15 September 2000

Optics Communications 183 2000. 479484 www.elsevier.comrlocateroptcom

Focusing limits of intense ultrafast laser pulses in a high pressure gas: road to new spectroscopic source
A. Talebpour ) , M. Abdel-Fattah, S.L. Chin
Centre dOptique, Photonique et Laser (COPL) and Dept. de Physique, Uniersite Laal, Quebec, Qc., Canada G1K 7P4 Received 11 May 2000; received in revised form 30 June 2000; accepted 19 July 2000

Abstract As a result of focusing ultrafast Ti:Sapphire laser pulses in a gas, a plasma column is created. Its density does not exceed some limiting value as a result of the defocusing by the electrons resulting from multiphoton ionization and the involvement of high order nonlinearities. The spectrum of the radiation from the excited species in the column has little contribution from the plasma continuum and the line broadening is less than that of the spectra radiated from a plasma generated by a long laser pulse. Thus, the spectroscopic source obtained using short laser pulses is an excellent source for spectroscopy. q 2000 Published by Elsevier Science B.V.
Keywords: Propagation; Ti:Sapphire laser pulse; Plasma; Spectroscopy; Photo-emission spectrum

1. Introduction Shortly after the invention of lasers, the phenomenon of the generation of self-luminous plasma in gases using Q-switched ruby lasers was reported almost simultaneously by different research groups w13x. During the years that followed the phenomenon was studied all over the world. These studies did not lead to great technological impacts which were expected in the early days. The main application anticipated from a laser generated plasma is in the spectroscopy for material analysis. However, laser generated plasma source for this application is associated with major difficulties which di-

) Corresponding author. Tel.: q 1-418-656-2131 ext. 4482; fax: q 1-418-656-2623; e-mail: talebpou@phy.ulaval.ca

rectly result from the mechanism of plasma generation. In this process the role of a long laser pulse is to generate some seed electrons by multiphoton ionization MPI. and to accelerate them by inverse Bremsstrahlung to collide with the gas molecules and generate new electrons through collisions. Avalanche or cascade ionization follows. Once the density of the electrons reached a critical value breakdown occurs and a high density plasma is created which has no memory of the laser pulse. On the other hand the mechanism of plasma generation in the interaction of an ultrafast laser pulse is free from the collisions due to the fact the duration of the pulse is short compared to average collision times. The electrons are exclusively created through the MPI in the leading edge of the pulse. As time passes, the increase in the energy of the pulse cannot lead to an appreciable increase in the density of the elec-

0030-4018r00r$ - see front matter q 2000 Published by Elsevier Science B.V. PII: S 0 0 3 0 - 4 0 1 8 0 0 . 0 0 9 0 3 - 2

480

A. Talebpour et al.r Optics Communications 183 (2000) 479484

trons, since the defocusing effects caused by the plasma and by the high order nonlinearities decrease the local intensity and thus stop the generation of more electrons w4x. Therefore, once the pulse passed, any point in the focal region will be left with luminous excited species immersed in a low density plasma, with very well defined boundaries. These species radiate to a particularly continuum-less spectrum, examples of which will be given in the present article and will be compared with the spectra obtained using long laser pulses. This new radiation source might have application in spectroscopy, especially when a gas with high pressure is under study. In the present article, through experimental observations on different gases, we will present and discuss the merits of such a spectroscopy using ultrafast Ti:Sapphire laser pulses. 2. Experimental results Our laser system consists of a Ti:Sapphire oscillator followed by a regenerative and two multi-pass Ti:Sapphire amplifiers that can deliver pulses with energies of up to 100 mJ. The duration of the pulses is 200 fs FWHM. and the central wavelength l is 800 nm. As long pulse source, we use an Nd:Yag
Fig. 2. A narrow portion of Fig. 1 which presents the strong line broadening in the case of long pulses. The identification of the lines are from w5x.

Fig. 1. Spectrum of Xe at a pressure of 10 Torr using long pulses 5 ns, Nd:Yag laser. with an energy of 100 mJrpulse and short pulse 200 fs, Ti:Sapphire laser. with an energy of 40 mJrpulse.

pulsed laser with a duration of 5 ns, l s 532 nm and maximum pulse energy of 800 mJ. The laser beam is focused using a 100 cm focal length lens into a chamber containing different gases with a variable pressure of 1700 Torr. The radiation emitted from the plasma is imaged onto the entrance slit of a single shot spectrometer armed with a CCD camera, with a spectral resolution of ; 3 A. When using long pulses the plasma is created in a small near spherical zone in the focal region and from one shot to another the position of breakdown varies along the propagation axis. The emission spectrum of this breakdown consists of a continuum modulated by the line emissions from atoms and atomic ions. When using the 200 fs laser pulses, spanning around the focal region, a faint filament with a length of about 10 cm is visually observed. The spatial shape of the filament is stable from shot to shot. In Fig. 1 we present the visible spectrum of Xe with a pressure of 10 Torr interacting with Ng:Yag laser pulses with a duration of 5 ns dotted line. and compare it with the case of 200 fs Ti:Sapphire laser pulses solid line.. As it is observed there are two distinct differences between the two cases. The first

A. Talebpour et al.r Optics Communications 183 (2000) 479484

481

difference is the smaller contribution of the continuum in the case of short laser pulses. In fact, we have noticed that the contribution of the continuum in the spectra of other target gases is even smaller as it will be remarked later in the case of D 2 . The second difference is the amount of broadening in the spectral lines. To show this point, in Fig. 2 we present a limited portion of Fig. 1 in the wavelength interval of 535548 nm. While the line broadening in the case of long pulses is very strong, the line widths in the case of short pulses, are limited by the resolution of the spectrometer. Therefore, the experimental results on Xe showed the merits of using short pulses as being the reduction in continuum radiation and reduced line broadening. In the following section we explain the physical origin of this superiority.

Fig. 3. The emission spectra of D 2 , with a pressure of 400 Torr, interacting with a 200 fs Ti:Sapphire laser pulse with an energy of 40 mJ.

3. Discussion As it was mentioned in the Introduction, the main reason behind the superiority of the spectra in the case of short laser pulses is due to low plasma density. To verify this statement first we determine the upper limit of plasma density in D 2 . Then we show that in the interaction of short laser pulses with a high pressure gas, the laser is not able to focus to its minimum focal size. Therefore, the peak intensity of the pulse remains so low that the density of electrons never reach the critical density necessary for the occurrence of breakdown. We will estimate the value of the reachable peak laser intensity in a gas at high pressure. Merits of choosing D 2 are the following. During the interaction of D 2 with the laser pulses there is a channel which dissociates the molecule into excited atom, DU w6x. The excited atom fluoresces and the bandwidth of the fluorescence can be used to characterize the plasma density. Since the Stark shift in hydrogen linearly depends on the external electric field, the fluorescence lines of H D. have the largest broadening. This allows the measurement of small plasma densities w7x. In Fig. 3 we present the spectrum of D 2 at a pressure of 400 Torr. There are two emission lines from DU , Da and Db . The mechanism of creating excited atoms radiating to these lines is described in Ref. w6x. Due to the linear Stark shift of the energy

levels of an atom, which occurs due to the fluctuations of plasma density, there will be a line broadening. In the case of Db line this broadening is calculated and its relation to the plasma density is found w7x. In Fig. 4 we have presented the experimental profile of the Db line open circles. along with the calculated line profiles in a plasma with a temperature of 10 4 8K at different densities. This temperature has been chosen because at low intensities of the order of 10 13 Wrcm2 the intensity of pulses at a high pressure gas. the MPI process takes place by the absorption of the threshold number of photons, just sufficient to overcome the ionization potential. Thus, the kinetic energy of the ionized electrons cannot exceed the energy of a single photon, which corresponds to a temperature of ; 2 = 10 4 8K. As it is observed from Fig. 4, the value of the density of the plasma is around 10 14 cmy3 . In the following we show that this density is many orders of magnitude smaller than what was expected if the laser were able to focus to its minimum size the situation which occurs in vacuum.. First we calculate the probability of MPI here tunnel ionization is included in the definition of MPI. of D 2 as a function of pulse energy using the experimental data on the ion yield of this gas. Since in the experiment the pressure of D 2 will be kept very low, the effects of self focusing by the neutrals and defocusing due to the plasma is negligible and

482

A. Talebpour et al.r Optics Communications 183 (2000) 479484

100 mm. We verified our absolute intensity calibration by comparing with the saturation intensity of He, which is known from theoretical calculations w8x. The yield curve of Fig. 5 can be used to calculate the probability of MPI as a function of the peak laser intensity. Once this quantity is known, a knowledge of the density of plasma will enable us to estimate the peak laser intensity reached in high pressure gas. Thus we proceed to calculate the probability of MPI from the ion yield curve of Fig. 5. We note that the ion yield, N, is related to the probability of MPI, Pi , through the following relation.
Fig. 4. Comparison of the profile of the Db line with theoretically calculated profiles at different plasma densities. The densities are in the unit of cmy3 .

N s r Pi r . dV

HV

1.

where r is the density of gas and V is that part of the interaction volume from where the ions are collected. To find Pi one should solve this integral

therefore the intensity of the laser can be calculated easily. The linearly polarized Ti:Sapphire laser was focused using fr100 optics 1 m focal length lens and beam diameter of 1 cm. into an ultrahigh vacuum chamber having a background pressure of 2 = 10y9 Torr. The gas pressure in the interaction chamber was controlled by a precision leak valve and ranged from 10y8 to 10y4 Torr. Ion species, which were created through the interaction of D 2 molecules with laser pulses, were collected using a time of flight TOF. mass spectrometer. Ion yield curves were produced by combining a series of intensity scans, each having a different fill pressure in the interaction chamber. In Fig. 5 the ion yield versus pulse energy upper scale. is presented. In this figure, the horizontal axis bottom. is in terms of the peak intensity of the laser. The laser intensity was determined by measuring the laser energy, the pulse width, and the focal spot size. The energy measurement was taken from a 4% reflection off a wedged beam splitter situated after the interaction region, and was detected with a pyroelectric energy meter. The pulse width was measured by a second order autocorrelator. To measure the spot size, the beam was attenuated by successive reflections from high-quality glass plates. The beam was then focused by the 100 cm focal length lens and the image of the focused area was analyzed by a microscopic objective associated to a CCD camera. The focal spot diameter at 1re 2 . was found to be

Fig. 5. Ion yield versus peak laser intensity and pulse energy. curves of Dq and Dq created as a result of interaction of D 2 with 2 a linearly polarized laser pulse. The theoretical yield curve of Dq, 2 calculated from the PPT model see text., has excellent fit with the experimental curve. Also the calculated probability of MPI as a function of peak laser intensity pulse energy. is shown in inset.

A. Talebpour et al.r Optics Communications 183 (2000) 479484

483

equation. However, this is impossible due to the fact that in the experiment only the relative value of the ions is measured. Thus, we chose another approach, i.e. we calculate Pi from the theory in such a way that when it is substituted in Eq. 1, the theoretically calculated curve fit the experimentally measured one. Previously w9x, we have shown that the rate of MPI of D 2 , W, is very well predicted by the PPT model w10x by assuming that the electron tunnels through a barrier given by Z e f frr y Fr instead of the pure Coulomb barrier 1rr y Fr which is used in the calculation of the MPI of atoms. Here, Z e f f s 0.72 for D 2 molecule. is a fitting parameter to simulate the effective Coulomb potential felt by the electron that tunnels out and F is the electric field of the laser. According to this model W is given by the following formula in atomic units. W s < Cn ) l ) < E

(
p
3r4

2 2 E . F

3r2

2 n ) y3r2

2 2 E . 3r 2

tion plotted in the inset of Fig. 5. From this curve it is observed that under the condition of low pressure, any pulse with energy above 13.35 mJ will ionize all molecules. Experimentally this is not the case when using high pressure gas. As we noticed before, the plasma density in 400 Torr gas using pulses with an energy of 40 mJ is around 10 14 cmy3 . This means that the probability of MPI does not reach above 10y5 10 14 r10 19 .. From Fig. 5 inset., this probability of MPI corresponds to a laser intensity of 5 = 10 13 Wrcm2 instead of 10 15 Wrcm2 which was expected in the vacuum. Thus, in a high-pressure gas the laser is not able to focus to its minimum focal size, and the peak intensity of the pulse remains so low that the density of electrons never reach the critical density necessary for the occurrence of breakdown. This explains why in the spectrum of gases radiated from a filament the level of continuum radiation and the line broadening are so low.

= 1 q g 2 .

A v ,g . ey

2F

g.

2.

4. Conclusion In conclusion by comparing the spectra radiated by self-luminous plasmas created by a laser pulse interacting with a gas target we have shown that the spectrum resulting from the interaction of ultra-short pulses is far superior in terms of low level of continuum and also smaller line broadening. This was attributed to the lower density of the plasma in the case of short laser pulses due to the defocusing effects which limits the intensity of the pulse in the interaction region and thus prevents further multiphotonrtunnel ionization of the gas. We verified our statement by measuring the upper limit of the plasma density in D 2 .

where E is the ionization potential of the molecule, v is the frequency of the laser and
2 n ) s '2 E rZ e f f

3.

Other quantities are given in w11x. The probability of the MPI is related to W through the following relation
q`

Pi s 1 y exp y

Hy` dtW

//

4.

Using this equation Pi is calculated as a function of laser intensity and pulse energy and presented it in the inset of Fig. 5. To see the accuracy of our semi-empirical model, we substituted the calculated value of Pi in Eq. 1. and presented the resulted ion curve in the Fig. 5 Dq Calc... As it is observed 2 there is a good overlap between theory and experiment note that at intensities above 2 = 10 14 Wrcm2 , corresponding to 13.35 mJ, the dissociation reduces the number of Dq ions to two Dq ions and therefore 2 it appears that the theory underestimates the experiment.. Thus we might have confidence on the theoretically calculated value of the probability of ioniza-

Acknowledgements It is our pleasure to acknowledge the technical assistance of S. Lagace and M. Lambert and fruitful discussions with C.M. Bowden and N. Akozbek. This work was supported in part by NSERC, le Fonds FCAR, US Army Research Office and the Department of National Defense of Canada.

484

A. Talebpour et al.r Optics Communications 183 (2000) 479484 w7x H.R. Griem, Plasma Specroscopy, McGraw-Hill, New York, 1964. w8x A. Talebpour, S. Larochelle, S.L. Chin, J. Phys. B 30 1997. 1927. w9x A. Talebpour, S. Larochelle, S.L. Chin, J. Phys. B 31 1998. L49. w10x A.M. Perelomov, V.S. Popov, M.V. Terentev, Soviet Phys. JETP 23 1966. 924. w11x A. Talebpour, J. Yang, S.L. Chin, Opt. Commun. 163 1999. 29, There were some unfortunate misprints in our original version. In Eq. 2, Z should be read Z e f f , formula 3. should be gs v 2 Ei r F, and in Eq. 8, Wn should be wn . The = sign should not be there.

References
w1x P.D. Maker, R.V. Terhune, C.M. Savage, Proc. 3rd Int. Conf. on Quantum Electronics, New York, 1963. w2x R.G. Meyerand, A.F. Haught, Phys. Rev. Lett. 11 1963. 401. w3x Y.B. Zeldovich, Y.P. Raizer, JETP 47 1964. 1150. w4x N. Akozbec, C.M. Bowden, A. Talebpour, S.L. Chin, Phys. Rev. E 61 2000. 4540. w5x A.R. Striganov, N.S. Sventitski, Tables of Spectral lines of Neutral and Ionized Atoms, IFI, New York, 1968. w6x Y. Liang, S. Augst, Y. Beaudoin, M. Chaker, H. Yu, A.D. Bandrauk, S.L. Chin, J. Phys. B 28 1995. 3661.

'