Kix,
Sirs,
first off, what a question, what a thread! So brilliantly answers and, to agree with Allan, this makes true fun and it shows me, there is something more beyond the horizon than that what the job brings all day (at least currently). A warm and heartfelt "Cheers" for bringing me on other thoughts with this great discussion!
Well, as Lawrence once again hit the nail on the head: "No physics, just rubber meets the road!" Physics whereas does only make sure that it - hopefully - gets the grip :-) (100% no offense Lawrence - Far from it!). Due to everything has been said by you great gentlemen on this topic, the only thing remaining to me is to congratulate you as well, Kix, for asking this excellent question.
I am humble certain that nearly everybody who is processing GMAW on aluminum and its alloys, and who is additionally using Magnesium alloyed fillers, comes to the point where he asks: "Where does this smut come from and why is its amount differing with the travel angle?" So did I - I must admit - a little longer ago. The problem that time was, that all the people I have spoken with, had different explanations, what stimulated me again to have a closer look upon the background of this phenomenon. As I said, what I could read and learn, was extensively and much of that I have - unfortunately - forgotten again. But I am honest when I admit, that I have attentively watched the progression in this field, since you know, the instruments for measuring physical phenomena are getting continuously better and more precisely with the time, and the more precise they are, the more mother nature let us having a look "behind the curtain".
Before I actually start, please let me honestly admit. This time I have no answer an your question, why a "pushed" bead is getting "...nice silver colored..." and a "dragged" bead "...turns brown and smoky...".
But... I have some ideas which I have considered round and round and when I finally conclude this post, I would like to ask that you all may be the jury judging what kind of coherences can cooperatively work on the phenomenon you, Kix, have described.
However, basically what there has been stated by now in this thread is reminding me on a composition of Wolfgang Amadeus Mozart. Consisting of many different notes linked together for yielding finally a perfect harmony. And everything what has been stated is exact and correct, each statement for itself and every single one indispensable!
There were the "mechanical" fundaments (excellent comparison and clarification - Allan), an extraordinary good and understandable explanation of the theory of fume and smut generation (brilliantly as always - Al) and last but not least, the inimitable description (founded on great personal experience) of the practical coherences to control the entire subject (simply unique... Lawrence). And - of course - not to forget... Dave Boyer, who has given an extremely good hint on what I would like to come to a bit later on.
Short and precise! A true pleasure to reading!
Now was my first consideration - just as always, I know - should I add something to this perfect "composition" and thus to venture the unforgivable risk to warp everything again? After having done that, I have decided to write a few words on your topic and I swear that I will hardly endeavor to not to destroy the fine and complex network of extremely good replies, stated by you great gentlemen.
To be with Allan
"(...)but I'm sure we'll make the rounds and get back to the aluminum at some point..."
let me speak of aluminum and its alloys again.
Meanwhile it has been more than once confirmed and is thus basically agreed, that the smut is being generated by the vaporization of Magnesium, or even other elements having low boiling points (e.g. Zinc). Mentioned by the way, it was found out, that the composition of the base material can have an influence on the smut's colour. This sounds reasonable, since the composition of the smut is influenced by the elements having been vaporized, and due to these elements have different optical properties, the optical properties of their condensates should show this well. SUGIYAMA has investigated that the smut founded basically on Magnesium is "black". The smut consisting of both Magnesium + Zinc whereas is more dark gray.
Well, before I would like to treat the "common" theories please let me report from one, which is completely different from all the other well-known ones. It has been suggested by Dr. Wolfgang Danzer - who is the head of the Laser Welding Department of LINDE Gas in Germany - at the ASTK 2004 which was an international Welding Colloquium in Aachen. He has surveyed the influence of shielding gases in welding (GMAW + LBW) of light metals and has in particular conducted investigations with respect to the composition of the black smut. Here the researchers wanted to having found out that the smut is nothing more than metal particles, having low volumes and masses and being projected from the areas the arc is working on. This result has - so Dr. Danzer - been proved by using High-Speed Videography and Scanning Electron Microscope- (SEM) and Microprobe Technique.
Dr. Danzer's theory implies that the "secondary" arcs (micro spots), often observable when welding aluminum and its alloys and been accepted for being responsible for the "Cleaning Area" beside the welding seam, have enough energy for vaporizing metal constituents - as Magnesium - and to project subsequently these metallic particles which condensate on the metal's surface. By investigating the "condensate" (smut) he has found out that there are areas of containing that high amounts of extremely small metal droplets, that these areas do no more reflect the light and thus appear to being "black" for the viewer. When using the technologies as mentioned above (microprobe), he could find out that there is no kind of "Soot" or "Magnesium-Oxide" contained within the condensate. Please see also the attached "Secondary_Arc_Trace", and "Projected_Material".jpg's.
This, as I mentioned, only for trying to show, that the topic, initiated by Kix, is truly not an "old hat", but rather more likewise today a quite interesting field of investigations.
But now coming back to the "common" theories of what the composition of the smut in aluminum welding is, and which should show, that those ones are quite contrary to what Dr. Danzer et al have found out.
Firstly however, I would like to emphasize that the theory (and practice) of how the smut is being formed, is exactly conform to what Al has explained in an impressive clear and precise way. Magnesium - or even other "easy to vaporize" - elements are evaporated within the arc atmosphere.
But, how does this actually work?
Well, firstly one should consider that it is a wonderful - for me the most fascinating miracle at all - but very intricate matter, of how a droplet is being formed when a Gas-Shielded Metal Arc is first ignited. There is a long row of variables acting together for that a welder can use the process for joining metals or materials, respectively. This is, by the way the reason for me to humble take my hat off for all the fellows doing this day by day by "ruling" this unique and fascinating tool "arc".
To make it quite short. Basically the vaporization of elements in "Metal Vapor Arcs" is necessary for providing an important share to forming the plasma. The metal vapors - as a constituent of the plasma - affect e.g. the electrical conductivity, thus the current density, by that again the temperature distribution of the arc and hereby again the droplet growth (in GMAW mainly at the anode) and penetration profile (in GMAW mainly at the cathode) at the work in a tremendous way. Most of them - although interesting but... - you know what now would come - should be not further treated hereinafter. Rather the most important thing that should be interesting for us, with respect to the subject of smut generated by metal vaporization, is the droplet growth on the one hand, and the in general heat content of both the droplet and the molten pool on the other hand.
What does this mean?
As well-known there are different modes of droplet transfer. Some of them were very precisely and vividly explained by Lawrence. The different characters of these modes affect as well the droplet growth. The larger the size the droplet grows to, the larger is its volume and... the larger its surface. This again increases the amount of vaporized elements, or metal, respectively, since the heat distribution over the surface is enhanced, or, the energy density drops. This means, the larger the drop, the higher its heat content and the larger the metal vaporization. And now it comes, different metallic materials do have different thermodynamical properties what means, that there could be elements (e.g. as alloying elements), having a lower boiling point etc.
What does this now mean for the droplet at the tip of a wire electrode?
Well, when having Aluminum and Magnesium it happens exactly that what Al has described. The lower boiling point of Magnesium compared with Aluminum makes sure that the amount of its expansion is drastically enhanced when the boiling point temperature is obtained. Therefore let us have a short look upon a very special thermal property of Aluminum and Magnesium - the boiling point. Now one could say Aluminum and Magnesium isn't equal to Aluminum and Magnesium, and he were probably right, since little variations in the chemical composition can change the thermal coherences partly drastically. Therefore we should first decide what kind of material composition we are talking about hereinafter. Since we have already heard by Al, that Magnesium is the main element making sure that smut is generated, we should choose a filler been already named by Allan. Thus I would like to propose to talk about the properties of:
- Pure Aluminum - Boiling point: 2793 K (=2519,85 °C = 4076,06 F)
- Pure Magnesium - Boiling point: 1368 K (= 1094,85°C = 2002,73 F)
- Alloy 5356 filler wire (AlMg-alloy ~ 5 % Mg) - Boiling point: ~ 1833 K (~ 1560°C ~ 2840F)
What can be seen already is that the Mg-alloyed wire has a boiling point laying under that one of pure aluminum but higher that one of pure magnesium. However, what one has to consider is, that the wire contains ~ 5 weight percent of Magnesium and this single element again has a quite low boiling point.
Now one sees very quickly that the values mentioned above have actually no "true" value, since we do not know - as already indicated above - what the fillers droplet temperature or the molten bead temperature actually is. Because it is confirmed by theoretical and practical investigations that the droplet temperature in GMAW is in general higher that one of the weld pool, I would like to propose to emphasize and treat hereinafter only the droplet temperature, which simplifies the entire subject substantially.
In other words, as long as we do not know the average droplet temperature, as long we have no chance to find out what the average amount of vaporized magnesium; and thus the average amount of the to be expected amount of smut is. This is easy to understand. Let's therefore consider, that the droplet temperature may be quite close to the boiling point temperature of the element being responsible for the smut, namely magnesium. Then we could probably expect that the amount of smut is "low" compared with a larger droplet having a higher heat content, since the amount of vaporization rises with the heat content of the droplet, and the heat content of the droplet increases again with the mass or volume of the droplet, respectively. What furthermore is important when speaking of vaporization is the increase of partial pressure of the vaporized material. When obtaining the boiling point, the metal vapor increases in volume and thus its appropriate partial pressure within the droplet is rapidly increased. This again makes sure that the droplet "explodes" formally. Spatter on the one hand and arc or process instabilities (turbulences etc.) are the result.
Moreover we should consider, that the droplet heat content is - in second order - a main factor for the weld pool temperature, since the droplets transfer their heat content into the weld pool when being absorbed by the bead. Always under the consideration that we are talking of, and treating the Gas Shielded Metal Arc Welding.
Now it's becoming a bit complicated.
Why..?
Well, for finding out the droplet temperature, first one has to measure the average heat content of a mass unit of droplet, consisting of the specific material one wants to survey, in our case, the filler wire Alloy 5356. This, to make it short, is normally accomplished by using an action called "calorimetry". I guess that the physical principle of a calorimeter is well-known, therefore I won't treat it further herein. What I would like to mention however is, that it is a quite tricky matter of realization. For measuring the droplet's heat content there are being used specific constructions of calorimeters, taking into account the specific peculiarities of the welding process. As a significant detail of these designs I would like to mention, that for melting the wire is a "torch combination" of GMAW- and GTAW-torch used. The average droplet heat content can then be calculated by measuring the mass of molten filler wire (depending to wire diameter and wire feed speed used) in relation to the heat quantity being received by the calorimeter. As I said, the calculation is intricate, since one has to consider that it's no pure metal (having very specific properties) but an alloy whose properties should be calculated. Avoiding the tricky details of the measures to be accomplished before the heat content of a filler metal droplet can be stated, I would rather like to state subsequently the height of an Alloy 5356 filler* wire having the following composition:
- .060 % Si (Silicon)
- .140 % Fe (Iron)
- .010 % Cu (Copper)
- .100 % Mn (Manganese)
- 4.960 % Mg (Magnesium)
- .100 % Cr (Chromium)
- .010 % Zn (Zinc)
- .100 % Ti (Titanium)
- Balance: Aluminum
(*) measured for GMA-Pulsed Welding
The average droplet temperature for this wire is converted at 1634 K (= 1360.85°C = 2481,53 F). As one can see by comparing the droplet temperature with the boiling point of the wire, it is close to that temperature, but not as high as it. But nonetheless it is higher than the boiling point temperature of pure magnesium (1368 K). Hereby one can a definitive amount of magnesium expect to be vaporized.
Well, by now you can ask or tell me:
"Stephan, why are you telling all this? Is it really so important to know about the droplet temperatures and all these details? Isn't it sufficient to know, that the magnesium is vaporized and thus the smut is generated by that? Why is it so important to be informed about the droplet's heat content? We know that the droplet does transfer its heat content into the molten pool and thus we achieve a weld pool! This means no news to us!"
O.k. o.k.! May be you are right - certainly you are! Perhaps I am on a wrong path again by considering all these little details, but you know? I like "Columbo". I am like him. All these little details have to match one to each other, before I can "sleep again" and the case has been resolved completely :-)! Hopefully I am able to explain finally why I am talking about droplet temperatures etc. for requesting you may feel pity for me.
Resume..?
Let me continue where we have stopped. Let us say that with a specific value of wire feed speed ( + wire diameter + base material's thickness + the electrical power supply parameters + welding speed + ...) we are adjusting a specific value of melting rate = weld metal deposit = weld pool volume = "heat input (what ever it is)" = heat content = ??? -- > SMUT. Shortly, we can adjust a specific amount of generated energy being usable as a variable for estimating the amount of vaporized elements - magnesium in this case.
Then I ask: "When we have adjusted all the named and unnamed variables above, for achieving this specific amount of energy which is responsible again for the specific share of magnesium being vaporized and (mainly) forming smut, why - then my question - is the amount of smut different when changing the torch travel angle?"
Hmmm, I have considered this and please try to follow my - perhaps - confused ideas...
Firstly let me conclude by here.
What can be said, is that the droplet's size can be used as a variable for the droplet's mass or volume, respectively. It has been furthermore confirmed, that the droplet's volume under given thermal conditions is a variable for its heat content. The heat content or the droplet's temperature respectively, are again a gauge for the amount of easy to vaporize elements (e.g. magnesium). This means, the larger the volume of the droplet - under given and constant conditions - the larger its heat content, the larger the amount of magnesium vapor and the higher the inner magnesium partial pressure and thus again the risk for obtaining spatter, turbulences and other negative effects. A sequence of spatter creation by drastically increased magnesium vapor partial pressure within a molten droplet in using an A 5356 aluminum wire electrode (diameter 1.0 mm) can be seen on the attached Partial_Pressure.jpg. However, for a specific amount of energy we achieve a specific amount of vaporization. Hence, the higher the amount of energy, the higher the amount of smut. So far...
Now it comes. Kix, you are talking about both conventional arc modes "Spray" and "Short Circuit". Both arc modes have - of course - different arc performance ranges. "High" performance (energy) for the first, "low" performance for the latter. Likewise can be said, that the performance can be changed in - within appropriate limits - by varying the values of arc voltage - as outstandingly good explained by Lawrence. And it is both practically and theoretically confirmed that Lawrence hit the nail with his explanations, and that the amount of fumes a/o smut has to do with the height of chosen arc voltage under other constant circumstances.
However, my question is going into the direction of your question, Kix.
"Does this explain finally why the welding seam looks "shiny silver" when the travel angle is "forehand" and "smoky" when it is chosen to be "backhand"?"
Hmmm...
And now I would like to come back to all the statements and posts by all the great gentlemen who have given them for clarifying this subject. You remember the W.A. Mozart "composition"?
I have started by considering what Allan has described when he said:
Quote
"... When you're out hosing off the driveway and you use the sprayer and point it in the direction of the area that you're trying to clean starting close to you and working away from you, the spray pushes most everything out ahead of it(this might somewhat describe the way the shielding gas acts when using a push angle). On the other hand if you start with the spray out a ways in front of you and bring it back towards where you are standing it doesn't do quite the same job of cleaning things(this might somewhat describe the gases action of coverage while utilizing a pulling gun angle). ..."
Unquote.
Then I went on by considering what Dave has stated:
Quote:
"...the push angle gets the argon out in front of the weld puddle so it can do it's cleaning action, and the material is clean by the time it melts. This won't happen with a drag angle..."
Unquote.
After this, I have thought about what Lawrence has posted and talking about the - please allow Lawrence - physical coherences of arc voltage and fume a/o smut generation of easy vaporizable elements (even magnesium in aluminum and others in steel welding).
Finally then I arrived at Al :-). His brilliantly explanation of the inner mechanisms of magnesium vaporization is the final result of what you, Kix, can find on or beside your welded aluminum seam!
Al these explanations have pointed us in the right direction, since all of them - as far as I am allowed to humble repeat that - are right and each of them provides a share for the complete answering of your question!
Now we could stop here... Could we really..? Hmmm...
So did I honestly think first. Since it is surely as I said. Everything has been said on this and so let it rest in peace... But you know me. There was only one very little detail which hasn't let me rest in peace.
And this was? Yes! It was the question: "Why is the "pushed" seam "shiny" and the "dragged" seam "smoky"?" Which of the posted answers you greatly appreciated fellows have stated, is responsible for this phenomenon?
And now I am finally coming to my very own considerations on this question and hopefully you will see, why I have firstly talked about droplet heat contents etc.
What I did first was that I have captured the possibility of executing a little practical welding attempt when I was in the training center of my company.
The following materials and adjustments were taken:
- Base material: AA 1350 (untreated = not wire brushed and not preheated)
- Thickness: 8.0 mm
- Manually Gas Shielded Metal Arc Welding
- Spray Mode
- Filler wire: A 5356
- Wire diameter: 1.2 mm
- Shielding gas: Argon 4.6
- Gas Flow Rate: 15 liter/min
- Wire feed speed: 11 m/min
- Travel angle: ~ -10° (dragging) / ~ 10° (pushing)
Under having a look upon the attached Top_View_Comparison.jpg one can recognize the phenomenon - mentioned by Kix - that the pushed welding seam shows a "smoky" (= smut) appearance beside the welding seam whereas the seam itself does have a (more or less) shiny appearance. The dragged seam whereas - accomplished by using equal conditions - shows both smut beside + on the seam surface itself. Just exactly as Kix has stated it.
Where could this come from..?
Now it's time to coming back on the relation or dependence respectively, of heat content of the molten droplet and weld pool and the amount of magnesium vaporization. I have considered further, that it is well-known - and by the way being taught - that by changing the torch travel angle the geometry of the welded seam as well as the depth of penetration or fusion respectively, do change. Likewise the width of the seam is being taught by the theory, should change by changing the torch travel angle. The width of a pushed seam should thereby be wider as a seam being created by using a negative or dragging travel angle. Therefore I have asked an appreciated colleague of mine, being also with my company, to be so kind and prepare two etches macro sections of the welded aluminum sheets to validate this behavior. Fortunately he has done, as to be seen on the attached Pushing_10°.- and Dragging_-10°.jpg's. I have measured out the macro sections for evaluating the fusion depth, width and height of the welded seams. Although this can not be representative due to a lack of numbers of samples (single probe) one can nonetheless see, that the pushing angle shows another, more shallow, penetration compared with the dragging travel angle. Also the height is less in pushing angle whereas the width of both pushing and dragging is quite comparable. O.k. this would slightly prove what the theory says, as mentioned above.
But what have these considerations to do with the phenomenon of a significant increased amount of smut on a seam surface when using a dragging travel angle?
Let me try to explain...
Let us imagine that we are using a travel angle of 0°, i.e. we are welding under processing the torch perpendicular to the work. Then the amount of energy - as mentioned above - been fixed for a specific combination of physical factors and sub items, should be equal to "1". Let us assume further that the workpiece area, being covered by the arc, when using a travel angle of 0° is approximately circular. This again means that the energy distribution when only considering the entire anode spot and neglecting the secondary "micro spots" (see Secondary_Arc_Trace.jpg), is approximately uniform. This means, each single area element unit within the covered arc area may be considered to have the similar energy input into the work piece or base material or weld pool, respectively. This again means, that the heat content of the likewise to be expected uniform weld pool (fusion zone) has a defined and specific value. This again means, that the value of evaporation of magnesium has an amount even specific to the chosen conditions. This yields likewise a specific amount of smut. Even specific to the travel angle of 0°. I.e. neither pushing nor dragging.
Coming now to negative (dragging) and positive travel angles (pushing) and thus as well to Dave's and Allan's statements.
Beginning with the pushing travel angle, e.g. 10°. Here - as Allan has so fabulously described - the energy of the arc has another distribution, compared with a travel angle of 0°, which has been considered to having a uniform circular distribution. Due to the arc plasma has however a specific "stiffness" when operated with higher performances, the energy is coupled into the weld pool unevenly. I have attached for a better understanding and visualization the Arc_Stiffness.jpg. Here one can - at least a bit -recognize that the amount of transferred arc energy is reduced and a specific share of the total energy balance is "lost" by radiation losses and different other mechanisms, e.g. convection. At all, the value of a pushing angle is different to "1".
A larger share - and here are Dave and Allan - of the arc energy interaction with the workpiece is being used for cleaning the surface (cleaning action) in front of arc by reducing - in particular when welding aluminum and its alloys - the oxides which are - nonetheless - very important since they stabilize the arc.
And now comes my first "hypothesis" you all may have to judge.
When the arc - when using a positive, i.e. pushing travel angle - has cleaned the surface of the work by the oxides, these constituents - consisting of hydrogen saturated aluminum oxide and other negative pollutions - were no more available to react with the vaporized amount of magnesium and thus the area of the welding seam, i.e. the areas where the arc has accomplished its cleaning action, appears to be shiny.
My second "hypothesis" whereas - and this is indirectly going into the direction which I would like to explain when I would like to treat the dragging angle - is the following one. By the changed distribution of transferred energy (heat) into the weld pool (losses in the ahead direction of the arc, see Arc_Stiffness.jpg), when using a pushing angle, the total heat content of the weld pool is that far reduced (by n/1), that the amount of this difference is similar to the amount of reducing the amount of vaporized magnesium and thus the amount of smut - generated mainly by vaporized magnesium - is reduced for the weld pool's volume and thus... the finally created weld seam. O.k. so far the pushing angle...
Coming now to the opposite direction, the negative or dragging travel angle, respectively.
First "hypothesis" for dragging.
When dragging the torch, the relations are inversed. Now we can assume that the cleaning action is not being accomplished by the arc, since its main energy share is directed upon the molten weld pool. Thus each region which is subsequently covered by the burning arc is still "polluted" with aluminum oxide which is again polluted with humidity, residues of separating agents, etc. These again do impetuously react with the metal or magnesium vapors respectively, coming from the droplets and weld pool for finally creating... smut!
Second "hypothesis" for the "smoky" seam surface appearance when using a dragging travel angle.
As having mentioned, the dragging travel angle creates a significant different weld pool geometry compared with the pushing angle. This means, also here is the value different to "1". Let us assume again that the distribution of thermal energy when dragging the torch is opposite to the pushing angle and thus the amount been lost when pushing the torch must now - on the contrary - be added (by n/1) to the weld pool's heat content. Hence the weld pool's average temperature is increased by the additional heat share. Hereby, I hope somebody can follow me, the amount of magnesium vaporization is... increased as well. Or isn't it?
In other words, by increasing the amount of the weld pool's heat content (the main arc energy share is being transferred onto the weld pool) its temperature is rise, thereby the amount of vaporization does change and thus the amount of smut - being pushed and condensed on the seam's surface - is increased, compared with the pushing angle. The surface of the weld seam appears "black".
Oh! I guess I am almost finished as I must recognize :-).
Before I come to a little conclusion of what I have considered and written down previously, please let me very briefly treat the question of Kix with respect to the composition of the smut. And thus coming back to what Al has stated in his fine explanation.
TONG et al have specifically investigated this subject and they have found out by using X-Ray Diffraction. They have - by the way on the contrary to Dr. Danzer - found out that the composition of the smut is exactly as Al said, consisting of aluminum oxide + magnesium oxide + metallic aluminum.
And - to confirm finally the statements of Lawrence - the shares of the constituents do significantly depend on the arc voltage being used. And the reason therefore has exactly been explained by Lawrence and nothing else more is needed to add on these explanations.
O.k. let me conclude my considerations - I hope you have read all the written by here...
- The generation of "smut" in GMA aluminum welding is founded on easy vaporizable elements. Mainly however Magnesium.
- The main share of the vaporized elements has its origin in the filler material, since the droplet temperature is considerably higher than the weld pool temperature.
- There were different approaches feasible and - as been proposed - to be judged by you all.
Basically these - personal -approaches can be separated in two different "classifications":
- The first is mainly founded on the mechanisms treating the cleaning action of the DCEP arc in its interaction with the surface oxides.
- The second approach treats the different arc energy distributions existing when changing the torch travel angle from 0° (perpendicular - energy distribution = "1") to either -10° (dragging - energy distribution = different to "1") or 10° (pushing - energy distribution = different to "1"). Hereby the weld pool's heat content and thus its temperature is increased or reduced which might lead to different magnesium vapor partial pressures within the weld pool a/o different amounts of magnesium vapor amounts in general and thus different amounts of smut.
Well, now I'm really at the end and would like to finish this post.
What remains, is only to say a hearty "Thank you" to all of you great fellows and friends.
Again you were the ones who have given me a chance to "stretch" my head a bit by busying myself with this fantastic topic.
I remain with best and honest regards to you all,
Stephan