J. Greses (1,2), P.A. Hilton (2), C.Y. Barlow (1) and W.M. Steen (3)
(1)Engineering Department, University of Cambridge
(3)Material Science Department, University of Cambridge, and Laser Group, University of Liverpool
Paper presented at the International Congress on Application of Lasers and Electro-Optics (ICALEO) 15-18 October 2001, Jacksonville, USA.
Under high power densities in laser welding, vaporised material is ejected from the keyhole and forms a plume/plasma above the weld pool. From previous studies, differences in the plume formation and extent of ionisation have been observed depending on the type of laser employed. In this study a comparison between CO 2 and Nd:YAG laser welding has been performed using the same energy density (~1.24MW/cm 2, produced using 3.5kW of power and a focal spot size of 0.55mm) under He, Ar and N 2 gas environments and vacuum. Spectroscopy measurements were made to investigate the temporal evolution of the plume temperature as a function of the atmosphere and plume/plasma control gases. The Boltzmann-plot method was used to analyse the temperature in the CO 2 plasma, while the Nd:YAG plume spectrum was fitted to blackbody radiation curves for the temperature calculation. The temperatures of both the CO 2 laser-generated metallic vapour and the Nd:YAG laser-generated plume remain quite stable, despite different gas environments and welding speeds. Plume/plasma evolution has also been recorded with a high-speed camera at 9000 frames/second and these results have been correlated with the characteristics of the weld shape.
Laser welding is becoming a relatively standard process in the modern factory. Initially CO 2 and more recently Nd:YAG lasers have increased their role due to their precision, low distortion, efficacy with difficult-to-weld materials, high production rate and process flexibility capabilities.  Welding experience in the mid-70s with CO 2 lasers led to the use of gas side jets of high ionisation potential to suppress or control the plasma formation.  Helium is the most efficient gas for plasma control although due to its high price, argon and nitrogen have also been used for this purpose.  Based on the experience of CO 2 laser welding, it was logical to apply the same approach for the control of plasma to the Nd:YAG laser welding process. However, Matsunawa  and Lacroix  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. Therefore there is a need to establish the differences between CO 2 and Nd:YAG plume/plasma formation during keyhole welding in order to optimise the weld penetration in mild steel. A comparison experiment was devised, keeping as many parameters as possible the same for both sets of experiments performed, except the wavelength of laser light used (10.6µm for the CO 2 laser and 1.06µm for the Nd:YAG laser).
Two different lasers were used for generating the plasma/plume conditions when laser welding. A CO 2 LASER ECOSSE AF5 fast axial flow laser, giving 3.5kW cw power at the workpiece with a multimode beam, was focused by a Zn-Se lens with a focal length of about 300mm, to a focal spot of 0.62mm in diameter. This relatively large focal spot was deliberately chosen for comparability with the Nd:YAG laser welding experiments. The Nd:YAG laser used was a GSI LUMONICS ASM series laser. The laser delivered 3.5kW cw to the workpiece, through an optical fibre and the beam was focused using 1:1 imaging to produce a 0.6mm diameter focal spot with a lens to workpiece distance of about 190mm. The power of both lasers remained unchanged throughout the experiments.
A modified glove-box was used to form a controlled atmosphere around the weld. The size of the chamber, with an octagonal section (length 120cm, sides 30cm) and a total volume capacity of around 2m 3, in conjunction with the short length of welds produced, minimised the effects of fume build up. A flat Zn-Se window positioned in one of the ports in the top of the chamber provided access for the CO 2 laser light and it proved convenient to seal the Nd:YAG focusing head to the same port, for the second set of experiments (See Figure 1).
Fig. 1. Sketch of the experimental configuration
The plasma/plume radiation was directly gathered by a fibre optic, although some of the plasma radiation when CO 2 laser welding was delivered by a telescope into the same fibre optic, connected to a 320 Princeton Instruments spectrometer with a slit aperture of 20µm. This system was calibrated with a mercury lamp for a diffraction grating of 2400 G/mm with a dispersion of 1.25nm/mm for 30mm focal plane coverage (37.5nm). The spectra were recorded by a ST-138 Princeton Instruments Optical Multichannel Analyser (OMA). For the Nd:YAG laser experiments no telescope was available, and the distance from the plume to the fibre optic was 250mm, effectively integrating over a plume length of 25mm. The number of spectra taken every time the system was triggered was 50 or 252, depending on the spectrometer configuration, with an exposure time for each spectrum of 0.01sec, plus an interval of around 16µsec between each reading.
A Kodak HS 4540 high-speed video camera positioned outside the chamber was used to record the plasma/plume behaviour at 9000 fps (frames per second). A 150mm focal length lens in conjunction with 80mm extension tubes were used allowing filming from outside the chamber. The camera was triggered in synchronisation with the spectrometer via the laser controller unit. A pulsed copper vapour laser (Oxford Lasers) was used to illuminate the weld pool and the plasma/plume. A narrow band pass filter was placed in front of the camera lens.
The sealable chamber allowed experiments to be performed in atmospheres of either air, helium, argon or nitrogen. In addition helium, argon and nitrogen side jets could be applied at different flow rates (10-40l/min), to a position about 1mm ahead of the laser beam-material interaction point. Nozzle diameters of 2 and 4mm were used, and the side jet was positioned at 0°, 45° and 70° to the horizontal.
The material used throughout the experiments was a mild steel with a thickness of 12mm.
Thermal sources, such as the plasma/plume generated in laser welding, can be divided into two classes: blackbody radiators and line sources. 
Blackbody radiators are opaque bodies or hot, dense gases that radiate at virtually all wavelengths. Line sources radiate mainly at discrete wavelengths. The vapour emission emerging from the keyhole when CO 2
laser welding has been described in several studies as lines sources. [7-9]
These studies agree with the theoretical process of vapour formation, but the experimental temperature and electron density results from that vapour show a great variance (more than 5000K depending on the parameters). Nevertheless from the results available made with both gas side jet and no side jet, the level of ionisation in the vapour is low, therefore the vapour is defined as partially ionised plasma. 
Under local thermodynamic equilibrium (LTE), the Boltzmann distribution of the excited states within the plasma are governed by the Saha equation. 
where N e,i,o
are the density of electrons, ions and netral atoms respectively; T e
is the electron temperature; m e
is the electron mass; g e,i,o
are the degeneracy factors of electron, ions and neutral atoms respectively; E e
is the ionisation potential for the neutral atoms, and k
is the Boltzmann constant.
The assumption of LTE in the plasma is valid since typical electron densities are two orders of magnitude higher than required in equation   :
is the largest energy gap of the atomic level system (eV).
In optically thin plasmas, i.e. no self-absorption occurring within the plasma, the spectral intensity of its optical emission ( I mn ) for the transition from the ground state ( m) to the upper state ( n), is given by:
where n m(x) is the density of atoms in the excited state m as a function of distance x through the plasma; A mn is the atomic transition probability between the upper and lower state and
is the energy of the photon emitted. Actually, some degree of self-absorption occurs for some spectral lines when the upper state is within an energy ~kT e of the ground state. 
The population of the excited states is given by the Boltzmann distribution:
with N being the population density; g the statistical weight, and Z the partition function.
Substituting equation  in , then:
where I mn is the spectral intensity for the transition from the ground state ( m) to the upper state ( n);
wavelength of the spectral line; g m the degeneracy factor of the upper state; A mn the atomic transition probability between the upper and the lower state; N the population density; Z the partition function; E m the energy of the upper state; k the Boltzmann constant, and T e the electron temperature.
If the left part of equation  is plotted against E m for several spectral lines, the slope of the line will be equivalent to -1/(kT e), from which it is possible to calculate the electron temperature ( T e ). This is known as the Boltzmann-plot method. However, it is possible to use a faster, though less precise, method for calculating the temperature if two appropriate spectral lines of the same element are chosen. The electron temperature by this pair-line method is given by  :
being the wavelength of the excited states s and m respectively.
For Nd:YAG laser welding of mild steels the emission lines of the vapour are known not to be as strong as in CO 2 laser welding,  making the identification of the spectral lines extremely difficult. Since its temperature is above absolute zero the Nd:YAG plume is considered a thermal source, so the maximum energy radiated from the vapour plume can be calculated by applying the blackbody radiation theory. The formula that describes the distribution of energy density per unit time per unit wavelength from the radiation of a blackbody has been given by Planck as :
where E is the energy distribution per unit wavelength (in J/m); λ the wavelength ( m); T the temperature ( K); c the speed of light; k the Boltzmann's constant, and h Planck's constant.
Blackbodies are theoretical objects with emissivity equivalent to 1, meaning they both are perfect absorbers and emitters of radiation. Hence, it is more correct to speak about real objects as 'greybodies', implying emissivities lower than 1. Nonetheless, most objects resemble a blackbody at certain temperatures and wavelengths. The radiated light from the weld pool has been used for monitoring purposes in arc welding applications.  Mueller et al.  showed that measuring the spectrum of the weld pool at different angles when Nd:YAG laser welding it was possible to distinguish between the emission of the weld pool and that of the plume. Nevertheless the minimum emission measured from lower angles was very similar to the spectrum of the plume alone, implying a very similar temperature for both.
Analysing the spectrum from only the plume formed above the weld pool and assuming an emissivity equal to 1, it was possible to fit and compare this to the theoretical blackbody spectrum at any given temperature (see equation ).
Plasma/plume temperatures evaluation
For the spectra readings, bead-on-plate runs in the down-hand position (PA) inside the chamber were made using speeds in the range 0.5-1.5m/min. For CO 2 laser welding, most of the measured spectra without the telescope were analysed in the region between 338 and 362nm ( Figure 2). Ten atomic lines were identified from the recorded spectra. More lines were available, but they were discarded because either their identification was uncertain or there was no corresponding data available in the literature. Figure 3 shows the Boltzmann-plot technique employing the ten lines, produced from data obtained when welding with air in the chamber and a helium side jet, and a speed of 0.5m/min. Each point corresponds to an average of 20 spectrometer readings. Using the relationship given in equation  and data from the NIST Atomic Spectra Database  from the gradient of the line, a plasma temperature of ~10680K was calculated. If for example some atomic lines (357.0097, 360.8861 and 358.1195nm) are left out of the plot, the resultant gradient indicates a temperature of ~7750K. Figure 3 also shows that the gradient might be determined by the relative intensity of just two lines, for example those at 339.9336 and 344.3878nm. Therefore, applying the pair-line method to several pairs of lines, it is possible to produce a rapid determination of the plasma temperature. The upper energy levels of the chosen pair of spectral lines have to be as distant as possible, so their intensity ratio is more sensitive to the variation of plasma temperature. Depending on which pair of spectral lines was used, the calculated plasma temperatures were in the region between 6800 and 11800K (See Table 1).
Fig. 2. Portion of the spectrum of the Fe plasma
Fig. 3. Boltzmann-plot method
Table 1. Plasma temperature variation with different combination of two spectral lines
|λ theoretical (nm)||E upper (eV)||λ theoretical (nm)||E upper (eV)||Temperature (K)|
The results for a pair-line determination of the plasma temperature (using atomic lines at 339.265 and 344.3878nm) under different gas environments are given in Table 2. These results show little variance in plasma temperature under different gas environments. The plasma temperature results made using the telescope, differ from those using the standard aperture of the fibre optic ( Table 3), although still being within the range of the plasma temperatures calculated with the Boltzmann-plot method in Figure 2.
Table 2. Plasma temperature variation with different gas atmospheres and speeds
| ||Air in Chamber|
He Side Jet
|Air in Chamber|
Ar Side Jet
|Argon in Chamber|
Ar Side Jet
|Argon in Chamber|
No Side Jet
Table 3. Plasma temperature variation with different gas atmospheres and speeds with the use of a telescope to focus the spectra into the fibre optic
| ||Air in Chamber with Ar Side Jet||Air in Chamber with He Side Jet|
|Temperature in K|
(344.099 & 354.1086)
|Temperature in K|
(344.387 & 339.2652)
|Temperature in K|
(344.099 & 354.1086)
|Temperature in K|
(344.387 & 339.2652)
The three experimental curves in Figure 4 represent the maximum, average and minimum spectra intensity for several readings using Nd:YAG laser light, recorded during a weld run in air conditions with an argon side jet. When the fitting procedure outlined in the previous section for the analysis of Nd:YAG laser welding spectra was used with the results obtained using all the different gas conditions and other parameters, the range of blackbody temperatures corresponding to the best fits lay between 2000 and 2200K. Since the spectrometer measures the intensity of the emitted light in relative units, this relative intensity can be modified to fit to a blackbody curve, as long as the intensity ratio at each particular wavelength is kept constant. The fitting of the spectra was difficult due to the small variations of intensity over the wavelength range, so that an average method for smoothing those variations was applied. This average was then visually fitted to the blackbody curves by multiplying it by an appropriate constant. The four temperature curves (2000-2300K) shown in Figure 4, are the energy distribution per unit wavelength given by the blackbody radiation equation , and they are thus fixed at that given temperature. The two sharp peaks in Figure 4 correspond to the two frequencies (511nm and 578nm) at which the copper vapour laser used for video illumination purposes emits.
Fig. 4. Blackbody radiation fitting to the shape of the Nd:YAG laser welding spectrum
The maximum intensity of the Nd:YAG spectra is reached around 560nm wavelength, after which the intensity decreases. That maximum intensity probably corresponds to a mixture of two high intensity atomic lines (at 561.5652 and 558.6763nm) that are not possible to fit to the spectra of the blackbody curve. Other peaks can be noticed in the spectra, but they were very difficult to identify, and non of those peaks were considered for the fitting.
The estimated plume temperatures, laying between 2000 and 2200K, vary little despite the use of different travel speeds, side jet gases (argon, helium and nitrogen), gas flows, impingement positions and side jet angles. Only with the side jet positioned horizontal (0 degrees) to the workpiece, a systematic increase in the plume temperature to 2300-2400K could be noticed. These results are in line with the model developed by Lacroix  on the influence of particle plume scattering for Nd:YAG laser welding showing theoretical temperatures of around 3000K.
The thermodynamic equilibrium for the presented temperatures was compared with the Stefan-Boltzmann law. The radiant emittance of a blackbody or the radiant energy emitted per unit time and unit area is proportional to the fourth power of the temperature  :
is the total radiant power at all wavelengths ( W
the area of the radiating surface ( m 2
the temperature ( K
), and σ the Stefan-Boltzmann constant ( Wm -2K -4
The plasma emerging from the keyhole can be modelled as a cone, but if observed horizontally, can then be approximated to a trapezoid. If the plume of vapour has a height of 50mm, a base of 0.6mm (laser spot size) and an aperture of 20° (approximated from the high speed images of the plume), the trapezoidal area is ~ 9.4x10 -4m 2. For a plume temperature of ~2000K the radiant power is just therefore over 1000W. This could be considered a high value of energy being radiated from the plume and it is highly sensitive to the temperature. If the same method is applied to the plasma temperature results when CO 2 laser welding, with the radiating surface an approximate ellipsoid of ~2x10 -4m 2 with an average plasma temperature of ~6000K, then the resultant radiant power, at ~1500W, is slightly higher than the Nd:YAG result.
Evolution of plasma/plume temperature with time
From the temporal distribution of the spectral readings, it was possible to follow the temporal evolution of the plasma/plume temperature. For CO 2
laser welding under air conditions with helium side jet the temperature evolution is shown in Figure 5
Fig. 5. Temporal evolution of the plasma temperature with He side jet (20l/min flow rate) and several speeds
From the Nd:YAG laser welding spectra fittings to the blackbody curves, the maximum plume temperature variance with time is between 25 and 100K, although the usual variance was ~50K.
Calculation of the plasma temperature when CO 2 laser welding is very sensitive to the chosen atomic lines. On this results a range of temperatures, varying by ~3000K could exist, depending on the chosen spectral lines. If the uncertainty in spectral line data (between 10 and 25%) is taken into account, then the variance in calculated plasma temperature values is even greater. All these uncertainties give rise to some concerns about the temperatures published in the literature and the degree of accuracy that can be achieved through spectroscopic measurements. For example by using the Fe I lines employed by Bermejo et al.  (at 442.731 and 442.257nm) then the calculated plasma temperature for air in the chamber with helium side jet and a speed of 1m/min is ~6000K, very similar to the results obtained by Bermejo  (~6200K). Further to this, Szymanski et al.  argued that low values of plasma temperature reported in the literature (between 5500 and 8000K) are due to the use of the atomic lines of the iron metal (Fe I). They argued that for more exact calculations the ratio between ionic and atomic lines should be used, since the emission coefficients for each line (atomic or ionic) reach a maximum at a predetermined temperature, which is usually higher for the ionic lines. Szymanski  also showed that by taking the intensities of the atomic lines, only the peripheral temperatures of the plasma could be calculated, and not the plasma core temperatures. Unfortunately, it was very difficult to identify the ionic lines (Fe II) in the recorded spectra readings. Nevertheless, the results obtained from the experiments reported here agree with the results obtained by other researchers in the field (see review by Duley  , pp171).
The evolution of the plasma temperature over time could provide more information about the effect of plasma on weld penetration (see for example plasma temperature behaviour at 0.75m/min in Figure 5) when CO 2 laser welding. Two periodical movements of the plasma could be distinguished from the temporal distribution of the plasma temperature and the high-speed video analysis. One of these movements corresponds to the plasma being created or emerged, and tilting from the keyhole (with a period of less than 5milliseconds, being quite stable in helium gas and more turbulent in argon). The other corresponds to the plasma detaching from the keyhole surface (only with argon gas), and is strongly dependent on the welding speed (i.e. the interaction time between the laser beam and the plasma).
The periodicity at which the plasma emerges and detaches from the weld pool and the frequency of the weld pool closing and opening must be related in some way. Probably when the plasma detaches from the weld pool, the keyhole no longer receives enough energy to remain open,  and thus collapses. The collapse may trap a region of plasma vapour at high temperature, thus rapidly heating and vaporising the surroundings.  The generated gas expands in both directions along the beam propagation axis causing changes in pressure along the keyhole. This may be responsible for increasing the upper part of the weld width, and producing changes to the penetration seen in the longitudinal section of the weld shown in Figure 6. With some spectroscopic readings, it was possible to see increases in intensity of the lines shortly after a detachment, indicating that ionised vapour emerged from the weld pool. For argon side jet welds, cross-sections of the welds show uneven shapes of the top of the weld pool, denoting turbulent weld pool dynamics.
Fig. 6. L-section of CO 2 laser welds with Argon (top) and Helium (bottom) side jets (each unit represents 1mm)
An explanation for the shape of the spectra emitted when Nd:YAG laser welding is not clear, specially the intensity decrease after 550nm. It could be argued that other physical parameters could affect the spectrum intensity, thus the spectrum differs from the Planck's blackbody radiation formula. The CCD detector, the intensifier tube and the mirrors of the spectrometer also have different quantum efficiencies over a range of wavelengths and their response at lower and higher wavelengths could alter the results. Unfortunately a full calibration of the spectrometer employed was not available.
Depending on the different gas conditions, when Nd:YAG welding completely different plume behaviours can be observed in the high-speed videos. It is remarkable, therefore that such a stable temperature is observed in the plume when Nd:YAG laser welding. It can also be observed that only under very high flows does the plume completely disappear, mainly due to the disruption of the weld pool bringing it close to cutting parameters. For all the other conditions the visible plume can only deviate from its path by the blowing force of the side jet gas. Unlike CO 2 laser welding, only very small variations in penetration depth can be observed in the longitudinal sections.
The characterisation and the effects of the vapour emerging from the keyhole under cw CO 2
and Nd:YAG laser welding of mild steel have been evaluated and the results obtained have allowed the following conclusion to be made:
- A partially ionised plasma with a temperature between 7000 and 11000K can be expected under cw CO 2 laser welding, with its shape characteristics changing depending on the gas employed as a plasma control mechanism. As widely recognised in the literature, helium proved the most effective gas to achieve a higher penetration weld and narrower heat affected zones.
- A plume consisting of a high-temperature (over 2000K) thermally excited gas can be expected for cw Nd:YAG laser welding, with its shape characteristics changing depending on the gas used as a control mechanism for the plume. At low welding speeds, argon proved the most effective gas to achieve higher penetration welds, perhaps due to its higher momentum effect when interacting with the plume.
- It has been demonstrated that the ionisation potential of the gas side jet in Nd:YAG laser welding does not have the same importance as in CO 2 laser welding, although other properties of the gas could play a role in the interaction with the plume.
- A better understanding of the interaction process between the gas side jet and the vapour being ejected from the keyhole is required to achieve higher penetration and narrower heat affected zones in deep penetration welds made with high power Nd:YAG lasers.
The TWI installations and the dedication of the staff of the Laser Department made this work possible. The authors would like to thank Mr D Russell for his laser welding expertise, and Mr R Lombardi and Mr C Crouchman for their help in the experiments and the construction of part of the equipment. This research was carried under EPSRC award No. 99313089.
- Steen, W. M. Laser Material Processing, Springer-Verlag, London, 1998.
- Russell D. Personal Communication, Former Head of EB and Laser Departments at TWI, 2000.
- Dawes, C.J., S.C. Kennedy, R.C. Crafer. Shielding Gases for CO 2 Laser Welding - An Experimental Study on Lap Welds in High Strength Low Alloy Steel. TWI Technology Briefing 398 - July 1989.
- Matsunawa, A., J.D. Kim, T. Takemoto and S. Katayama. Spectroscopic Studies on Laser Induced Plume of Aluminum Alloys. Proceedings of ICALEO'95, 1995.
- Lacroix, D., G. Jeandel 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.
- Young, M. Optics and Lasers, Springer-Verlag, Berlin, 1984.
- Dowden, J., P. Kapadia, N. Postacioglu. An Analysis of the Laser-plasma Interaction in Laser Keyhole Welding. J. Phys. D: Appl. Phys., 22:741, 1989.
- Poueyo-Verwaerde, A., R. Fabbro, G. Deshors, A.M. de Frutos and J.M. Orza. Experimental Study of Laser-Induced Plasma in Welding Conditions with Continuos CO 2 Laser. J. Appl. Phys. 74:5773, 1993.
- Szymanski Z., J. Kurzyna and W. Kalita. The Spectroscopy of the Plasma Plume Induced during Laser Welding of Stainless Steel and Titanium. J. Appl. Phys. 30:3153, 1997.
- Duley, W.W. Laser Welding, John Wiley & Sons, Toronto and New York, 1999.
- Griem, H.R. Plasma Spectroscopy, McGraw-Hill, New York, London, 1964.
- Mueller, R.E., Gu, H., Ferguson, N., Ogmen, M. and W.W. Duley. Real Time Optical Spectral Monitoring of Laser Welding Plumes. ICALEO'99, 1999.
- Hecht, E. Optics, Addison-Wesley, Reading, Mass., 1998.
- Chen S.B., Y.J. Lou, L. Wu and D.B. Zhao. Intelligent Methodology for Sensing, Modeling and Control of Pulsed GTAW: Part I - Bead-on-plate Welding. Weld. J. 79(6):151S-163S, JUN 2000.
- Fuhr, J.R., G.A. Martin and W.L. Wiese. Atomic Transition Probabilities. J. Phys. Chem. Ref. Data, Vol. 17, Suppl.4, 1988.
- Sears, F.W. Principles of Physics III: Optics, Addison-Wesley Press, Cambridge, Mass., 1946.
- Bermejo, D., R. Fabbro, L. Sabatier, L. Leprince, J.M. Orza; 'Spectroscopic Studies of Iron Plasma Induced by Continuous High Power CO 2 Laser'; SPIE 1279:118, 1990.
- Beyer, E., K. Behler, G. Herziger. Plasma Absorption Effects in Welding with CO 2 Lasers. SPIE 1020:84, 1988.
Meet the authors
José Greses is a second-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).