effects of ambient gas pressure and laser fluence - Laser-based ...

Thin Solid Films 446 (2004) 178–183

Deposition of particulate-free thin films by two synchronised laser sources: effects of ambient gas pressure and laser fluence ¨ a, I.N. Mihailescua, M. Kompitsasb,*, A. Giannoudakosb E. Gyorgy a Lasers Department, Institute of Atomic Physics, P.O. Box MG 54, 76900 Bucharest, Romania National Hellenic Research Foundation, Theoretical and Physical Chemistry Institute, 48 Vasileos Konstantinou Ave., 11635 Athens, Greece

b

Received 19 December 2002; received in revised form 29 August 2003; accepted 18 September 2003

Abstract The micrometer and sub-micrometer sized particulates present both on the surface and inside of pulsed laser deposited thin films and structures stand for the main drawback of the method in view of technological applications. We applied a two-laser system in order to withdraw the particulates in case of Ta and TaOx thin films. The Ta targets were irradiated by the first UV laser, while the second IR laser was directed parallel to the target surface, aiming to heat and evaporate the particulates. The morphology of the obtained thin films was studied by scanning electron microscopy. For the TaOx films, the ambient gas pressure influences, besides the size and density of particulates, their propagation velocity. This in turn results in the variation of the optimum delay time between the ablating UV and the second IR laser pulse. For the Ta films we found that a threshold fluence of the IR laser pulse exists, above which completely particulate-free films were deposited. 䊚 2003 Elsevier B.V. All rights reserved. PACS: 81.15.Fg; 68.37.Hk Keywords: Laser ablation; Synchronised laser sources; Particulate-free thin films

1. Introduction Pulsed laser deposition (PLD) allows, due to its simplicity and flexibility, for the growth of thin films of practically any material. Nevertheless, despite its versatility and relatively long research history, PLD proved to be still problematic for large-scale implementation in technological fields w1–5x. Indeed, besides the well known advantages of the PLD (accurate control of the film thickness and stoichiometry as well as the purity of the deposited material), it still shows a major drawback related to the presence of various types of particulates, both on the surface of the films as well as embedded into the bulk. We note that in high-performance electronic and optical devices particulate-free films are required. For this reason, the particulate origin and the different procedures aiming for their density reduc*Corresponding author. Tel.: q30-210-7273834; fax: q30-1-6012525. E-mail address: [email protected] (M. Kompitsas).

tion until the complete elimination still remain important topics of further research. The origin of these particulates has been assigned to different physical mechanisms taking place during laser irradiation, depending on concrete processing conditions w1–8x: (i) material dislocation caused by the subsurface superheating of the target, (ii) liquid phase expulsion under the action of the recoil pressure of the ablated substance (vapour and plasma), (iii) condensation (clustering) of the evaporated material in the expanding plumes, (iv) blast-wave explosion at the liquid (melt)– solid interface, andyor (v) hydrodynamic instabilities at the molten target surface. The proper choice of the processing conditions, such as laser fluence, wavelength, and ambient gas pressure, proved to be the key parameter for the particulates density reduction. We studied w9–14x the formation mechanisms, structure and composition of particulates present on the surface or embedded into the bulk of SiC, TiC, TiN, CxW and CNx thin films obtained by PLD.

0040-6090/04/$ - see front matter 䊚 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2003.09.071

¨ E. Gyorgy et al. / Thin Solid Films 446 (2004) 178–183

In order to minimise the particulate density, in addition to optimising laser parameters (laser wavelength and fluence) and target surface properties (molten, highdensity or well-polished), several other solutions have been proposed and tested during last years. Among these are mechanical shutters used as velocity filters w15x, offaxis geometries of the substrate so that direct impinging of the particulates on it is avoided w16–19x and particulates deflection by a supersonic pulsed gas-jet, synchronized with the laser pulse w20,21x. All these techniques can be applied in particular cases only, because they often reduce the deposition rate andyor require the presence of a gas. Fragmentation of particulates by means of an additional laser beam propagating parallel to the substrate surface or dual-laser beam ablation from a single target have also been reported. In the first case w22,23x a UV laser beam followed the IR ablating laser with a fixed delay. Nevertheless, the particulates ejected from the target were only partially removed. In the second case w24,25x a long (t;200 ns) IR CO2 laser pulse was temporally overlapped with a UV (t;20 ns) ablating laser. It served for both melting the target and vaporization of particulates in plasma plumes. Droplet filtering has been performed also by cross-beam PLD w26x. In this technique, laser plumes from two different targets ablated by two synchronised lasers intersect in the proximity of the targets forming one single plume. A diaphragm is placed perpendicularly to the expansion direction of this sole plume to shadow the substrate from particulates. We recently reported w27x the deposition of Ta thin films in vacuum, using two laser sources whose controlled delay was within the range 0–1.5 ms. First, we used a ns UV laser source to vaporise the target. A second ns IR laser beam was directed parallel to the target surface, intersecting the ablation plasma. The IR laser pulse were sent with different controlled delay times after the ablating UV laser pulse, aiming for heating and vaporising the particulates present in the ablation plasma. Scanning electron microscopy (SEM) investigations of the deposited Ta thin films indicated a strong variation of particulates density on the films surface as a function of the time delay between the two laser pulses. This proved the crucial role of a proper delay time selection in particulates elimination. Our selection for a second pulsed laser source in IR is motivated by the much higher absorption at these wavelengths by vapours, plasma and particulates. This is mainly driven by the inverse Bremsstrahlung (IB) process in the initial plasma (i.e. a few millimeters in front of target surface) w2,3x. Indeed, the characteristic absorption coefficient is proportional to the square of the incident laser wavelength. Correspondingly, this effect is minor for wavelengths in UV. Furthermore, because of the supplementary kinetic energy gained from radiation by IB, free electrons excite and ionise

179

Fig. 1. Scheme of the experimental set-up.

the vapour species by electron collisions w4x. Accordingly, the second IR laser pulse plays a double role: (i) direct heating, breaking and evaporation of particulates present in the ablation plasma, and (ii) increasing kinetic energies of the plume species. This last effect directly contributes to the particulates evaporation. Moreover, it induces a higher degree of excitation of species. This allows for enhanced gas phase reactions and ensures better films adhesion, crystallinity and morphology w1– 3 x. In the present work we go one step further by applying the two-laser method described above to deposit particulate-free Ta oxide films from a Ta target in ambient oxygen as reactive gas. We investigate the influence of the reactive gas pressure on particulates formation as well as on the optimum time delay between the UV and IR laser pulses for the particulates density reduction. We also study the effect of the IR pulse energy on the particulates elimination. We determine the minimum IR laser fluence necessary for complete vaporisation of particulates during their transit from the target to the collector, leading to deposition of completely particulatefree thin films. 2. Experimental procedure The experimental set-up used is depicted schematically in Fig. 1. The laser irradiations were performed in a stainless steel vacuum chamber. Prior to each deposition, it was evacuated down to a residual pressure of 7=10y4 Pa. For Ta targets ablation we applied the UV laser pulses generated by a Lumonics Mo. TE-861T excimer laser. It was operated either at ls248 nm (KrF*, tFWHM;12 ns), or at ls193 nm (ArF*,

180


tFWHM;10 ns). The laser pulses succeeded to each other at a repetition rate of 10 Hz. The laser beam incidence angle onto the target surface was of ;458. The UV laser beam was focused on the target surface in a spot of 0.95 mm2. IR pulses were generated by a controllable, timedelayed Quantel Mo. YG851 Nd:YAG (ls1064 nm, tFWHM;10 ns) laser. The IR laser beam propagates parallel to the target surface and crosses the flux of the ablated substance at a distance of 2 mm above the focus of the UV laser beam. A cylindrical lens of 10 cm focal length, placed normally to the plasma flux, is focusing the IR laser beam on the plane containing the plasma column’s symmetry axis. So there is a volume in front and behind the focal plane where the IR laser fluence exceeds a threshold value, for which efficient particulates evaporation can take place. The controlled delay between the UV and IR laser pulses was set with a stable and high precision digital delay generator (EG and G Mo. DDG 9650). The two laser pulses were detected by a fast photodiode (rise time -1 ns) from Motorola (MRD 500) and monitored on a 350 MHz Tektronix oscilloscope. The time jitter between the two laser pulses was of a few ns. Correspondingly, the delay of interest in our experiments within the range 10 msy1 ms could be fixed up with a relative accuracy of 5=10y4 to 5=10y2. To avoid fast drilling, the targets were mounted on a vacuum-compatible computer controlled XY translator, so that the laser spot scans the target surface both vertically and horizontally. A total surface of 10=10 mm2 was uniformly scanned. The movement was selected such that every target point was irradiated five times. Then, the target was translated to the next irradiation location, 0.5 mm apart. The substrate holder was also connected to the XY translator, performing an identical movement as that of the target, in order to improve the thickness uniformity of the deposited films. The first series of experiments were performed in oxygen as reactive gas. The incident laser fluence of the KrF* laser source was set to a value of 6.3 Jycm2. For the deposition of one film the Ta target was submitted to the action of 15 000 subsequent UV laser pulses. The energy of IR laser pulses was 140 mJ. After focusing on the plane containing the plasma columns’ symmetry axis we got a fluence of 7 Jycm2. The oxygen gas pressure measured with a MKS Baratron gauge was set at values of 10, 16, or 20 Pa. The time delay between UV and IR laser pulses was increased up to 1 ms. The second series of experiments was performed under a 7=10y4 Pa residual pressure. The radiation generated by the ArF* laser source was focused on the target surface to an incident fluence of 0.75 Jycm2. Due to the low fluence value and thus low ablation rate, the number of laser pulses applied for the deposition was increased to 50 000. The IR laser pulse energy was

varied in the range 50–240 mJ. The corresponding fluence incident on the plasma was of 2.5–12 Jycm2. In this case, the time delay between UV and IR laser pulses was set at a unique value of 200 ms. The ablated substance was collected on glass slides placed parallel to the target surface at 35 mm separation distance. During depositions, the collectors were kept at room temperature. The obtained structures were characterized from topographical point of view with a Sloan Dektak IIA stylus profilometer. The morphology of the deposited films was investigated by SEM with a Cambridge S120 and a JEOL TEM Scan 200 CX instruments. 3. Results and discussion According to the profilometric determinations, the deposited films exhibit large thickness uniformity (better than 3%). This was achieved by a special design of the experimental set-up, coupling the movement of the target with respect to the UV ablating laser spot, in correlation with the change of the substrate position with respect to the ablation plasma. 3.1. Effects of the ambient gas pressure In Fig. 2a–c we present the SEM micrographs of reference thin films, deposited by multipulse KrF* laser ablation of Ta targets, in the absence of the second IR laser action. The density of particulates deposited on the films surface gradually decreases from approximately 106 ycm2 at 10 Pa (a) to 5=104 ycm2 at 20 Pa (c). Nevertheless, their mean size increases with pressure from several hundreds of nm at 10 Pa, to a few micrometers at 20 Pa. A similar decreasing tendency of the particulates density with the oxygen pressure was reported by Hino et al. w28x. They performed SEM investigations of Ta oxide thin films deposited by KrF* laser irradiation of Ta targets. Moreover, it was demonstrated that when the background pressure is sufficiently high (G7 Pa), ablated particulates can be ‘blown away’ by the gas flux, causing a significant decrease of particulates density on the surface of deposited films w29x. The gradual increase of particulates size with the increase of gas pressure could be the effect of enhanced vapor phase condensation and coalescence of particulates during their transit from target to collector. In addition, laser induced oxidation on target surface at higher oxygen pressures gradually modifies its thermophysical and optical properties. This may result in a thicker molten surface layer w1,3x. Besides size and density of particulates, the ambient gas pressure can influence their propagation velocity as well. As already reported, the ablation plasma contains micrometer and sub-micrometer sized particulates and droplets having velocities up to 103 –104 cmys. This is


181

O2 with a delay time of 750 ms when we observed the complete elimination of the particulates (Fig. 3h). This corresponds to an average particulates velocity of 2– 3=102 cmys. For larger delays than the optimum values 500 and 750 ms at 10 Pa (Fig. 3b–c), and 1000 ms at 16 or 20 Pa (Fig. 3f, i) we observed the re-appearance of particulates on the surface of deposited films. 3.2. Dependence of particulates density on the IR laser fluence

Fig. 2. Typical SEM micrographs of thin films deposited by multipulse KrF* laser irradiation with a fluence of 6.3 Jycm2 in O2 at 10 Pa (a), 16 Pa (b), and 20 Pa (c), respectively. The magnification is identical for all images.

about two orders of magnitude lower than the velocity of atomic and molecular species w25,30x. Consequently, for each pressure value we optimised the time delay between UV and IR laser pulses. After preliminary tests, the IR laser fluence was fixed at 7 Jycm2. As can be followed from Fig. 3a–i, the optimum time delay for which significant diminution of particulates density is observed increases with gas pressure from 250 ms at 10 Pa (a) to 750 ms at 20 Pa (h). These rather long delay times are congruent with particulates velocities within the range of 102 –103 cmys. The shift of the optimum delay time towards higher values is clearly indicative for the slowing down of the particulates with increase of gas pressure. The best results were obtained in 20 Pa

In the next series of depositions we used for ablation an ArF* laser source. The irradiation of Ta targets was conducted under a 7=10y4 Pa residual pressure (vacuum). As well known, an important parameter that influences the particulates size and density is the ablating laser fluence. In good agreement with previous results reported in literature w1,24,25x, we found that particulates size and density rapidly decreases when the ArF* incident laser fluence is decreased from 11 to 3 Jycm2 w27x. In accordance with this tendency we observed in the new experiments smoother surfaces covered with particulates having a density smaller than 102 ycm2. The strong reduction of the particulates density accompanying the decrease of the laser fluence can be attributed to the lower recoil pressure of the laser ablation plasma. For technological and economical reasons, any energy losses have to be avoided. It is therefore mandatory to establish the minimum IR laser fluence necessary for the complete elimination of particulates. To this purpose the ablating ArF* UV laser fluence was kept at 0.75 Jy cm2, while the time delay between UV and IR laser pulses was set at 200 ms. The IR laser pulse energy was varied from 50 to 240 mJ, corresponding to an incident fluence of 2.5–12 Jycm2 on plasma. Micrometer and submicrometer particulates are still visible on the surface of the films obtained after interception by 50–100 mJ IR pulses (i.e. for 2.5–5 Jycm2 incident fluence (Fig. 4a,b)). Nevertheless, the particulates density decreases with the increase of the IR laser pulse energy. Moreover, when the IR laser pulse energy was increased in excess of 150 mJ, (i.e. for 7.5 Jycm2 laser fluence and higher), the complete elimination of particulates was observed (Fig. 4c,d). 4. Conclusions We developed a system with two-synchronised laser sources, which allows for the efficient diminution till the complete exempt of particulates in PLD. The first laser source (UV) is used for ablation while the second one (in IR), directed parallel to target surface, heats vapours and plasma and vaporises (breaks) the particulates. Our choice is based upon the enhanced absorptivity by plasma and vapours of IR laser radiation. The elimination of particulates by the second IR laser pulse

182


Fig. 3. Typical SEM micrographs of thin films deposited by KrF* multipulse laser irradiation with a fluence of 6.3 Jycm2 in 10 Pa O2 for (a) 250 ms, (b) 500 ms, and (c) 750 ms delay time between the UV and IR laser pulses; in 16 Pa O2 for (d) 500 ms, (e) 750 ms, and (f) 1000 ms time delay between the two laser pulses; in 20 Pa O2 and (g) 500 ms, (h) 750 ms and (i) 1000 ms time delay between the two laser pulses. The IR laser pulse energy was 140 mJ and the incident fluence on plasma 7 Jycm2, respectively. The magnification is identical for all images.

Fig. 4. Typical SEM micrographs of thin films deposited by multipulse ArF* laser irradiation in vacuum. The ablation fluence was set at 0.75 Jycm2, and the time delay between UV and IR laser pulses was of 200 ms. The IR laser energyyfluence was of (a) 50 mJy2.5 J cmy2, (b) 102 mJy5.1 J cmy2, (c) 159 mJy8.2 J cmy2, (d) 203 mJy10.2 J cmy2. The magnification is identical for all images.


was possible by appropriately choosing the optimum time delay between the two laser pulses. In turn, the later is determined by the particulates velocity and strongly depends on the deposition parameters, i.e. ambient gas pressure as well as UV laser wavelength and fluence. The method proved very efficient in case of PLD in vacuum, when we observed the complete elimination of particulates. Some residual, very few, particulates are still visible in case of PLD in gases. We demonstrated that both the ablation process and the particulates elimination are governed by threshold values, while a quite narrow window exists in all cases for obtaining optimum decrease of particulates densities up to their complete elimination. All procedures have been optimised in respect with the energy budget and experimental geometry. Acknowledgments The authors acknowledge with thanks the financial support of a NATO PST.CLG.977325 Collaborative Linkage Grant and additional financing from a GreekRomanian Joint Research and Technology program on ‘Multiwavelength laser plasma investigations for applications in thin films deposition and processing’. References w1x D.B. Chrisey, G.K. Hubler (Eds.), Pulsed Laser Deposition of Thin Films, Wiley, New York, 1994. w2x M. Von Allmen, A. Blatter, Laser-Beam Interactions with Materials, 2nd ed, Springer, Berlin, 1995. w3x D. Bauerle, Laser Processing and Chemistry, 2nd ed, Springer, Berlin, 1996. w4x W.W. Duley, UV Lasers: Effects and Applications in Materials Science, University Press, Cambridge, 1996. w5x Ion N. Mihailescu, Eniko Gyorgy, in: T. Asakura, (Ed.) ‘Pulsed Laser Deposition: An Overview’, Invited contribution to the 4-th International Commission for Optics (ICO) Book ‘Trends in Optics and Photonics’, (ICO President) Springer Series in Optical Science, p. 201, 1999. w6x K. Murakami, in: E. Fogarassy, S. Lazare (Eds.), Laser Ablation of Electronic Materials – Basic Mechanisms and Applications, Elsevier, Amsterdam, 1992, p. 125. w7x R. Kelly, J. Rothenberg, Nucl. Instrum. Methods B 7–8 (1985) 755.

183

w8x A.B. Brailovsky, S.V. Gaponov, V.I. Luchin, Appl. Phys. A 61 (1995) 81. w9x V.S. Teodorescu, I.N. Mihailescu, E. Gyorgy, A. Luches, M. Martino, L.C. Nistor, J. Van Landuyt, J. Hermann, J. Mod. Opt. 43 (1996) 1773. w10x I.N. Mihailescu, A. Lita, V.S. Teodorescu, E. Gyorgy, R. Alexandrescu, A. Luches, M. Martino, A. Barborica, J. Vac. Sci. Technol. A 14 (1996) 1986. w11x G. Leggieri, A. Luches, M. Martino, A. Perrone, R. Alexandrescu, A. Barborica, E. Gyorgy, I.N. Mihailescu, G. Majni, P. Mengucci, Appl. Surf. Sci. 96 (1996) 866. w12x N. Chitica, E. Gyorgy, A. Lita, G. Marin, I.N. Mihailescu, D. Pantelica, M. Petrascu, A. Hatziapostolou, C. Grivas, N. Broll, A. Cornet, C. Mirica, A. Andrei, Thin Solid Films 301 (1997) 71. w13x I.N. Mihailescu, V.S. Teodorescu, E. Gyorgy, A. Luches, A. Perrone, M. Martino, J. Phys. D: Appl. Phys. 31 (1998) 2236. w14x I.N. Mihailescu, E. Gyorgy, V.S. Teodorescu, Gy. Steinbrecher, J. Neamtu, A. Perrone, A. Luches, J. Appl. Phys. 86 (1999) 7123. w15x D. Lubben, S.A. Barnett, K. Suzuki, S. Gorbatkin, J.E. Greene, J. Vac. Sci. Technol. B 3 (1985) 968. w16x E. Agostinelli, S. Kaciulis, M. Vittori-Antisari, Appl. Surf. Sci. 156 (2000) 143. w17x N. Inoue, T. Ozaki, T. Monnaka, S. Kashiwabara, R. Fujimoto, Jpn. J. Appl. Phys. 36 (1997) 704. w18x N. Inoue, S. Kashiwabara, S. Toshima, R. Fujimoto, Appl. Surf. Sci. 96–98 (1996) 656. w19x R.J. Kennedy, Thin Solid Films 214 (1992) 214. w20x S.J. Barrington, T. Bhutta, D.P. Shepherd, R.W. Eason, Opt. Comm. 185 (2000) 145. w21x P.K. Schenck, M.D. Vaudin, D.W. Bonnell, J.W. Hastie, A.J. Paul, Appl. Surf. Sci. 127–129 (1998) 655. w22x G. Koren, R.J. Baseman, A. Gupta, M.I. Lutwyche, R.B. Laibowitz, Appl. Phys. Lett. 56 (1990) 2144. w23x H. Chiba, K. Murakami, O. Eryu, K. Shihoyama, T. Mochizuki, K. Masuda, Jpn. J. Appl. Phys. 30 (1991) L732. w24x P. Mukherjee, S. Chen, J.B. Cuff, P. Sakhivel, S. Witanachchi, J. Appl. Phys. 91 (2002) 1828. w25x S. Witanachchi, K. Ahmed, P. Sakthivel, P. Mukherjee, Appl. Phys. Lett. 66 (1995) 1469. w26x A. Tselev, A. Gorbunov, W. Pompe, Appl. Phys. A: Mater. Sci. Process 69 (1999) 353. w27x E. Gyorgy, I.N. Mihailescu, M. Kompitsas, A. Giannoudakos, Appl. Surf. Sci. 195 (2002) 270. w28x T. Hino, M. Nishida, T. Araki, Surf. Coat. Technol. 149 (2002) 1. w29x T.E. Itina, A.A. Katassonov, W. Marine, M. Autric, J. Appl. Phys. 83 (1998) 6050. w30x B. Hopp, N. Kresz, Cs. Vass, Z. Toth, T. Smausz, F. Ignacz, Appl. Surf. Sci. 186 (2002) 298.