Vacuum Web Coating

Rotatable vs. planar magnetron sputtering burn-in

I have been wondering just how sensitive rotatable aluminium targets are to developing a oxide layer requiring burn in to break through to the metal as compared to planar targets.

In my case planars burn in almost instantly while there is a lot of technique applied by operators to get rotatable targets to finally burn through.

Answer

Removing the oxide is not only dependent upon the cathode design, magnetic arrangement and strength, power applied and sputtering pressure but is also dependent upon the history of the target.  The targets for the rotatables will have a different manufacturing history and so are likely to have a different oxide thickness to the planar targets.  It is probably more difficult to make the tubular shape and so more heat will have been used in the processing which will have increased the oxide thickness.  As the oxide sputters at a slower rate than the metal, the target surface area several times larger on the rotatable than the planar and the rotation of the target allowing time between sputtering for the surface to re-oxidize it can be expected for the rotatable to take more time to clean up compared to the planar.

If you want to try to minimize the burn-in time it may be worth mechanically cleaning up the target surface just before installing then into the vacuum system.  I know people have used abrasives to clean the surface to get to bright metal. This needs care as the dust can be a fire or an explosion hazard and as it generates dust it is important that the target is cleaned well afterwards to make sure the dust does not cause sealing problems or arcing on the target surface. The target will immediately oxidize but the oxide thickness will have been minimized.  I would expect a target precleaned in this way to clean up in a similar time (per unit area) to a planar for the same sputtering conditions.

An expensive way of investigating the target differences would to use one of the surface analysis techniques to determine oxide thickness of the different types of target.

Supplementary question

On the burn in issue for “new” targets I can concur on the variables you mentioned, what I can’t understand is that after the machine has produced a batch for several hours (and now targets are supposedly really clean) then the machine vented, the need for substantial burn in between batches is always present.  The targets just do not get to a high enough voltage to sputter until they finally “break thru.”  Again I do not see this with the planar aluminum…metallize for hours…vent…pump down and I can immediately hit them with full power and the voltage is high and they sputter…a couple minutes of burn and they are stable and ready again.

Answer

The burn in of the rotatable is no surprise even on a well established target.  Once the sputtering has stopped the oxide will build up. Nominally it will be the same thickness as on the planar but on the planar the target does not move and so the negative aspect of the planar which is the erosion profile that is often a 'V' shape becomes a benefit to burning in the aluminium. When the aluminium switches between oxide and metal it does not switch instantaneously across the whole racetrack but it breaks through where the magnetic field is strongest and parallel to the surface then as the metal suddenly sputters very quickly some will be backscattered to the target surface as well as the metal track widening and so it gives the appearance of breaking through across at least most of the racetrack.  The rotatable may have a similar magnetic design and strength but before the breakthrough takes place the target has rotated slightly and moved into a slower sputtering rate position and so the oxide is either removed at a slower rate or it may even begin to build up again.  The front surface of the target will be hot and so will rapidly oxidise. Thus for your rotatable it will take many more passes across the sputtering zone before the target breaks through.

I suspect if you were able to stop the rotation that you would be able to get breakthrough at a similar speed to the planar but obviously not around the whole cylindrical target.

Question - Seasonal variations – are they a problem?

I have a query regarding bond strength in metallized film. We have long rainy season and unfortunately we don’t have any humidity control system in our metallization hall, we are also producing PET chips & PET film at same premise but since past few days of rainy season we are continuously getting low bond strength in all different types of PET metallized film even in copolymer coated metallized for which we never get bond strength less than 600 grm force/inch currently getting 150~200 grm force, we have tried to increase solid content in copolymer coating we got some improvement and value raises to 300 grm force/inch & in other type of films like plain PET & corona treated PET we are constantly getting ~ 50 grm force/inch earlier we are getting around 110 grm force/inch. Can high humidity like 80% RH can play this big role or we need to look for some other causes?

Answer

Humidity has all sorts of effects throughout the process and so if the ONLY difference between the good bond strength and the low bond strength is the season and the increase in humidity then your answer is yes the humidity is the cause of the problem. You do need to be sure that this is the only difference.

If we look at some of the variations that are related to humidity we can see how it can affect the process.

PET chips will absorb moisture from the surrounding atmosphere and so in high humidity will contain more water than in dry conditions.  PET chip is usually dried before extrusion but it depends on what controls there are on the drying as to whether the PET chip enters the extruder with the same moisture content or is slightly wetter in the high humidity season.  If the chip is only dried for a set time there will be variations. If the drying is done using dry air and the output moisture content monitored and the chip always used once the moisture is below a set level the content ought to be similar irrespective of the season.  The moisture level in the PET chip will be reflected in the moisture content of the PET film produced as well as giving small variations in the polymer chemistry that can affect things like the oligomer level too. Thus the more consistent the input material the more consistent the film output can be expected to be.

As the film is wound on the production line it will trap the surrounding atmosphere between the layers and this will have the high or low humidity depending on the season. The PET will absorb the moisture and so in the high humidity season the moisture content will be higher.  If the film is corona treated there can be variations in the treatment level depending on the corona treatment settings and ambient humidity.  High humidity air is more conducting than low humidity air and so the corona treater, if it is left with the same settings, will produce different treatment levels at high humidity than for low humidity.  Again if this is known about and the settings are changed, to compensate for this conductivity difference, the treatments may be similar.

Between the film manufacture and corona treatment (if any) and the vacuum deposition process the film will be stored and during this time the film will also equilibrate with the surroundings and will either absorb or desorb moisture depending on the humidity levels.

Once the film reaches the vacuum system and it is pumped down there may be observable changes such as the base pressure is not as low as for film pumped down in the dry season. This may be difficult to check as there is also the factor of the stray deposition that absorbs moisture and so the base pressure can also be related to the system cleaning and quality of the cleaning. Without any film in the system if the system is pumped down it will reach a different base pressure if the system is clean from when the system has stray deposition coating shields and other surfaces.  The stray deposition is porous and so has a very high surface area and so will absorb much more moisture than the cleaned surfaces.  The difference in surface area can easily be a factor of ten or more.  Thus this may hide the fact that there is also a higher moisture load from the roll of film made in the high humidity season however both will be related to some extent as the both the absorbed moisture in the stray deposition coating and the roll of film will be significantly higher in the high humidity season than the dry season.

It is likely that to allow for variations in cleanliness and seasons you have a base pressure set that once the system is below this level it is OK to begin metallizing.  What you may see more easily is the time to reach this base pressure is longer in the high humidity season than in the dry season.

Once the process starts there will be a higher level of moisture given up as the trapped air layer is revealed as the web unwinds. This moist air will have the moisture captured by any cryosurfaces within the system but the background moisture level will be higher than in the dry season. This higher moisture level will affect any plasma cleaning process in as much that the moisture will be a proportion of the gases involved in the plasma thus if you think you have a particular ratio of argon to oxygen it will really be a ratio of argon to oxygen to water where the water content is dependent upon the season.  The water will be split to oxygen and hydrogen and so will change the argon to oxygen ratio.  Depending on how much of a change this is will depend on how much of a change it will make to the plasma treatment but if the plasma treatment has been optimised in one season it could well mean that it is no longer optimised in a different season.

If the moisture level of the polymer is not controlled during film making and so there is a variation in oligomer level in the PET film then this will be seen as a different level of oligomer on the surface which may also mean the plasma treatment needs to be modified to take account of the different level of oligomer.  The worst case scenario is if some of these changes are cumulative such as if in the high humidity season there is more oligomer, and the corona treatment has been optimised in the dry conditions and so is worse in the high humidity season and also the moisture level is higher in the vacuum system which also has had the plasma treatment optimised in the dry season and so this too is not optimised.  The cumulative effect of all this could lead to a surface that is very different to that produced in the dry season and so the wetting and adhesion is poor by comparison.

Thus there are a number of things that you may be able to check if you have been monitoring and recording them over time such as system base pressure and pump down time.  The pump down time is likely, on average, to be longer, even after cleaning, in the high humidity season than in the dry season.  If you carry out a corona treatment check the settings to see if they are fixed or are modified with changes in humidity.  If they are fixed it would be worth planning to check the surface energy as a routine measurement over a year to relate the performance to seasonal changes such as humidity and temperature. Plotting the temperature and humidity against the system pumpdown time and noting the cleaning points too is also worth doing.  It would be difficult to plot the surface energy after the vacuum plasma treatment and so it may be that measuring the bond strength of the coating against daily humidity would be easier.

Measuring oligomer levels may be more difficult as this would involve extraction and weighing the oligomer to a high precision and this may be very time consuming.

I know this does not solve your immediate problem but it will give you the information in the future to know which areas to concentrate on in improving monitoring and control.

In the short term I would check that the corona treatment and plasma treatments are both optimised for the current conditions.  Both may have been optimised in the past but as explained above this optimisation may no longer be correct.

New type of deposition source.

In the past I have advocated using a slot deposition source instead of the more standard resistance heated evaporation boats.  The reason for this is that the deposition efficiency of the standard boat sources can be anywhere from 35% - 60% depending on the geometry adopted.  This is poor compared to using a slot source where the deposition efficiency can easily be >95%.  This high efficiency deposition improves costs not least because the stray deposition is minimised and so the system cleaning can be reduced and the scrap is minimised.

The lack of interest in developing slot sources has been for many reasons. The crucible material has always been put forward as a problem.  Molten aluminium is very corrosive and is part of the reason why resistance heated boats only last 10 – 15 hours.  I have never been convinced about this argument as Mitsubishi and Hitachi both developed air-to-air systems that could operate for >100 hours using alumina coated boats.  Another argument for not developing the source was that problem of feeding the source.  In an enclosed source any hole is a potential exit route for aluminium vapour. If the source is fed by an aluminium wire then where the wire enters the source the vapour could escape and condense and potentially weld the hole shut. The alternative approach is to have the whole aluminium inventory in the crucible from the start of the process. This significantly increases the thermal mass and slightly slows down the heat up time but more significantly slows down the cooling time. This is particularly slow if there is an emergency stop and so venting the chamber can be delayed making the recovery time significantly longer.

These excuses have been enough for machine manufacturers not to take much interest in developing a slot source. However things are now changing.  The cost of energy is rising and so machine efficiency is of increasing importance. Reducing system wastes is also becoming more important as the environmental issues such as disposal is becoming more costly and recycling is preferred.  Also the high energy process for manufacturing aluminium will increase the materials costs and so minimising aluminium cost will become of more interest as costs continue to rise.

This brings be on to the new source.  This source uses the novel idea of having magnetic levitation instead of a crucible. Hence all the problems of materials choice are dispensed with by having no crucible at all.   A couple of coils are used to produce magnetic fields that pin the metal in place.  A slug of metal can be fixed in place to start the process off.  A high frequency alternating current can be used to heat and melt the metal whilst it is pinned by the magnetic field. The vapour is not constrained by the magnetic field and so can escape. By enclosing the source and leaving an escape slot the vapour can be directed towards the substrate.  The molten metal can be replenished by a wire feed. As with the evaporation boats the wire feed has to be balanced against the evaporation rate.  The wire feed still has to be managed to allow the wire to enter and yet restrict the vapour escaping. This source has been used on web substrates and the material efficiency has been demonstrated to be well in excess of 95%. 

Unfortunately this source has been developed by a user company rather than one of the machine manufacturers and so at this time does not look like becoming widely available as a replacement for resistance heated boats.  However at least one manufacturer is developing a jet vapour source, also a slot source, and this holds out more hope that this, when they finally start to publicise it more widely, will become available to those who want it.

Questions re. backsurface treatment & starry film

1.         We are getting back treatment in our PET film, for 52+ corona treated film we are getting 48+ dyne value on non treated side, what can the possible reasons for this problem?

2.         How this back treatment can affect metal bond strength as we are using this film for metallization?

3.         For such corona treated film after metallization we are seeing Star-Sky defect under light box, is this problem is related to back treatment or what other reasons possible for this problem.

Answer

I would suggest we separate out the different problems.

The star sky sounds to be a simple case of many pinholes. Pinholes are primarily caused by dust or debris on the polymer surface that is metallized and then as it gets moved away it leaves a small area that is not metallized behind that is seen as one of your stars when the film is placed on a light box.

A secondary source of these demetallized areas can be through pickoff. This can be where the filler protruding from the back surface is pressed hard against the front surface and when next unwound the pressure has cause the metal to transfer from the front surface to the back surface causing the loss of a small amount of metal that is also see as a pinhole of star on your light box.  If the back surface of your web has been corona treated to raise the surface energy it will also raise the adhesion and may make this back surface pickoff more likely than if the back surface were left untreated.

Corona treatment is generally carried out on the front surface only but there are circumstances where the resistance between the roll and ground or film and the roll are altered sufficiently that not only is there a corona plasma produced on the front surface of the web but also another one is produced on the back surface of the web too. It is unlikely to be as powerful as the front surface corona as is possibly being demonstrated by the slightly lower surface energy of the back surface.  This back surface treatment can lead to some winding problems because the higher surface energy tends to make the back surface less slippy and so slightly harder to wind and so this can lead to more stick-slip behaviour during re-winding and this can encourage some of the movement of debris which you see as an increased number of stars.

Thus I would go back to the corona treatment station and check on the set up, paying particular attention to the connections and also the conductivity of the roll to make sure it has been set up correctly to provide a front surface corona only.

Problem of Static

We are getting higher Static Charge development in the slit reels, more in lower thickness and smaller widths.  Please advice, How to get it reduced. Our slitting machines are equipped with mechanism of discharging. Yet it is not working properly in thinner films and lower widths.

Answer

Do you measure the static charge on the web?

When you mention a static neutralizer is this a simple carbon brush, tinsel wire, or an electrostatic discharge bar. If it is the latter is this a fixed voltage system or is there some kind of feedback or tuning?

The reason I ask these questions is that there are a variety of possibilities.  If the web material is the same for the thicker and thinner films it may be that all you are seeing is that the static charge is the same on both the films but the lower stiffness of the thinner film is showing the effects more visibly than the thicker film.  Typically if the webs are the same material then over the same rollers you would expect the triboelectric charge that is generated will be of the same magnitude. 

The charge generated can vary also with winding tension and well as humidity.  Humidity is a mechanism to dissipate charge and so on days of high humidity the charge generated will be lower than on days of low humidity. 

Static will be generated at every roller in the winding system. The static charge is cumulative and so it is important to have a measure of the charge on the web so that the positioning of the neutralization system can be optimized. It may even be that more than one neutralization system is needed.  The static charge will be generated on both sides of the web and so it is also important that the neutralization is done on both sides of the web. 

So I would suggest that you measure the static charge during winding at different positions along the web and look at how effective the neutralization is. This will also tell you if the position of the neutralization is in the correct position. In some cases it may be necessary to add a second neutralization station just before the rewind.

If the neutralization is an active one then it is also possible that it is set too high and instead of just neutralizing the surface it is in fact charging the surface but with the opposite polarity.

Seed layer for coating silver on glass

We are currently trying to deposit Ag on glass and have been told to deposit a layer of Nichrome first for better adhesion, maybe 5 to 20 angstroms. Is this necessary or should a glow discharge cleaning be good enough to get adhesion to the glass we want a nice bright coating, second surface, and the nichrome goes down a little dark. Any advice or experience would be greatly appreciated.

Answer

There is a tendency for the silver to not wet the glass well and so the coating will have to be quite thick before it becomes continuous and this also causes a higher surface roughness and this can reduce the reflectance.  A glow discharge clean will raise the surface energy of the glass and improve the wetting and hence produce a smoother coating as well as a continuous coating at a lower thickness. The adhesion between silver and glass may not necessarily be good and hence the recommendation to use a tie layer. The choice of tie layer is usually as a result of many peoples experience in what works well and what doesn't and hence the recommendation for the nichrome. However if you look at the work done with heat mirror coatings using titania/silver/titania you will find that to be able to use thinner silver coatings and keep them stable in the vacuum deposition environment some companies use either tie or protective thin layers on either side of the silver.  They too do not want the thin tie layer to become part of the optical performance and so the layers are kept very thin and often they try to match materials. So that if they are using a silica or silicon nitride layer next to the silver they will use a silicon thin tie layer but if they are using titania next to the silver they will use titanium. The thickness is also as has been recommended to you at less than 3nm. The expectation is that these tie layers will be discontinuous at this thickness and that they will also be easily oxidised and so optically they will disappear.

Thus for your silver coating I would plasma clean the glass, use the nichrome but aim for a thin layer of  1.5 - 2nm before depositing the silver.

One other thought that does cross my mind is, how are you determining the thickness of the nichrome layer? It is very easy to think you have a very thin layer but in reality to have quite a thick layer that then acts differently. How is the nichrome being deposited, by sputtering or some other method? If you are sputtering it will be easier to reproducibly deposit a thin layer but if you are using an electron beam gun there can be problems of deposition rate stability which can lead to thicker tie layers.   

So I would also check carefully that you are depositing a tie layer that is thin enough and not too thick.

I hope these thought help

Question re. Plasma treatment

I want to know how and up to what extent the distance of plasma target plates and substrate surface can affect treatment  & is there is any possibility of back treatment during plasma treatment because we are getting both side 70+i.e. water hold on both side of metallized film if yes then can it impact bond strength.

Answer

It depends on the geometry of the plasma treater and the positioning of the web as it passes the surface.  The pressure of the process defines the mean-free-path of the gases and if the gap between either surface is larger than the mean-free-path then it will be possible to generate a plasma on both sides of the web and hence yes it would be possible to be plasma treating both sides at the same time.

As to whether it can affect the bond strength the answer is yes. Any plasma treatment can affect the bond strength if the treatment is on the surface that is involved.  If you are asking can plasma treating the back surface affect the bond strength on the opposite surface the answer is no although it can have other effects that may cause problems. The back surface may have low molecular weight material present either by design as in the inclusion of slip agents or by default as in oligomers.  This low molecular weight material may aid the handling characteristics by lowering the coefficient of friction.  Plasma treating the back surface and removing this material may raise the coefficient of friction and cause the film to no longer slide over itself so easily.  This in turn can cause an increase in pick-off or scratching damage.  Neither of these effects will affect the bond strength.

The front surface treatment is the one that will affect bond strength and so long as this has been optimised you will get the best possible bond strength from this treatment. If the treatment has not been optimised the bond may be lower than ideal and it may also vary with varying conditions such as humidity of the air wound in the film used.  The plasma does not get diluted by being shared between the back surface and front surface. Essentially the plasma will have the same effect on all surfaces. 

I hope this helps with your problem.

Pattern metallization – AIMCAL Fall Conference preview

For those of you interested in pattern metallization there are three papers that will be of interest to you at this years AIMCAL Fall Conference in Myrtle Beach, USA, Oct 19-22. 

The first of these has been long awaited and is from Nick Copeland of Bobst General Vacuum and is about their ability to build in pattern metallization that can be kept in-register with existing patterning on the web.  This becomes important where webs have been either pre-printed or pre-embossed with a pattern.  Using a reference marker they are able to positioning of the web to keep the metallization pattern in register with these existing patterns to a high degree of precision.  This increases the flexibility of the process and allows it to be used in a number of the more challenging markets including security markets.

            The other two papers are both by Leybold Optics, the first by Christopher Schmitt on the measurement of patterned coatings and the second by Anye Chifen on modifications to the oil masking process to reduce the oil required and improve the quality of the patterning process.  The measurement of patterned materials can be a problem. Most measurement systems are taking an average measurement either optically they are averaging the light passing through an area or electrically they are averaging the conductivity based on an area of coating. In either case if the area is varying because of changes in the pattern shape the measurement will also be changing. If the measurement is changing it is then difficult to use this measurement as a feedback signal to control the deposition process and so control the coating thickness.  In this instance they have developed a method of taking Optical Density (OD) measurements from a very small area and so are more easily able to choose an area that is not fluctuating in size.  The second paper describes a different method of supplying the oil to the oil masking process. Typically this is done by heating the oil to vaporize it inside the vacuum system but in this case a metered flow of oil is injected into a hot evaporation box inside the system where the oil can be vaporized and via nozzles onto the web to provide masking.

            So for anyone either already involved in pattern metallization or those who are thinking about adding this to their skills there are three more good reasons for attending the coming conference.

Flake pigments – AIMCAL Fall Conference

Just to whet your appetite for the Fall Conference in some of the future posts I will highlight some of the papers that are expected to be given later this year at the AIMCAL Fall Conference in Myrtle Beach in the USA (Oct 19-22).

For this first tempter I will tell you a little about a paper being presented by Craig Engle of the South Western Research Institute (SWRI) on nano-platelets.  First let me say a little about the topic of nano-platelets.  This covers a wide range of materials that can vary in size range quite considerably. The platelets you are all probably most familiar with are those that are used in inks and paints.  Aluminium inks and paints have improved in reflectivity and brightness over the last few years.  The original pigments were balls of aluminium that were put in a ball mill with some very hard spheres that would flatten the softer aluminium into flakes. These flakes were not perfectly flat but were much more wrinkled and are often referred to as cornflakes because of the similar appearance. The next improvement was to polish some of these cornflakes and make them smoother. These too were not perfectly flat but were more lenticular in shape because the flakes would rock as they were being polished. The latest method of producing flakes is to coat a release layer onto a roll of substrate, often polyester film, and vacuum metallize aluminium onto the release layer. This metallized film is then passed through a liquid that dissolves away the release layer freeing the aluminium up. The aluminium is very thin and the film breaks up into flakes.  These flakes are as flat as the polymer film and so the reflectivity and brightness are far superior to the other methods of flake manufacture but the vacuum metallizing is a more expensive process that the simple ball milling.

            This paper from SWRI adds some new developments onto the manufacturing process.  The standard method of producing flakes in this way does nothing to control the size and shape of the flakes. The flakes produced will be random in shape and the sizes will be from hundreds of microns across to fines that are dust.  To make the flakes useful they have to be sized or milled and sized with the different size ranges giving different optical properties when they are used in the inks or paints.  Where these high aspect ratio flake materials can be used in other applications there is a preference that they are not only size more uniformly but also the shape is more regular than random.  What SWRI have done is to modify the surface that the metallization is done onto so that they can control the shape and size. Also as is implied by the title they have done this at a very fine level so that they can control the flakes down to the nanometre level.

Question about too much (?) cooling

I want to know the effect of over cooling during metallization, how does the cooling impact on bond strength and other quality parameters of metallized film.

Answer

When the substrate enters the deposition zone it is heated up by the radiant heat from the red hot evaporation boats and well as the heat of condensation of the depositing metal. What you are achieving with the cooling is to reduce the temperature rise that the film reaches.  Without any cooling the film would typically exceed 100 degrees Centigrade and for a thick coating at high speed could even exceed 150 degrees Centigrade.  By using a cooled deposition drum the substrate is pre-cooled often to subzero temperatures and even for the same rate of heating this will reduce the maximum temperature by the same difference as the minimum temperature is below room temperature. So if room temperature is 20 deg C and the deposition drum uses a glycol/water mix and allows cooling down to -20 deg C then the difference is 40 deg. So if the maximum temperature the film was reaching without a cooled deposition drum was 125 deg C then the maximum temperature with the cooled drum would be 85 deg C.   In reality this would be better than this as the film continues to be cooled throughout the deposition zone and not just pre-cooled.

The cooling continues after the deposition zone as the film is still in contact with the cooled deposition drum and this brings the temperature of the film back down to room temperature.

There is the possibility that the cooling available brings the temperature back down to below room temperature. Usually the film as it then passes over other rollers, which are at room temperature, will equilibrate to room temperature before it is re-wound into a roll.   If the film does not equilibrate and is wound cold then it will have the problem that as it rises in temperature on the roll it will loose some tension as the polymer expands and during the temperature change not all the layers will be at the same temperature and so moving it at this time can lead to slipping of the layers (telescoping).  This is also true of rolls have been stored in a cool warehouse and moved into a warmer area for deposition.

In terms of the deposition process the difference in temperature between an un-cooled film and a cooled film at the point of nucleation is only around 40 degrees and so there will only be a small difference in the nucleation size, with the hotter film having the slightly higher crystal size. Similarly with the continued growth the hotter film would tend towards a larger crystal size over the cooler film.  But this will be marginal compared to the difference that can be created by plasma treating the film surface to change the surface energy and hence the wetting of the film.

Adhesion, it could be argued, would be better for a cooled film as the differences in thermal contraction of the polymer and metal will be lower from the lower peak temperature.  The coefficient of thermal expansion for a metal is much lower than for a polymer and so as the temperature is reduced the polymer shrinks more than the metal coating does.  This can put a strain on the interface. If the temperature peak is reduced there is less contraction and so less strain on the interface. Adhesion failure occurs when this interfacial strain exceeds the adhesion and so minimising the strain is useful.

I hope this answers your question