J. Greses (1,2), P.A. Hilton (2), C.Y. Barlow (1) and W.M. Steen (3)
(1)Engineering Department, University of Cambridge (2)TWI (3)Material Science Department, University of Cambridge, UK and Laser Group, University of Liverpool
Paper presented at ICALEO 2002, Scottsdale, Arizona, USA, 14 - 17 October 2002
During high-power cw Nd:YAG laser welding a vapour plume is formed containing vaporised material ejected from the keyhole. The gas used as a plume control mechanism affects the plume shape but not its temperature, which has been found to be less than 3000K, independent of the atmosphere and plume control gases. In this study high-power (up to 8kW) cw Nd:YAG laser welding has been performed under He, Ar and N 2 gas atmospheres, extending the power range previously studied. The plume was found to contain very small evaporated particles of diameter less than 50nm. Rayleigh and Mie scattering theories were used to calculate the attenuation coefficient of the incident laser power by these small particles. In addition the attenuation of a 9W Nd:YAG probe laser beam, horizontally incident across the plume generated by the high-power Nd:YAG laser,was measured at various positions with respect to the beam-material interaction point. Up to 40% attenuation of the probe laser power was measured at positions corresponding to zones of high concentration of vapour plume, shown byhigh-speed video measurements. These zones interact with the high-power Nd:YAG laser beam path and, can result in significant laser power attenuation.
When a high power Nd:YAG laser beam is incident on a target plate, a plume (a luminous body composed of evaporated particles, Fig.1) is observed to occur above the plate surface. The characteristics of the plume will depend on the applied energy density  . Some researchers [2-4] have indicated that the vapour ejected from the keyhole under high energy densities in Nd:YAG laser welding is a high-temperature thermally excited gas rather than a partially ionised plasma. Greses et al  studied the effect of gas side jets and gas environment on the plume formation and the process efficiency  , concluding that a side argon gas jet, delivered from the side, at a high flow-rate through a small diameter, precisely positioned nozzle, optimises penetration. A possible explanation for this was that the higher momentum of the heavier argon molecules could 'blow' the plume away.  Several studies [6,7] have been made to understand the fluid-dynamics of the interaction between a gas side jet and a laser welding plume, although much basic research is still needed in this area. Furthermore, it is believed that the plume is composed of very small particles, with an average diameter less than 50nm  , and that the particles cluster together as they condense  . This paper aims to further understand the mechanism of plume vapour formation during high power cw Nd:YAG laser welding, by measuring and calculating the attenuation effect of the plume on the laser beam.
A description of the experimental set-up (shown in Fig.2) used to produce a controlled atmosphere around the weld and the composition of the mild steel, have been described elsewhere. [4,5] A description of the high-speed video employed for filming the laser welding process and the spectrometer used to measure the plume spectra is also given in Greses et al. 
Fig.1. Nd:YAG-induced plume formation
Fig.2. Sketch of the experimental configuration
Two HAAS-LASER HL3006D series and one HL4006D series Nd:YAG lasers were used for welding. The output from each laser was combined in a 3-in-1 beam combining unit (BCU) manufactured by HIGHYAG Laser technologie.  A combined cw output power of up to 8kW was delivered to the workpiece through a step index fibre optic of core diameter 1mm and the beam was focused using recollimating and focusing lenses, to produce a 1.0mm diameter focal spot, with a lens to workpiece distance of approximately 150mm. The total power was obtained by adding the results of the three separate measurements. Some work was also conducted using a CO 2 fast axial flow gas laser, as described in ref.  .
Fig.3. Experimental set-up
A 9W Nd:YAG probe laser beam was arranged horizontally incident across the plume generated by the high-power Nd:YAG laser, at various positions with respect to the beam-material interaction point ( Fig.3). A power meter measured the probe laser power fluctuations, after interacting with the plume, transmitting the reading to a digital voltmeter with an incorporated graph plotter. The probe laser was a LEE LASER 703Tseries Nd:YAG laser, giving a 40W of cw power in a multimode beam. The use of a 1.2mm aperture in the laser reduced the available power to 9W. The resulting laser beam cross-section was elliptical, approximately 2mm in height and 3mmin width, with a nominal beam divergence of 1.5mrad. The power of the probe laser remained unchanged throughout the experiment. The inherent power stability of this laser was around 2%.
The particles emerging from the weld pool during welding were sucked from the vapour plume with a small vacuum pump positioned about 10cm from the plume and deposited on an inbuilt synthetic filter. For analysis in a Philips CM30transmission electron microscope (TEM) (LaB 6 filament at 300eV), the particles were flushed with methanol and transferred to a holey carbon film using conventional techniques. The shape and size distribution of the particles were examined with statistical techniques.
Particle formation, clustering and size
The mechanism, during laser welding, in which the metallic atoms evaporate from the keyhole walls and then grow by coalescence and/or form clusters is not well understood. At high temperatures and atmospheric pressure, vaporising atoms form a Knudsen layer over the metal surface, which controls the net rate of vaporisation and re-deposits material back onto the surface.  When high temperature plasma is formed the vaporisation rate may be inhibited due to a charging effect between the weld pool surface and the plasma, enhancing re-deposition at the Knudsen layer and reducing the net vaporisation rate.  Kotake  explained that the dynamic processes of particle clustering are influenced by the dynamic state of the atom, the kind of colliding atoms or molecules, and their dynamic state of motion. For Fe particles, their clusters show narrowly distributed sizes, since the clustering process is affected by the cohesive energy. The larger the cohesive energy the smaller the clusters, leading to low values of condensation or deposition rates in Fe. Detail of the condensation process of clusters is out of the scope of this paper, but can be found in Kotake. 
Matsunawa et al 
studied the production of ultra-fine particles (up to 150nm in diameter) generated when using a pulsed Nd:YAG laser beam. They concluded that a high evaporation rate, corresponding to a high power density, increased theproduction of ultra-fine particles and the particles were ejected perpendicular to the workpiece. They also noticed that the particles had a tendency to be linked together forming chains with the projected shapes of the Fe particles being mainly hexagons and octagons. The average size of the particles was 20nm, for an argon atmosphere, with about 95% of the particles being less than 40nm in size. For a helium atmosphere, the average particle size was 12nm, with an arrower particle size distribution than found in argon. Lacroix et al 
also measured the size of the particles present in a cw Nd:YAG laser generated plume, showing that the plume vapour consisted of very small particles of around 50nm in diameter condensing into larger agglomerates (around1µm in diameter), on the glass cover used to collect the particles.
Results and discussion
In Figures 4, 5 and 6, which show TEM pictures of the particles produced using both CO 2 and Nd:YAG laser wavelengths, it is possible to appreciate the even distribution of the plume particles and the way they link together to form particle chains. The chain formation may occur during the preparation method to obtain the TEM photographs, due to the surface tension between the particles and the methanol. The particles observed directly on a cover glass by Lacroix  did not form chains, but those analysed by Matsunawa  by TEM also did.
Figure 7 plots the distribution of particles in the plume against particle size observed when CO 2 laser welding, showing an average particle size of under 4nm. Figure 8 plots the same distribution from the particles in a plume created by the Nd:YAG laser welding process. For the later, the average size of the particles is larger at 40nm, with 82% of the analysed particles being under55nm in diameter. Measurements of the particle size made with different gas side jets and gas environments, did not significantly change the average size of the particles produced when Nd:YAG laser welding, as can be seen in Table 1. It should also be noted that, due to the maximum resolution of the TEM, a great number of very small particles (less than 10nm) could not be precisely analysed. If those particles were considered, then the realaverage size of the particles would be much smaller.
Table 1 Plume particle size in different gas environments when Nd:YAG laser welding
|Gas Side Jet (l/min)
||N 2 (20)
|Weld Speed (m/min)
|Particle Min Size (nm)
|Particle Average Size (nm)
|Particle Max Size (nm)
As can be seen by simple direct observation of the laser welding process, plume formation using a high power CO 2 laser is smaller than that observed when using a high power Nd:YAG laser. Mundra and DebRoy  explained that the vaporisation rate when using the CO 2 wavelength is low due to the electrons drifting back from the plasma and producing a negative charging effect in the liquid metal. The negatively charged liquid metal attracts positive ions from the Knudsenlayer, increasing re-deposition and, therefore, decreasing the net vaporisation rate. As can be seen in Fig.7, the average size of the particles produced when CO 2 laser welding was smaller and more constant, probably because the particles were ejected across the high temperature plasma. The high temperature gradient of the plasma, when CO 2 laser welding, may also stop the growth by coalescence of the particles.
Attenuation and scattering by the vapour plume
The attenuation (absorption plus scattering) of the probe laser power, is presented in Table 2, for different Nd:YAG laser welding powers. The position of the laser beam with respect to X and Y is shown in Figure 3. In Table 2, the values represent the average attenuation of the probe laser power during welding. The values in brackets represent the maximum percentage of power being lost during welding.
The results for the three different powers are very similar, except for the attenuation close to the top of the keyhole, where the values range from 12% to 38%. Temporal instabilities in the ejection of the plume, observed in high-speed video measurement, could account for this variance.
Table 2. Attenuation of probe laser by a plume for different incident laser powers and probe laser positions.
Tsubota el al  measured the attenuation of a probe laser (different laser types were tried: CO 2, Nd:YAG and He-Ne) crossing a plume perpendicular to the main laser beam. For the CO 2 probe wavelength, the beam attenuation was nearly constant, at 5%, along the length of the plume. For the Nd:YAG probe, the beam attenuation varied from over 50%, at a height of 3.5mm from the weld pool, to around 20% at 15mm. For the He-Ne probe laser, the attenuation was even higher than observed for the Nd:YAG wavelength, as Matsunawa and Ohnawa  also measured. These results are in line with Rayleigh scattering theory (eq. (2)), where the scattering loss is inversely proportional to the 4 th power of the wavelength. The shorter the wavelength the more dominant the scattering effect will be, for similar particle size.
Nevertheless, when the laser light encounters particles much smaller in size than the wavelength of the incident radiation (as in the case of Nd:YAG laser welding, with an average size of particle of 20nm compared to the 1.06µmwavelength of the laser light), the energy lost (E Attenuation) is caused by scattering and absorption  :
where P is the laser power, r the average radius of the particles, N the density of the scattering particles, z the distance through the particles and
where m is the complex refractive index (m=n+ki) and λ the laser wavelength. In equation (2) , Q SCA
the efficiency term in the Rayleigh scattering formula, is the same term used by Matsunawa and Ohnawa 
but using the complex refractive index rather than the dielectric constant.
For evaporated  Fe at 1.03µm m=3.81+4.44i and considering an average particle radius of 10nm, then Q SCA=3.15x10 -5 and Q ABS=2.2x10 -2. The scattering can thus be considered negligible compared to the absorption.
From Figure 9, it can be recognised that the highest concentration of particles in the vapour plume is at the tip of the keyhole. In the case of Fig.9, for example, the plume dimensions, for maximum particle concentration, are approximately 20mm in height with a base of 1mm in diameter (the laser spot size), forming a cone, with a volume of ~5.25x10 -9m 3. The maximum evaporated mass by a single pulse of Nd:YAG laser was measured by Nonhof  as 48µg, corresponding to ~5x10 11 particles of 10nm in diameter. It should be noted that in cw laser welding the evaporated mass would probably be less due to the effect of the Knudsen layer and the pressure balance inside the keyhole.Nevertheless, with that value, the density of particles in the region was ~3x10 19 (around 3.5% in volume ratio). Applying the coefficients from equations (2) and (3), and the density of particles into eq. (1), for a particle radius of 10nm and over a plume length of 20mm, the attenuation is~70%. This is a high value when compared to the results presented in Table 2 for the region at the very top of the keyhole, with maximum recorded attenuation in this work being ~40%. The cross-section area of the probe laser beam is around 25% bigger than the cross section area of the plume at the co-ordinates (0,0). Therefore, the attenuation values of the probe laser, in Table 2, at these co-ordinates will be artificially low.
Hansen and Duley  calculated the extinction, α EXT (or attenuation), caused by absorption and scattering by small particles using Mie theory for the Nd:YAG and CO 2 laser wavelengths. Mie theory becomes practical when the wavelength and the radius of the scattering spheres are of similar magnitude. Their results for the respective absorption ( α ABS) and scattering ( α SCA) coefficients are presented in Table 3.
Table 3. Comparison of scattering and absorption terms for Fe particles  .
|λ (µm)||Size Range||α SCA/ α EXT||ABS/ α EXT|
For an intensity of 10 10Wm -2 Hansen and Duley 15 calculated an absorption coefficient of 2.7m -1. Thus from Table 3, and for small particles, α SCA=0.761 m -1 and the overall α EXT =3.46 m -1. Applying a similar formula to eq. (1) to α SCA and α ABS:
then for a length of 5cm, the extinction or attenuation is of the order of 16%. The difference in attenuation between the values calculated with the method of Hansen and Duley  and Nonhof  can be attributed to the different value of the particle density employed.
Lacroix et al.  , calculated that the refractive index for an Nd:YAG laser-induced plume at atmospheric pressure was almost equal to unity, and its α ABS=3m -1. For this case, Lacroix believed that there is no plasma formation. His computer model of the scattering effect in the plume yielded temperatures between ~7000K at the top of the keyhole and ~4000K at a height of 5cm, for particles of 50nm in diameter and 5% volume ratio (ratio of particle volume to plume volume). The scattering produces a significant defocusing effect on the laser beam, which increases with high volume ratios. For larger particles (around 1µm) the plume temperature decreases (due to less absorption, as also seen in Table 3).
A different approach for understanding of the effect of particles in the plume on the laser beam, was studied by Russo et al  . For a single Nd:YAG laser pulse (i.e. no keyhole formation), a Huygens-Kirchhoff diffraction model was used with several plume shapes, at atmospheric pressures and the plume temperature decreased from the boiling point of themetal at the top of the weld pool, to ambient values several cm above the surface. The laser power at the centre of the plume was significantly reduced and the laser power intensity redistributed over a larger radius. Unfortunately,the effect of diffraction in cw keyhole Nd:YAG welding was not included in this research, but this should be further investigated for a better understanding of the effect of the plume on the laser beam.
In Table 4, the extinction coefficients [3,13,15] and the energy loss, from eq. (1) and (4), are in line with the transfer efficiencies ( η Transf and E Loss, Table 5) calculated by Greses el al  for welding with a 4kW Nd:YAG laser with a spot size of 0.6mm (Intensity ~1.24x10 10Wm -2).
Table 4. Coefficient of extinction, transfer efficiency and percentage of energy loss from Mie and Rayleigh scattering theories.
| ||Mie Scattering||Rayleigh Scattering|
||Hansen and Duley 15
|E Loss (for z=5cm)
In Table 5, the maximum attenuation or energy loss (30%), (or 70% in transfer efficiency), is achieved with a helium side jet at low speed (0.5m/min). At low speed the particles remain in the plume for a longer time in the path of the laser beam, building up the particle density and increasing the scattering and absorption distance. As the speed increases or a heavier gas (i.e. argon) is employed to 'blow' the plume away from the laser beam path, the energy loss is reduced.
Table 5. Efficiencies for 4kW Nd:YAG laser welding. 
|Gas Side Jet|
|HAZ Area (mm 2)||η Transf||E Loss|
|η Melt||E Melt|
The average size of the particles that constitute the vapour plume ejected in laser welding has been measured. Particles produced in high power CO 2 laser welding have, as an average, smaller size than particles produced with a high power Nd:YAG laser, due to the plasma effect over the Knudsen layer in CO 2 laser welding. Nevertheless, the average particle size produced with both lasers is smaller than 50nmm, much smaller than either the laser wavelength used.
A high concentration of these very small particles, in a small volume at the top of the keyhole, attenuates the incident laser beam power. Mie and Rayleigh scattering theories predict the attenuation of the laser power to be between10% and 70%, depending on the particle size and more significantly, the volume ratio. For an Nd:YAG incident laser wavelength, and using an Nd:YAG probe laser, transverse to the beam, the attenuation of the incident laser beam power was found to be as high as 40%. The attenuation decreased along the plume, away from the top of the keyhole as the volume ratio decreases. An increase in attenuation of the incident beam reduces weld penetration and fusion area,although the use of a gas side jet to 'blow' the plume away from the laser interaction path is an effective method of reducing this attenuation.
TWI equipment and the dedication of the staff of the Laser Department made this work possible. I would also thank my colleague Ton Van Helvoort for his help with the TEM analysis. Many thanks also to the EPSRC Engineering Instrument Pool, that kindly provided part of the equipment. This research was carried under EPSRC grant No. 99313089.
- Matsunawa, A., Katayama, S., Susuki, A. and T. Ariyasu. Laser Production of Metallic Ultra-fine Particles. Trans. JWRI, Vol. 15, No. 2, 1986.
- Matsunawa, A., Kim, J.D., Takemoto, T. and S. Katayama. Spectroscopic Studies on Laser Induced Plume of Aluminum Alloys. Proceedings of ICALEO'95, pp. 719, 1995.
- Lacroix, D., Jeandel, G. and C. Boudot. Solution of the Radiative Transfer Equation in an Absorbing and Scattering Nd:YAG Laser-induced Plume. J. Appl. Phys. 84:2443, 1998.
- Greses, J., Hilton, P.A., Barlow, C.Y. and W.M. Steen. Spectroscopic Studies of Plume/plasma in Different Gas Environments. Proceedings of ICALEO'01, E808, 2001.
- Greses, J., Barlow, C.Y., Hilton, P.A and W.M. Steen. Effects of Different Environments on CO 2 and Nd:YAG Laser Welding Process Efficiencies. International Congress on Laser Advanced Materials Processing (LAMP2002), 27-31 May 2002, Osaka, Japan. Proceedings to be published by SPIE, 2002.
- Reilly, J.P. Debris Plume Phenomenology for Laser/Material Interaction in High Speed Flowfields. Proceedings of SPIE 1397:661, 1990.
- Douay, D., Daniere, F., Fabbro, R. and L. Sabatier. Study of High Power Laser Welding: Effects of Plasma Blowing. Proceedings of SPIE 2789:202, 1996.
- DebRoy, T., Basu, S. and K. Mundra. Probing Laser Induced Metal Vaporization by Gas Dynamics and Liquid Pool Transport Phenomena. J. Appl. Phys. 70:1313, 1991.
- Mundra, K. and T. DebRoy. Calculation of Weld Metal Composition Change in High-Power Conduction Mode Carbon Dioxide Laser-Welded Stainless Steels. Met. Trans. 24B:145, 1993.
- Kotake, S. Formation of Metal and Non-metal Clusters through Laser and Electron Beam Vaporisation. Heat and Mass Transfer in Materials Processing, ed. by I. Tanasawa and N. Lior, Hemisphere Publishing Corporation, 1992.
- Tsubota, S., Ishide, T., Nayama, M., Shimokusu, Y. and S. Fukumoto. Development of 10kW Class YAG Laser Welding Technology. Proceedings of ICALEO'00, C-219, 2001.
- Matsunawa, A. and T. Ohnawa. Beam-Plume Interaction in Laser Materials Processing. Trans. JWRI, Vol. 21, No. 1, 1991.
- Nonhof. C.J. Material Processing with Nd-Lasers, Electrochemical Publications Ltd, 1988.
- American Institute of Physics Handbook, ed. by E.G. Gray, MacGraw-Hill, 1972.
- Hansen, F. and W.W. Duley. Attenuation of Laser Radiation by Particles during Laser Materials Processing. J. Laser Appl. 6:137, 1994.
- Russo, A.J., Akau, R.L. and J.L Jellison. Thermocapillary Flow in Pulsed Laser Beam Weld Pools. Weld. J. 69:23s, 1991.
- Olivier, C.A., Hilton, P.A. and D.K. Russell. Materials Processing with a 10kW Nd:YAG Laser Facility. Proceedings of ICALEO'99, D233, 1999
Meet the Authors
José Greses is a final year PhD student at the Engineering Department of the University of Cambridge (UK), sponsored by TWI (UK), with previous Mechanical Engineering studies at the Polytechnic University of Valencia (Spain)and an MSc in Marine Technology at Cranfield University (UK). E-mail: firstname.lastname@example.org