Vacuum Web Coating

Aluminium Oxide transparent barrier coatings

Aluminium has long been the main vacuum coated barrier material but it does, of course, have the property of being opaque. So for those wanting a transparent barrier coating they have always looked to other materials of which the most popular has been some version of silica.  Most time this is a sub-stoichiometric version of the silica and in the case fo the plasma enhanced chemical vapour deposition (PECVD) process it also contains an amount of carbon, anywhere up to 20%, or more, depending on process conditions.

The cost of these alternative materials has been high with the lowest still being of the order 2x to 3x the cost of aluminium metallizing.  This is because of two significant costs, the capital cost of either a PECVD or an electron beam deposition system is a much higher cost than a resistance heated boat type metallizer and the deposition speed is often less than a half that of aluminium metallizing.

The one company that appeared to have solved this problem was Camvac who managed to produce an aluminium oxide coating from their modified standard metallizer.  It would appear that by them carefully introducing the oxygen at the right point and in the right quantity they could control the oxidation fo the coating but also not damage their resistance heated boats. 

Needless to say this technology has been patented and so for many years this has been the only low cost process that I have known to be operating.  However in recent times I have heard that there have been two others who have managed to develop a similar process. One I know has been checked out to make sure it was not infringing any patents and that they were clear to run production the other I do not know if it has been checked out but they are prepared to talk about the process and will be giving a paper on the process at the 2009 SVC conference in Santa Clara.  This process was developed by the Fraunhofer Institute in Dresden (FEP) in conjunction with a polypropylene converter and the Applied Materials built a machine using the technology for use in production.  The FEP specialise in using additional plasmas during deposition to activate any reactive gases and to also promote surface reactions and densification of the growing coating.  Thus I expect that this process uses that expertise to evaporate the aluminium quickly but then use the plasma to activate the oxygen to speed up the conversion of the metal to the oxide.

As aluminium oxide is less dense than the metal this conversion from metal to oxide can also help improve the barrier as it will add a compressive component to the coating, as it swells, that will close pores down and make diffusion through grain boundaries slightly harder.  If this technology is now available through one of the system builders then this could mean that the cost of obtaining transparent barriers could be due for a fall in the next couple of years as these coatings start to become available.

Air-to-air vacuum coating systems.

With the surge of interest in vacuum coating machines for the deposition of photovoltaic materials there has been the development of evaporation sources.  This has included those evaporation sources that have a very high material efficiency.  This type of source can offer opportunities to change the design of the vacuum system.  The source no longer needs to be brought to atmosphere after the deposition of each roll.  It is only the unwind and rewind polymer rolls that need to be changed.  This leads to the option of using load-lock chambers for the unwind and rewind rolls which would allow the main chamber to be held under vacuum during the roll change over.

There is a trade-off between the system cost of a more complex winding system and the load-lock design and the possible reduction in some of the pumping capacity as the main volume does not need to be pumped out as frequently and so it is usually allowed slightly longer pumping time thus requiring a slightly smaller pump set. 

A second design option is to go to an air-to-air winding system. This still uses a load lock for roll changes but there is no need to vent and pump any vessel around the unwind and rewind rolls.  However the pumping requirement is much higher as to get the film into and out of the vacuum system there is, in effect, a continuous air leak that has to be pumped.  The pressure is reduced as the film passes through a series of chambers. Each chamber is up to two orders of magnitude different to the adjacent chambers.  Thus not only the pumping system is more substantial but also the winding system has considerably more rolls included. The air-to-air systems, although it has attractions, may also have problems such as increased contamination as the air that enters the vacuum system continuously will contain airborne particles that can become pressed into the film as the film passes through the nip roll that restricts the quantity of the air passes into the system.  A second detraction that has been an irritation to a number of operators in the past is that of noise.  The velocity of the air is high and this can cause a continuous loud noise requiring ear defenders .

Both of these system options also allow the deposition sources to be kept under vacuum and hot during roll changes.  This can reduce the down time between deposition runs as there is no need to cool the source and then re-heat of which the cooling generally is a rate limiting part of the process.  There is usually a small reduction in temperature in order to slow down the evaporation rate whilst maintaining the bulk of the heat. This smaller temperature variation also helps in reducing any source temperature variations and can thus be of benefit in improving deposition uniformity.

As more of these different systems are built more experience will be gained and the designs refined with the best features retained.  The economics of the different capital cost and running costs can be evaluated and this may lead to changes in the design of metallizers. So do not be surprised if these is a new batch of air-to-air metallizers produced in the near future.

Lifetime of metallized film.

Hi sir, follow up question on the lifetime of metallized but this time on the adhesion of the metallized coating on the cpp or pet film. Some customer complains that after a year their un-used laminated opp/vmcpp film delaminate. We observed that the metallized coating completely transfers to the opp. Did the bond between the cpp and the metallized was overcome by the bond of adhesive between opp and metallized coating. Did surface contamination plays part on this after a long period of storage.

Answer

As with all adhesion failures the recommendation is to first confirm the plane of the failure. The reason for this is that you do not want to be trying to solve the wrong problem.  Assuming the failure really is at the interface it is possible that this relates to the initial adhesion and then is time and temperature dependent.  Ideally the metal is bonded to the polymer at all possible points uniformly across the whole surface.  If the surface is not suitably prepared the metal will only be intermittently bonded to the polymer. In the spaces between bonds it is possible for material to migrate into the space and it is this material that can degrade the adhesion.  This migrating material could be moisture or low molecular weight unpolymerised fragments or any additives that might have been added to the polymer.

The rule of thumb is generally that the higher the original adhesion the longer the lifetime of the metal adhesion. The lower the adhesion the easier it is for material to migrate into the space between the unbonded metal and polymer swelling the space and leading to premature delamination.  Heating the material can accelerate any degradation process as it increases the rate of migration or diffusion through the polymer of any potential contaminant.

If you have a known problem of failures after a year of storage it should be possible to plot the progressive decrease in adhesion over the year.  The force used for delamination should reduce progressively.

Thus examining the surfaces may give some indication about contamination. If you have some material that is known to fail then sending this to a surface analytical laboratory and allowing them to delaminate some material in their controlled environment and then examine each of the metal and polymer surfaces by a technique such as X-ray Photoelectron Spectroscopy should allow you to identify the chemical composition of both surfaces.  If there has been some contamination it is likely that you will have the same material present on both the metal and polymer surface and it will not be the same chemistry as the original polymer surface.

There is also another factor that may be coming occurring and that is the adhesive between the OPP and MCPP may be also aging but in this case may be slightly improving in adhesion. It depends on the adhesive type but some do not fully cure immediately but take some time and so with time can increase in adhesion.  Thus if the adhesion on one side of the metal is reducing and on the other side it is increasing the failure may switch from one side to the other over time.

I hope this gives you something to work on.

Question about retrofitting a glow discharge

Hi, We have a very old Dynavac Evaporative Metalliser and would like to obtain better adhesion in our substrates by building a Glow discharge. While the principle is simple (DC transformer connected to an Aluminum rod and the chamber, then keep the vacuum level low while energizing). Looking at your diagram on page 229 of your book the ideal setup would be to have Vacuum at 10 -1 with voltage at 900 DC, to achieve this vacuum we would install a bleed value at the opposite end of the chamber to the pumps. My questions are does this setup sound ok and can we use filtered air instead of an inert gas, or do you suggest some further reading before we jump into this?

Answer

The glow discharge sounds as if it would work however if I were to be modifying a system I would aim for a magnetically enhanced plasma to treat the surface. The power requirement can be the same but the plasma density with the magnetic enhancement will be greater as it can run at the lower voltage but will carry a greater current.  The design is very similar to a magnetron source with the magnets making a magnetic circuit for the electrons to race around.  The higher density plasma will give a faster treatment time than the glow discharge and so is more useful at the faster winding speeds.  It will also operate down to lower pressures and so can be more flexible in where it is sited in the system.

As to gases, any gas will allow a plasma to be struck but the surface treatment will be dependent upon the gases present.  Argon is a heavy atom and when it strikes a polymer surface it can break bonds but argon is inert and so cannot react with anything on the surface. Thus it can hit contaminants on the surface and may crosslink them to the substrate or may further fragment any short chain molecules or oligomers. To remove any hydrocarbons the plasma needs to contain a reactive gas, oxygen for preference which can react with the hydrocarbons to form volatile species that will desorb from the surface and can be pumped away.  If you have a roll of polymer in the system there will always be some oxygen around from the air trapped in the roll and also from any moisture in the air and absorbed into the polymer. The water will be cracked in the plasma and more oxygen released.  Thus you introducing dry, filtered air will provide the surface with two gases both of which are oxidising gases with oxygen being more active than the nitrogen.

Practically if the pumps are sited to the side of the winding system there will be a small pressure gradient from one side of the chamber to the other. When you introduce the gas there is a danger that you will exaggerate the pressure gradient. Sometimes I have seen a tube used with holes drilled in the tube to allow the gas out. By suitable positioning of the tube and varying the size of the holes the pressure can be balanced. There may still be some variation across the web from plasma generated by-products which will be present at a higher proportion towards the pump. 

If you use a magnetically enhanced plasma the system can be run at a lower pressure and this will reduce this effect somewhat. 

Deposition sources

There are changes coming in deposition sources largely driven by the new enthusiasm for high rate deposition for photovoltaic devices.

The photovoltaic materials generally have to be deposited to a thickness of greater than 1 micron and so the deposition process has to be fast.  Although some of the companies are using sputtering the more successful companies depositing the copper indium gallium diselenide (CIGS) materials are using evaporation.  The CIGS compound is evaporated from a series of sources of the individual elements enabling the compound stoichiometry to be graded through the thickness.  Although the compound has to be graded through the thickness the requirement is for the thickness and stoichiometry uniformity across the web to be precise.  Thus there has been work done to improve the stability of the deposition process and control the thickness uniformity. 

There have been two different approaches to the sources. One has been to develop the jet vapour source where the evaporation is done into an enclosed volume where all the internal surfaces are kept hot to prevent condensation and the exit slot is the full width of the substrate.  The internal vapour pressure evens out any evaporation variations so that the exiting vapour uniformity is very good.

The second approach has been to take existing high stability vapour sources that have well known characteristics and to use arrays of these sources to provide the uniformity.  The Knudsen source is the basis of this type of source.  The semiconductor industry has used this type of source for many years and it has been modelled extensively so that the interaction of multiple sources is well understood. The design of these sources has been developed and optimised so that the temperature is controlled to a fraction of a degree and so the evaporation rate is more precisely controlled than most evaporation sources. 

What becomes clear is that these sources could be adapted to the deposition of aluminium. The benefits of doing so would be not only that the uniformity of the deposition would be improved but also that the material efficiency would be significantly improved.  Currently evaporation from resistance heated boats can have an efficiency of anywhere from 35% up to more than 50% depending on the system design including source to substrate distance and deposition drum size.  The vapour jet source, in particular, can have a material deposition efficiency of greater than 95%.

Where these new sources are unproven is in two aspects. One is the replenishment of the sources. Neither of these sources has a replenishment facility. The whole inventory of material has to be loaded and heated at the start. This is because any feed process has to be done via a hole where vapour can escape which is not only a material loss, a cooling point and a possible problem through condensation of the vapour that could close up the hole causing feeding problems.  The Knudsen sources can usually hold sufficient material for several deposition runs to be completed.  This leads to the second area of process uncertainty which is that of source cooling.  At the end of the first deposition run the source needs to be cooled to a safe temperature for the system to be vented.  As these sources are designed for temperature uniformity they usually include radiation shielding which both limits the energy losses but also slows down the cooling because of this minimisation of heat losses. In the past gas quenching has been used to accelerate the cooling process.  In some systems the need for cooling the sources has been followed by using load locks for the un-wind and re-wind rolls with the main vessel kept under vacuum for the multiple depositions the inventory of the source can allow. As the material efficiency is so high the need for shield cleaning is reduced and so keeping the main vessel under vacuum is possible.

Another advantage of using this different evaporation source is that the range of materials that can be evaporated is increased.  It has always been one of the limitations of resistance heated evaporation sources that the range of materials is very limited.  Sidrabe have developed an alternative source where they extended the range of materials by using tungsten rods as the core of the evaporator and refractory materials for the enclosure.

I hope this gives a glimpse of what is coming. It may take a few years for these developments to filter through to the aluminium evaporators but it will eventually be adopted.  As the material efficiency can be improved so much the energy requirement can similarly be reduced and this will become increasingly important as energy process continue to rise.

Wrinkling - additional monitoring

Recently I was visiting a well known metallizer manufacturer and this was an observation that came from some of our conversations.

I had been looking at metallizers and thinking that in so many ways they had not really changed significantly in decades.  The basic wire fed evaporation source was easily recognised as working in the same way.  There have been changes behind the panels in that the power supplies are more sophisticated and power is applied better and more uniformly and similarly the wire feed is better than it was. 

I was looking round at things that make the life of the machine operator easier. One change stood out as being both simple and helpful and that was the view into the system.  I have lost count on the number of occasions where it would have helped in diagnosing a problem if I had only been able to see better into the system to look at different parts of the winding to see where winding problems were beginning.  Generally we had one or possible two windows to look through and possibly a strategically placed mirror added inside the system to give some view of an obscured part of the system.  This has all changed with the miniaturisation and price reduction of camera systems.  I had previously worked on a vacuum process for roll coating explosives where we controlled the system from behind a blast wall and so our only view of the system was via cameras. This made me familiar with split screen multiple view displays but even the 10 – 15 years ago the camera technology was still quite bulky and so these cameras were sited outside the system pointing in at various angles to give us the necessary vision.  This was never as much or as good as we wanted.   Now the cameras have become small enough that they can be fitted with easy within the system and cheap enough that it is possible to consider using many cameras within the system to give all the necessary views to help with diagnostics.  What is more it is almost as cheap to use colour cameras as it is to use black and white.  The same rule applies as ever it did and that is that it is critical the lens of the camera is protected from any stray deposition.

Thus it is now conceivable to have a view of the film as it comes off any high wrap, spreader or bowed roll as well as out of the metallizing zone. This means that identifying where wrinkles start should no longer be a guessing game. Our first thought is always that the deposition has caused the problem but there are occasions where that is not true and it is always expensive in time and money to be trying to solve the wrong problem.

Retrofitting is generally more problematic than installing during the system build but as most of this is just cabling this should not be too difficult.

So if this has persuaded any of you to retrofit some cameras or for those of you ahead of the game that might have already done so I would be delighted to hear how you have got on with them and if you have experienced any problems.

Web Processing for Barrier

AIMCAL Converting School          Web Processing for Barrier   October 9 - 10, Cleveland, Ohio

Room Rate Guarantee Deadline is September 24!

The Web Processing for Barrier course provides an overview of the technologies that can be applied to deliver barrier materials for markets such as food packaging, display, and photovoltaic (solar) applications. Initial discussion focuses on the fundamentals of diffusion and permeation to provide an understanding of the contributions and limitations of materials and processes. Advanced topics include newer technologies such as nanotechnology, scavengers, and indicators that result in "smart" materials.

Course Outline
Day 1: 8:30AM - 5PM
Day 2: 8:30AM - 4PM

Introduction
The Basics of Barrier
Markets
Terminology
Techniques and standards for barrier measurement
Materials
Performance
Blends and laminates
Nanomaterials
Selection
Smart/active packaging
Indicators
Scavengers
Technologies
Film making and extrusion
Coating
Lamination
Vacuum deposition
Substrates, surfaces and quality
Relating materials and surfaces to barrier
Pre-treatment (wetting, adhesion)
Cleaning and barrier
Variation in barrier during processing
Who will benefit from this course?
Anyone working at facilities that convert or use barrier materials including engineers, designers, quality control, stability and production personnel.

Date for this course:
| October 9 - 10, Cleveland, Ohio |

Comparison between CPP and BOPP

As we know CPP and BOPP exhibits different properties. Especially sealing properties are quite different. CPP exhibit better seal strength than BOPP. I request you to enlighten me on technical grounds why such difference is exhibited?

What will be the performance if we use same tool to seal CPP and BOPP?

Answer

The acronyms are as follows CPP = Cast PolyPropylene  and BOPP = Biaxial Oriented PolyPropylene.

Cast polypropylene is produced by extruding polypropylene and chilling it and so the polymer chains within the film are randomly distributed throughout the film in all three dimensions.

BOPP starts out in the same way, the polypropylene is extruded and chilled but then the polymer is re-heated and stretched in two directions hence the 'bi-axial orientation' in the description.  This stretching is done when the polymer is softened and so the polymer chains are able to rearrange themselves to some extent and so they become oriented into the two stretching directions.  This changes the mechanical performance of the polymer. The tensile performance is higher along the polymer chain length than it is across the polymer chains thus in either of the two stretched directions the tensile performance is improved but in the third axis the tensile performance is reduced.   In the ordering of the polymer chains many of them become ordered enough that they become crystalline. This too effects a change in the performance, as crystalline material is a better barrier than amorphous material.  Crystalline polymer is denser than amorphous polymer and the amorphous polymer will soften at a lower temperature than the crystalline.

Thus using the same heat sealing tool may well work OK but may need a higher temperature for the BOPP than for the CPP.

I hope this answer helps.

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 about pressure variation when using plasma treater

We have metallizer. When operating plasma treater there is a variation or rise in evaporation zone pressure throughout the metallizing cycle. Theoretically the pressure of winding should rise as plasma treater is located in winding section, but the winding zone pressure is not varying, what is the reason behind this?

Answer

This sounds to me to probably be related to a problem of cooling.

I would monitor any cooling systems that you are using. If you have different cooling water to the plasma treater, the deposition drum, protective shields and the evaporation boats I would monitor each separately. I would check both the input temperature and output temperature.

Typically once you have pumped out your system the residual gas is almost all water that is out gassing from all surfaces including from the unwinding roll from trapped air and from water contained within the polymer.  Once the roll is unwinding there will be an increase in the gas load because of the release of the trapped air as the roll keeps presenting fresh surfaces. 

As you suggest this is a fairly constant process and so you would expect everything to remain constant throughout the deposition run.

If you have a problem with a cooling system what happens is that initially everything is cool as the process has only just started and there may be a large thermal mass that has to be warmed up. If you are recirculating the water (or coolant, which may be a water /glycol mix or something similar) there is also the thermal mass of the water.  What happens is that the water passes through something that is heating up and so the water is heated up, but usually to a lower temperature than whatever they are passing through. The coolant then returns to the chiller that returns the coolant to a constant starting temperature.  If the heating load is higher than the chiller capacity you will progressively see that the coolant is no longer returned to the same starting input temperature but this temperature gradually creeps up.  As the coolant temperature creeps up it no longer cools as well as it ought to and so the internal surfaces become hotter.  If these surfaces have any stray deposition on them they will have a very high surface area and so can contain a lot of water. As the temperature increases water can be released from the surface thus raising the pressure.

Most systems have cryopanels that will pump away the water very well but it does depend on the water having access to the cryosurfaces.  It is common to see systems with only cryopanels in the unwind zone whereas it has been shown by Telemark Cryogenics  Ltd as presented at AIMCAL that it is better if the cryopanels are distributed across different zones.  The deposition zone will always heat up the web and internal surfaces more that anything occurring in the unwind zone. The deposition zone also has the greatest potential for having coated surfaces that can mop up very large quantities of water.

If the problem does not appear what the plasma treater is not used it could be that some of the cooling systems are linked and that it is only when the plasma treater is used that the chiller cannot keep the temperature low enough.  Even if the coolant systems are not linked the plasma treater does add another heat load within the system. It is possible that the hot plasma treater is radiating heat to another surface that is the releasing the water.

Another option is that the extra heat is reaching the cryopanel surface and that over time the surface of the ice no longer remains frozen but stays as a liquid with a correspondingly higher vapour pressure.  The cryopanels are always a balancing act as the core remains cold but the growth of the ice can affect the total capacity of the panel. If the water is released within the system very quickly the ice may be very porous and this is then has a worse thermal conductivity than if the ice was built up slowly and was very dense.  The dense ice will allow more water to be collected than the porous ice. If there is too much water within the system over time the ice will build up to a thickness where no more water can be condensed and so the system pressure will rise.

Finally it is possible you have a small water leak that is normally closed but under heating opens up slightly and leaks water out.

I hope this helps you sort out the problem.