra measured at 20, 204, 362, 609, and 1057 kHz normalized on the most intense OH(A2Σ+−X2Πi) (0−0) transition. The spectra at 20 kHz were taken near the ultrasonic probe where the SL intensity is highest. The shape of these spectra is quite similar to those reported recently.5 In the spectral range of 280−350 nm, the vibrational bands (0−0), (0−1), (1−0), (1−1), (2−1), (2−2), 4861



Article

Table 1. The Best Fitting Wavelengths of OH(A2Σ+−X2Πi) Vibrational Bands in Comparison with Literature Data literature λ, nm

a

fitted λ, nm

transitions

Sing9

Crosley10

Lifbase6

Dieke11

20 kHz

204 kHz

362 kHz

609 kHz

1057 kHz

(0−0) (0−1) (1−0) (1−1) (1−2) (2−1) (2−2) (3−2) (4−3)

306.38 344.77 281.89 312.41

306.4 342.8 281.1 312.2 348.4 287.5 318.5 294.5 302.2

308.90 346.00 282.79 314.53 351.00 289.13 320.00 296.87 303.00

306.4 342.8 281.1 312.2 318.5a 287.5 318.5 294.5 302.2

309.3 346.0 283.6 314.5 351.0 289.7 320.0 296.5 301.3

309.3 346.0 283.8 314.5 351.0 290.5 320.0 297.1 301.1

309.1 347.0 282.8 314.5 351.0 290.0 320.0 296.5 301.1

309.3 347.0 282.6 314.5 351.0 290.6 320.0 296.1 301.1

309.4 347.0 283.6 314.5 351.0 290.3 320.0 297.1 302.1

288.27 319.49 295.25

Probably aberrant value.

Figure 2. Deconvoluted SL spectra of the OH(A2Σ+−X2Πi) system. Dotted line: experimental spectrum; solid line: fit. The Gaussian centered at 336 nm is introduced to take into account the small influence of N2 [C3Πu (v′ = 0) → B3Πg(v″ = 0)] emission band, which can be seen in SL spectra at high-frequency ultrasound due to the presence of traces of air. Inset: Lifbase simulation of the OH(A2Σ+−X2Πi) thermalized system at T = 5000 K and P = 200 bar.

in the literature6,9−11 and summarized in Table 1 show big disparity and are far from being satisfactory to correctly fit the experimental SL data. For example, the λ values for the (0−1) transition vary by up to 3.2 nm for different reports. It should be emphasized that the stability of our numerical model depends greatly on the judicious choice of the transition wavelengths. After several tests, the best fitting wavelengths for each ultrasonic frequency are chosen with a 1.5 nm maximum tolerance referring to those given by the most recent Lifbase database6 except for the (4−3) transition (Table 1). The best fit for the latter transition is obtained at λ = 301.1−301.3 nm,

which is closer to the Crosley10 and Dieke11 data than to that reported by Lifbase.6 Figure 2 demonstrates the SL spectra with subtracted baselines and their best fits calculated with Wolfram Mathematica 7 software using the above-described approach. The choice of optimized wavelengths for the vibrational transitions allowed us to get a best fit in a wide range of the SL spectrum for all studied ultrasonic frequencies. For comparison, Figure 2SI in the Supporting Information demonstrates some unsuccessful fits obtained with nonoptimal λ values. 4862



Article

Figure 3. Relative vibrational population distribution of the OH(A2Σ+) state as a function of vibrational energy for different ultrasonic frequencies.

relative population distribution obtained from the SL spectra deviates strongly from the equilibrium Boltzmann distribution. At 20 kHz, the vibrational population distribution of OH(A2Σ+) state appears to follow a hyperbolic function known as Brau distribution typical for weak vibrational excitation.15 At higher ultrasonic frequencies it follows a Treanor exponentially parabolic distribution function which describes strong vibrational excitation.15 The overpopulation of higher vibrational states in a nonequilibrium plasma with Treanor behavior occurs through an anharmonic vibration−vibration (V−V) exchange. Recently, the Treanor effect has been observed for CO disproportionation in water treated with 20 kHz ultrasound17 in full agreement with the results obtained in this work. Electronic (Te) and Vibrational (Tv) Temperatures Calculation. The finding of nonequilibrium plasma formation during multibubble cavitation raises the question of the temperature within the collapsing bubbles. Obviously, the average gas temperature around 5000 K usually reported12,13 cannot be applied to a plasma far from equilibrium. Such kind of plasma is characterized by multiple different temperatures related to different plasma particles and different degrees of freedom. The electron temperature (Te) often significantly exceeds the vibrational (Tv), rotational (Tr), and translational (T0) temperatures: Te > Tv > Tr ≈ T0.15 The spectral analysis of the OH(A2Σ+−X2Πi) emission bands presented in this paper can be used to determine the Te and Tv of the ultrasonically generated plasma. The calculation code developed here to obtain the synthetic spectrum of the OH(A2Σ+−X2Πi) system is based on several assumptions. First, we assume that the excitation of the electronic upper level is mainly due to electron impact and that the vibrational populations of the upper level result from the ground state population distribution multiplied by the appropriate Franck−Condon factors. Second, we consider that the rotational population of the upper level is significantly the same as that of the ground state. Then, we presume local equilibrium between vibrational and rotational populations. Finally, the vibrational transitions are considered as the only processes of radiative decay. Their probability coefficients (Av′v″) are listed above. These assumptions that somehow simplify the considered plasma model are widely used in calculations of plasma states.15 The electronic, rotational, and vibrational temperatures are the input parameters for the calculation of the population distribution functions, and have a great influence on their relative intensities. For the vibrational level calculations of the ground state X2Πi, Klein−Dunham coefficients proposed by Luque and Crosley16 and by Herzberg as quoted in the NIST physical database18 are taken (see Table 1SI in the Supporting

The inset in Figure 2a shows a Lifbase-simulated thermalized spectrum of OH(A2Σ+−X2Πi) system at T = 5000 K and P = 200 bar which are usually referred to as intrabubble conditions presuming quasi-adiabatic collapse during multibubble cavitation in aqueous solutions.12,13 This Lifbase-simulated spectrum is dominated by (0−0), (1−0) and (1−1) transitions in full agreement with flame spectroscopic data.14 By contrast, the SL spectrum measured at 20 kHz exhibits also strong emission from higher vibrational excited states, such as (2−1) and (3−2) transitions which can be seen clearly around 289.7 and 296.5 nm respectively (Figure 2a). The vibrational overpopulation is even stronger at high ultrasonic frequency as follows from Figure 2b-e. For instance, at 1057 kHz the intensity of (1−1) transition is comparable to that of (0−0) transition (Figure 2e). These data clearly show that the multibubble SL spectra cannot be understood in the frame of a simple adiabatic heating model whatever the ultrasonic frequency. The deconvoluted spectra shown in Figure 2 offer the possibility to obtain the populations relative to N0 of the OH(A2Σ+) vibrational states from v′ = 1 up to v′ = 4. These values can be calculated as the ratio of the population densities of the emitting vibrational level ν′, Nν′, and of the level ν′ = 0, N0, using the well-known equation for an optically thin plasma:15 Nv ′ =

Iv ′ v ″ A v ′ v ″ × ΔEv ′ v ″

(2)

where Iν′ν″ is the intensity of the deconvoluted molecular band, Aν′ν″ is the corresponding radiative transition probability, and ΔEν′ν″ is the energy difference between the two levels involved in the optical transition. The N0 value is calculated from the normalized intensity of the (0−0) transition. The Aν′ν″ values (A00 = 1.451 × 106 s−1; A10 = 4.643 × 105 s−1; A21 = 6.852 × 105 s−1; A32 = 6.928 × 105 s−1; A43 = 5.495 × 105 s−1) are taken from the Lifbase database.6 Vibrational energies (cm−1) are computed using the Klein−Dunham coefficients determined by Luque and Crosley.16 The maximum uncertainty on relative vibrational populations of OH(A2Σ+) are estimated in the range of 15−20%. Note here that the relative populations Nr1 and Nr2 cannot be calculated properly from the (1−1) and (2−2) transitions despite the high intensities of these bands due to the large bandwidth of the virtual Gaussian which induces a high uncertainty on their intensities. The calculated values of Nrv are plotted in Figure 3 as a function of the vibrational energy for each studied ultrasonic frequency. For comparison, data computed with Lifbase for the equilibrium distribution (P = 200 bar and T = 5000 K) are also included. Figure 3 shows unequivocally that, in contrast to a thermalized system, the 4863



Article

Figure 4. Simulated and experimental deconvoluted SL spectrum of the OH(A2Σ+−X2Πi) system in the 275−360 nm wavelength range for 20 kHz ultrasound frequency. The relative intensity of the theoretical vibrational band (0−0) is normalized to its experimental value. The virtual band related to the continuum emission and the low-intensity (0−1) transition are not considered in the simulated spectrum.

transitions are computed with respect to the selection rules. Results for the main branches are presented in Table 6SI of the Supporting Information. The resulting spectra were then convoluted with the experimental slit function assumed to be a Gaussian distribution.

Information). To evaluate the rotational energy levels of this fundamental state, we use the nonequilibrium spectroscopic parameters for both rotational and vibrational distortion constants performed recently by Bernath and Colin (see Table 2SI).19 Even if in most cases the rotational constants show little variations from paper to paper, the choice of the set of Bernath and Colin19 data for the ground state X2Πi is motivated by the fact that their values often agree in digits with the other sources and contain a significant number in the decimal place. Appropriate analytical approximation taking into account the Λ doubling effects and the fine structure in the estimation of the rotational energy of the upper levels X2Πi is used. Although the nonequilibrium rotational constants of the excited state A2Σ+ have been the subject of many studies, the selection of these molecular data is sometimes not obvious. In our synthetic spectrum calculations, the values proposed by Moore and Richards20 are used, which appear to more accurately yield the observed line positions (see Table 3SI in the Supporting Information), while for the vibrational equilibrium constants, the Lifbase database is used (see Table 4SI in the Supporting Information). The rotational energy levels of the A2Σ+ doublet state are computed from the matrix elements for the 2Σ+ effective Hamiltonian given by Douay et al.21 using the system of equations (eq 1SI) presented in the Supporting Information. The simulation of the Σ state is quite simple since it is wellknown that Σ states do not allow Λ splitting (Λ = 0, S ≠ 0) and belong systematically to Hund's case (b). Each rotational level is simply split by the spin degeneracy. The vibrational term values (Tv′v″) are calculated using traditional spectroscopic developments based on the Dunham expansion series with accurate Klein−Dunham coefficients. The Franck−Condon factors (qv′v″) evaluated by the Rydberg−Klein−Rees (RKR) potential method are all taken from Luque and Crosley16 (see Table 5SI in the Supporting Information). Line positions are calculated from the energy differences between the levels of the transitions (eqs 1SI−2SI in the Supporting Information). The spectral band intensities given by eq 3 are computed by using the appropriate Höln−London factors described in the (a/b) Hund intermediate case by Arnold et al.22 Our calculation model considers satellite branches as well as the main branches of P, Q, and R. The wavelengths of the different

Inn″′ vv ″′ JJ′″ =

SJ ′ J ″ 2J ′ + 1 64π 4cν 4 ·N0· · 3 2J ′ + 1 (2 − δ0, Λ̅ ) ·Q rot e

−hc / k(

FJ ′ Fn ′ Fv ′ ) + + Tel Tvib Trot · A

v′v″

(3)

In eq 3, SJ′J″ is the Höln−London factor, Qrot is the rotational partition function of the considered molecule, Av′v″ is the vibrational transition probability, K is the Boltzmann constant, N0 is the number of molecules at the fundamental state, δ0,Λ is the Kronecker delta, and the factor (2 − δ0,Λ) is due to λ-type doubling. The terms Fn′, Fv′, and FJ′ are the electronic, vibrational, and rotational energies, respectively. Finally, the simulated and the deconvoluted experimental spectra are compared, and Te and Tv temperatures are adjusted until a best optimized fit has been achieved. Figure 4 shows as an example the simulated spectrum at 20 kHz. The results obtained for other ultrasonic frequencies are presented in the Supporting Information (Figure 3SI). In most cases, a satisfactory agreement is observed for band intensities and positions, except for the relative intensity of the (3−2) transition band at 1057 kHz (Figures 3gSI and 6hSI in the Supporting Information). This discrepancy can be traced back to the poor resolution of the SL spectrum at this ultrasonic frequency. In general, the satisfactory simulation of the deconvoluted SL spectra confirms the correctness of the molecular data incorporated in the numerical code as well as in the matrix elements used for the 2Σ+ effective Hamiltonian implemented by Douay et al.21 Table 2 summarizes the optimized values of the electron and vibrational temperatures calculated from the simulated SL spectra. It shows that the nonequilibrium plasma generated during the sonolysis of water sparged with argon obeys the classical inequality Te > Tv. The Te value increases with the ultrasonic frequency from ∼8000 K (∼0.7 eV) at 20 kHz to ∼12 000 K (∼1 eV) at 1057 kHz. For comparison, the typical electron temperature of flames is much lower (∼0.2 eV).15 The plot of Tv versus Te (Figure 5) shows that the vibrational 4864



Article

close vicinity of the probe. In order to study the effect of the distance on the vibrational population distribution, the SL spectra obtained with 20 kHz ultrasound are collected for different axial positions with respect to the ultrasonic probe. The temperature inside the reactor is maintained at about 10− 11 °C and the absorbed acoustic power, determined by calorimetry, is 30 W. The SL emission cartography shown in Figure 6 demonstrates sharp decrease in total SL intensity with

Table 2. Calculated Plasma Temperature Parameters versus Ultrasonic Frequency in Argon- Sparged Water at 10−12 °C frequency, kHz

Te, K

20 204 362 609 1057

8000 ± 500 9500 ± 400 10000 ± 800 11000 ± 600 12000 ± 700

Tv, K 5000 7600 8450 9050 9800

± ± ± ± ±

200 300 500 200 400

Figure 5. Variation of vibrational temperature as a function of electron temperature for different ultrasonic frequencies at 10−12 °C in the presence of argon.

Figure 6. Evolution of the SL spectrum at 20 kHz, Pac = 30 W, in water saturated with argon at 10−11 °C as a function of the distance from the ultrasonic horn. The deconvoluted spectra are shown in the Supporting Information (Figure 4SI) and the relative population distributions are plotted in Figure 5SI.

temperature drastically increases in the low electron temperature region (20 kHz → 204 kHz) then reaches a smoother progress in the Te region corresponding to high ultrasound frequency. Such kind of Tv−Te dependence is typical for nonequilibrium plasmas with Te ≤ 1 eV.15,23 The data summarized in Table 2 clearly indicate that the acoustic collapse creates more drastic conditions inside the bubbles at higher ultrasonic frequencies. However, several papers reported that the total SL intensity in argon-saturated water reached a maximum between 200 kHz and 400 kHz and then progressively diminished.5,24 Such apparent discrepancy can be related to a dramatic decrease of the bubble volume with ultrasonic frequency. The resonance radius of the cavitation bubble in water is equal to 10.0 μm at 358 kHz and to only 3.3 μm at 1071 kHz.24 Consequently, the bubble volume at 1071 kHz is almost 30 times smaller than that at 358 kHz, leading to a much smaller light emitting plasma core. It is noteworthy that several other parameters can influence the total intensity of SL. For instance, the cavitation event occurs at a faster rate at higher ultrasonic frequency. Also, the density of light-emitting bubbles is most probably not the same for different frequencies. In practice, it is difficult to take into account the influence of all these parameters. Consequently, the total SL intensity is hardly applicable as a probe of intrabubble conditions. By contrast, the spectroscopic analysis based on the determination of the relative vibrational level populations of electronically excited species does not depend on total SL intensity. Some Examples of Experimental Results Interpretation. To illustrate the potentiality of the developed method, it is applied to estimate the evolution of OH(A2Σ+) vibrational level populations and of Te and Tv values with the distance from the ultrasonic probe at 20 kHz and with the bulk temperature at 204 kHz. Effect of the Distance from the Probe. Low-frequency 20 kHz ultrasound produces a very inhomogeneous cavitation zone with the highest concentration of cavitation bubbles in

the distance from the probe. However, the corresponding Nrv (Table 3) and Te, Tv (Table 5) values reveal the absence of Table 3. Relative Population Distribution of OH(A2Σ+) ν′ = 1−4 Vibrational States Normalized to the 0 State in Water Sparged with Argon at Different Distances from the Ultrasonic Horn Operated at 20 kHza log Nrv b

position z (mm)

N0

N1

N2

N3

N4

15 16 17 18 19

1 1 1 1 1

0.38 0.36 0.37 0.37 0.37

0.31 0.29 0.30 0.31 0.36

0.28 0.26 0.28 0.30 0.35

0.21 0.20 0.20 0.25 0.31

a

The uncertainty is equal to 15% for N2 and N3 and 20% for N1 and N4. bz = 15−19 indicates a stepwise decreasing distance between the horn and the focus position.

plasma parameter variation in the studied range of distances except very close to the probe where higher vibrational levels seem to be more populated. Therefore, the strong drop in total SL intensity with the distance from the probe (15 ≤ z ≤ 18) is related to the decrease of cavitation bubbles density and not to changes of the plasma state inside the bubbles. Effect of Bulk Temperature. Figure 7 shows that the total SL intensity at 204 kHz sharply decreases with increasing water temperature, which is in line with previous studies.12 By contrast, the variation of bulk temperature has no significant influence either on the relative vibrational population distribution of OH(A2Σ+) species (Table 4) or on the Te, Tv values (Table 5). Such behavior cannot be explained by a 4865



Article

the probability of collisiononal quenching is increased, leading to the observed sharp drop of total SL intensity.

■

CONCLUSIONS We have undertaken a thorough spectroscopic analysis of the emission spectra generated in water in the presence of argon under the effect of low- (20 kHz) and high- (204, 362, 609, and 1057 kHz) frequency ultrasound with the main emphasis on methodological aspects concerning the relative population computations of the vibrational levels of the OH(A2Σ+) state. In the spectral range from 280 nm up to 350 nm, the vibrational bands (0−0), (0−1), (1−0), (1−1), (2−1), (2−2), (3−2), and (4−3) of the OH(A2Σ+−X2Πi) system have been identified using the Lifbase database. The relative populations of OH(A2Σ+) vibrational states from v′ = 1 up to v′ = 4 have been calculated after deconvolution of the observed SL spectra. A dramatic effect of ultrasonic frequency is observed with increasing relative populations of the higher excited levels when the frequency is increased. To make more detailed optical diagnostics, the vibronic temperatures (Tv, Te) of the OH(A2Σ+−X2Πi) system have been estimated for all studied ultrasonic frequencies. This research provides new insights into the origin of multibubble SL. We have found new evidence for nonequilibrium plasma formation during water sonolysis in argon. At low ultrasonic frequency, a weakly excited plasma is formed with Te ∼ 0.7 eV and Tv around 5000 K. By contrast, at high-frequency ultrasound, the plasma exhibits a Treanor behavior typical for strong vibrational excitation. The Te and Tv values increase with ultrasonic frequency, reaching Te ∼ 1 eV and Tv ∼ 9800 K at 1057 kHz. The developed approach offers the possibility of an intrabubble plasma diagnostics that is independent from the total intensity of SL.

Figure 7. Evolution with bulk temperature of the SL spectra at 204 kHz, Pac = 53 ± 2 W, in water in the presence of argon. Deconvoluted spectra are presented in Figure 6SI of the Supporting Information.

Table 4. Relative Population Distribution of OH(A2Σ+) ν′ = 1−4 Vibrational States Normalized to the 0 State in Water Sparged with Argon at Different Bulk Temperatures for 204 kHz Ultrasounda log Nrv

a

bulk temperature (°C)

N0

N1

N2

N3

N4

12 15 17 20 25 30

1 1 1 1 1 1

0.53 0.44 0.45 0.40 0.51 0.54

0.47 0.53 0.53 0.53 0.54 0.55

0.55 0.59 0.60 0.55 0.60 0.56

0.66 0.77 0.67 0.69 0.71 0.67

■

Additional details as described in the text. This information is available free of charge via the Internet at http://pubs.acs.org

Table 5. Calculated Te and Tv Temperatures as a Function of (i) the Distance from the Ultrasonic Probe at 20 kHz (10− 11°C, Pac = 30 W) and (ii) the Bulk Temperature at 204 kHz (Pac = 53 W) conditions 20 kHz: z, mm 19a 17 16 204 kHz: T, °C 15 19 30 a

Te, K

Tv, K

8000 ± 500 7900 ± 500 7600 ± 500

5000 ± 200 5000 ± 200 4900 ± 200

9500 ± 400 9600 ± 400 9700 ± 500

7600 ± 300 7700 ± 400 7800 ± 400

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by French ANR program (ANR-10BLANC-0810 NEQSON).

■

z = 19 mm indicates the closer distance to the probe.

REFERENCES

(1) Mason, T. J.; Lorimer, J. P. Sonochemistry. Theory, Applications and Uses of Ultrasound in Chemistry; Halsted Press (Wiley): Chichester, U.K., 1998. (2) Adewuyi, Y. G. Environ. Sci. Technol. 2005, 39, 3409−3420. (3) Young, F. R. Sonoluminescence; CRC Press: Boca Raton, FL, 2004. (4) Seghal, C.; Sutherland, R. G.; Verral, R. E. J. Phys. Chem. 1980, 84, 388−395. (5) Pflieger, R.; Brau, H.-P.; Nikitenko, S. I. Chem.Eur. J. 2010, 16, 11801−11803. (6) Luque, J.; Crosley, D. R. LIFBASE: Database and Spectral Simulation Program (Version 1.5), 1999, SRI Int. Rep. MP 99-009 (http://www.sri.com/cem/lifbase). (7) Felenbok, P. Ann. Astrophys. 1963, 26, 393−428.

progressive decrease of gas temperature inside the bubble as predicted by the adiabatic heating model. However, it can be understood in terms of OH(A2Σ+) collisional quenching by water molecules: OH* + H 2O → OH + H 2O

ASSOCIATED CONTENT

S Supporting Information *

The highest uncertainty, obtained for 30°C, is estimated to 25%.

(4)

Reaction 4 was found to be effective in atmospheric pressure plasmas25 and in combustion processes.26 In our case, heating of the bulk causes water partial vapor pressure to increase. Consequently, more water molecules enter into the bubble, and 4866



Article

(8) Nelis, T.; Payling, R Glow Discharge Optical Emission Spectroscopy: A Practical Guide; The Royal Society of Chemistry: Cambridge, U.K., 2003. (9) Sing, M. Astrophys. Space Sci. 1987, 138, 79−86. (10) Crosley, D. R.; Lengel, R. K.; Quant., J. Spectrosc. Radiat. Transfer 1975, 15, 579−591. (11) Dieke, G. H.; Crosswhite, H. M.; Quant., J. Spectrosc. Radiat. Transfer 1962, 2, 97−199. (12) Didenko, Y. T.; McNamara, W. B., III; Suslick, K. S. J. Phys. Chem. A 1999, 103, 10783−10788. (13) Suslick, K. S.; Eddingsaas, N. C.; Flannigan, D. J.; Hopkins, S. D.; Xu, H. Ultrason. Sonochem. 2011, 18, 842−846. (14) Alkemade, C. Th. J.; Herrmann, R. Fundamentals of Analytical Flame Spectroscopy; Hilger: Bristol, U.K., 1979. (15) Fridman, A. Plasma Chemistry; Cambridge University Press: Cambridge, U.K., 2008. (16) Luque, J.; Crosley, D. J. Chem. Phys. 1998, 109, 439−445. (17) Nikitenko, S. I.; Martinez, P.; Chave, T.; Billy, I. Angew. Chem., Int. Ed. 2009, 48, 9529−9532. (18) NIST Physical Reference Data, http://www.nist.gov (accessed February 16, 2012). (19) Bernath, P. F.; Colin, R. J. Mol. Spectrosc. 2009, 257, 20−23. (20) Moore, E. A.; Richards, W. G. Phys. Scr. 1971, 3, 223−230. (21) Douay, M.; Rogers, S. A.; Bernath, P. Mol. Phys. 1988, 64, 425− 436. (22) Arnold, J. O.; Whiting, E. E.; Lyle, G. C. J. Quant. Spectrosc. Radiat. Transfer 1969, 9, 775−798. (23) Ono, S.; Teii, S. J. Phys. D: Appl. Phys. 1983, 16, 163−170. (24) Beckett, M. A.; Hua, I. J. Phys. Chem. A 2001, 105, 3796−3802. (25) Bruggeman, P.; Iza, F.; Guns, P.; Lauwers, D.; Kong, M. G.; Gonzalvo, Y. A.; Leys, C.; Schram, D. C. Plasma Sources Sci. Technol. 2010, 19, 1−7. (26) Hall, J. M.; Petersen, E. L. Int. J. Chem. Kinet. 2006, 38, 714− 724.

4867


Nonequilibrium Vibrational Excitation of OH Radicals Generated

Recommend Documents