Two arguments could be the followings: one is when the keyhole stability is lost, this causes pressure alterations, which in turn changes the surrounding fluid flow and final weld geometry; another one is that at low powers you can't create enough pressure to generate the keyhole and you have the simple convection mode, which ends up in non-thick joined configurations.

So you need enough power to go through all the required thickness, but it's actually the pressure that keeps the stability of the process. You can check that by applying a higher power laser beam welding on a rather thinner sectioned materials, you will noticed the fluctuations of the keyhole, when the beam exits the bottom. When the beam exits the bottom you will also notice considerable amount of material ejected via shear at the bottom.

I am very upset by your statement that the loss of stability of the keyhole causes a change in pressure, since it is equivalent to the belief that the wind is blowing from the fact that the leaves are stirring.

I'm in shock.

Dear Rinat Seidgazov,

Years ago we did experiments to analyze the composition of the vapor that came out of the keyhole. We condensed a portion of the metal vapors inside a both end open quartz tube held co-axial with the laser beam. The relative amounts of the elements in the condensate gave us the temperature inside the keyhole ( X. He, P. Fuerschbach and T. DebRoy, Probing Temperature during Laser Spot Welding from Vapor Composition and Modeling, Journal of Applied Physics, 2003, vol. 94(10), pp. 6949-6958). The temperatures gave us the sum total of vapor pressures of all alloying elements inside small keyholes. The values of the pressures were higher than the ambient pressure for spot welding.

We also did similar experiments much earlier with moving laser beam ( P. A. A. Khan and T. DebRoy, Alloying Element Vaporization and Weld Pool Temperature During Laser Welding of AISI 202 Stainless Steel, Metallurgical Transactions, B, 1984, vol. 15, pp. 641-644) and the sum total of the partial pressures was higher than the ambient pressure for linear welding.

Does the higher pressure "displace liquid metal"? My understanding is both yes and no. Please allow me to give an example where the higher pressure does displace liquid. Depending on the power density and the interaction time, a laser beam can generate sufficient recoil force to displace liquid metal. The most common proof is the ejection of metal droplets from the liquid pool when the recoil force exceeds the surface tension force ( X. He, J. T. Norris, P. W. Fuerschbach, and T. DebRoy, Liquid Metal Expulsion during Laser Spot Welding of 304 Stainless Steel, Journal of Physics D: Applied Physics, 2006, vol. 39 (3), pp. 525-534). So my "yes" is conditional - need to have a combination of power density and interaction time that exceeds a critical value. My "no" is also not absolute - please allow me to explain why I say so. The primary driving force for the convective transport of liquid metal within the liquid pool is Marangoni stress resulting from the variation of surface tension with distance. Since temperature changes with distance from the axis of the laser beam on the liquid surface, there is a spatial gradient of interfacial tension which is a stress known as the Marangoni stress and this stress drives a strong recirculation of liquid metal within the molten metal (S. A. David and T. DebRoy, Current Issues and Problems in Welding Science, Science, 1992, vol. 257, pp. 497-502). So, the displacement or convection of liquid metal is not caused by the vapor pressures to a large extent. There is however a secondary effect of vapors coming out of the keyhole that contributes in a smaller way to the convection. The vapors exert a sheer stress on the surface of the keyhole and this shear stress modified the convection driven by Marangoni force. However, this effect is not very strong.

I wish you well in your search for an answer.

Tarasankar DebRoy

University Park, Pennsylvania.

In my study, the width of tracks changes little when under conduction mode, however when under transition mode and keyhole mode, the width increases rapidly. I think recoil pressure may play a more important role in displacing melt compared to Marangoni effect.

Dear prof. Debroy, many thanks for your opinion. Sorry for the delay with the reply. It took time for his preparation.

As you noticed in your publications of the 90s, "experimental data indicate that the vaporization rate under most welding conditions is five to ten times lower that the rate predicted by the Langmuir equation” (T. DebRoy, David Physical processes in fusion welding, Rev. Mod. Phys., 1995, Vol. 67, No. 1). You explained such a discrepancy by the imperfection of the classical theory of evaporation with reference to the description of the welding process. I can agree with this because all the formulas of the classical theory of evaporation are obtained for equilibrium conditions althought the laser welding conditions are strongly nonequilibrium. Therefore, the application of the classical theory of evaporation for the description of laser welding is a risky intrusion into the poorly studied field of physics when the researcher can not be sure on reliability of the calculated results. Unfortunately this did not have the proper effect on further research. Alas, this often happens. The discrepancy you found is a warning against the absolutization of conclusions based only on calculated data without reliable experimental confirmation. It should be done the conclusion that to understand the complex hydrodynamics of the welding process, it is preferable to rely on direct measurements. I also do not understand why the alternative hydrodynamic mechanism, not related with an evaporation and melt displacement by the vapor pressure, is not considered by you as another possible reason for the discrepancy you have established. For example, a thermocapillary effect with a shear flow structure that leads to a deep penetration effect, as it was shown in a number of publications (R. D. Seidgazov , Yu. M. Senatorov Thermocapillary mechanism of deep melting of materials by laser radiation . 1988 Sov. J. Quantum Electron. 18 396 , Seidgazov R. D. Thermocapillary mechanism of melt displacement during keyhole formation by the laser beam. J Physics D: Appl Phys, 2009, v.42, No 17 (175501), (7 pp)). It is impossible to ignore the thermocapillary mechism with share flow structure, given its high accuracy of reproduction the wide range of experimental data.

Argumenting your position, you indicate that the calculated vapor pressure is higher than the ambient pressure. I can not accept your optimism for the following reasons:

The data were obtained by calculation with using of some measured values on the base of assumption that the hydrodynamics of the welding process is completely determined by evaporation and vapor pressure. The possibility of an alternative hydrodynamic mechanism ( thermocapillary effect with a shear flow) is not considered. As a consequence, the calculated values of superfluous vapor pressure inside a keyhole are in the range of (1 ÷ 10) atm. Really, these values correspond to the dynamic pressure of the melt flow at a speed of 7,5 - 16 m/s, observed in the welding experiments of steel with laser beam power of 6-14 kW (I. Ericsson, J. Powell, A.F.H. Kaplan. Measurements of fluid flow on keyhole front during laser welding. Science and Technology of Welding & Joining. October 2011). However, the values of the pressure inside a keyhole obtained as a result of direct measurements are very small. They are in the range of (19 - 570) Pa when welding of Al-alloys by an electron beam with a power of 1.1 ÷ 3 , 85 kW (Bondarev А.А., Voropai N.М. Fizika i chimiya obrabotki materialov, 1974, No. 2, pp 50-55 (in Russian)) and in the range of (310 - 426) Pa when welding of armco- iron by an electron beam with a power of 7,2 kW and of 404 Pa for welding of titanium alloy (Verigin AM, Erochin AA, Shavyrin AN, Reznichenko VP Fizika i chimiya obrabotki materialov, 1980, No. 2, pp 145-146 (in Russian)). Such a low pressure in the penetration channel proves irrefutably that the real hydrodynamic mechanism of the welding process can not be associated with evaporation. For the sake of persuasiveness, I will give the values of the speed of the thermocapillary flow for the experimental conditions (Ericsson, etc., 2011) according to the thermocapillary model. These values are 6,3 m/s with a power of 6 kW (in experiments of 7,5 m/s), 8.1 m/s with a power of 10 kW (in experiments 10,77 m/s) and 9,6 m/s with a power of 14 kW (in the experiments of 16,24 m/s). As you can see, some differences are within the realistic picture of hydrodynamics. They increase with increasing of welding speed and, possibly, are associated with an increase of plasma influence through the electrocapillary effect. This has yet to be investigated if someone shows interest.

As another argument, you call the ejection of drops during welding, representing it as a result of vapor pressure action on the melt, when the force of vapor recoil pressure exceeds the force of surface tension. In this case, you also do not consider other possible physical mechanisms. But the ejection of droplets can be a consequence of thermocapillary acceleration of the melt with the shear flow structure. There occur a metal splash as a result of the collision of the thermocapillary flow with the solid boundary of the fused zone with the formation of an outlet on the edges with traces of detachment and ejection of the droplets. It was confirmed by comparing the calculated values of the melt flow rate using the thermocapillary model (12 m/s) and the measured velocity of the droplet ejection (13-16 m/s) Сейдгазов Р.Д., Низьев В. Г., Гофман В. Э. О механизме удаления расплава импульсом ТЕА СО2-лазера. Поверхность. 1992. №3. С. 18-21 (in Russian)). From the condition of detachment of the drop, an estimation of the surface tension in the neck of the drop was made . It turned out to be very small and amounted to 10 % of the maximum value at the melting point (i. e. the surface structure approaches the state of destruction). Such a drop in the surface tension value can not be explained only by thermocapillary effect and signals the existence of an additional cause for the decrease of surface tension. I think such an cause can be the electrocapillary effect, which is acted when liquid metal and plasma are in contact. The rationale for the possibility of an electrocapillary mechanism was discussed in my recent report with an analysis of many experimental observations accumulated over 50 years of investigations of deep penetration welding (R. D. Seidgazov Электрокапиллярная гидродинамика при сварке с глубоким проплавлением. Вторая международная конференция "Электронно-лучевая сварка и смежные технологии", Национальный исследовательский университет "МЭИ" 14-17 ноября 2017 года, Москва (in Russian)). The report shows that under some specific welding conditions (in a vacuum, under the action of an external electric field), the thermocapillary mechanism can be significantly enhanced with an appearance of electrocapillary effect which is comparable by value. Even the grounding of the sample can have a noticeable effect on the formation of the crater formed by laser action. The possibility of an electrocapillary mechanism directly follows from the results of your experiments in which a decrease in the surface tension of the molten metal in the presence of plasma was observed (P. Sahoo, T. Debroy. Interfacial Tension between Low Pressure Argon Plasma and Molten Copper and Iron. Metallurgical Transactions B, vol. 18B, 1987, 597-601 pp.). I would gladly have brought other arguments, but the RG chat format has limitations.

Dear prof. Rinat Seidgazov and prof. Tarasankar Debroy, how can we distinguish the effect of thermocapillary between vapor if they exist at the same time? Athough some research has shown that without vapor, the thermocapillary effect can deepen the melt pool. I'm still confused when using the same thermocapillary model under conditions with intense vapor, i.e. LPBF, since the melt velocities estimated by simple thermocapillary model and Bernoulli model are comparable.

Dear Wang Yafei. Please forgive me for the too long time delay in answering.

There is a counter question: is it necessary to do a calculation of the vapor pressure at all if it is possible to reliably determine that the thermocapillary effect is dominant under the conditions of welding? After all, no one comes to mind to take into account the light pressure of the laser beam.

It is necessary to pay attention to the following circumstances:

1. The general misconception about the determining influence of vapor is based mostly on the application of the classical theory of evaporation under the assumption that the evaporation process is equilibrium in laser welding. But it is clear that the process of laser welding is a highly non-equilibrium process, for describing which the formulas of the classical theory of evaporation are inapplicable.

For example, Professor Debroy mentions his experiment on the condensation of a portion of metal vapors inside quartz tubes held coaxially with a laser beam. But then, based on the relative amount of elements in the condensate, a calculation was made of the temperature and pressure of steam inside the keyhole (note - assuming equilibrium evaporation!). The pressure values ​​were higher than the ambient pressure for spot welding (note - this does not imply a conclusion about the dominant effect of vapors on the melt hydrodynamics).

2. All experimental data are in strong contradiction with the assumption of the dominant influence of vapors on the melt hydrodynamics. The discrepancy between the calculation and the experimental data ranges from 102 to 103 times (the vapor flow rate is 102 times, the vapor pressure is 103 times, the cost of beam power for evaporation is 102 times, the loss of mass for evaporation is 102 times). This allows us to state with confidence that the effect of vapors on the hydrodynamics of the melt is negligible.

3. The capillary model (including thermocapillary and electrocapillary effects) provides a very good correlation between experimental and calculated data.

If you take all this into account, then there is no need to take into account the effect of the vapor.