Ion Assisted Deposition

As coating specifications and tolerances become more challenging, you may find yourself unable to meet customers’ needs with a standard evaporative coater. With this in mind, many evaporative coating facilities are looking into Ion-Assisted Deposition (IAD) to provide added value to their existing equipment. The benefits of IAD are plentiful, and ion beam sources can be surprisingly easy to use, too!

This article describes the benefits of IAD such as improved adhesion, stoichiometry, and defect mitigation. IAD also plays a key role in improvements such as durability, damage threshold, and reducing spectral shifting. In the “Handbook of Plasma Processing Technology”, J.J. McNally[1] discusses several types of IAD and their benefits in detail, including the basic physical reasons for each idea. For users new to the idea of Ion Beams, this chapter provides a great deal of clarification and can be very valuable.

A Physical Description

     A typical evaporation process requires low-energy particles of a desired material to adhere to the substrate. The energy that the particles have is low – in the 0.1eV range. Due of the low energy state, the particles tend to form crystalline structures (columns) which are somewhat brittle and have large numbers of defects and pores. Once the film is exposed to atmosphere, water vapor naturally migrates into those pores causing a spectral shift and further weakening the coating.
Evaporation also suffers from poor adhesion. Any particulate contamination on the substrates before deposition weakens the coating bonds, and can lead to flaking. Contamination can also be in the form of native oxide layers. Achieving oxide stoichiometric films is challenging with evaporation and may require either extremely high oxygen flow, reduced evaporation rates, or an acceptance that the stoichiometry will not be ideal.

Ion Beam Solution

     An ion beam source directs high-energy oxygen ions at the substrate. These incoming ions have far greater energy than the evaporate (on the order of 100 – 200eV), and upon striking the substrate will deposit this energy into the existing layers of the coating. Importantly, the energy is high enough to embed the oxygen ions down several nanometers into the coating, providing a dose of reactive gas to regions which may not have been fully oxidized before being covered by evaporate. Further, when the energy of the incoming ion is deposited into the coating, surface atoms from the coating are liberated to move and shift along the surface. This helps create an amorphous solid as opposed to a crystalline one, removing the columnar structure that promoted water absorption and creating a more dense overall structure. The absence of defects and better stoichiometry also reduce light absorption, increasing the damage threshold and overall performance of the coating.
As an added benefit, an ion beam source can be directed at the substrates in vacuum conditions before the coating process begins to pre-clean the surfaces removing any native oxides and particulates. For some cases, ion beam energy can be increased to provide surface texturing. Both methods promote adhesion of the evaporate.

Citations

     Don’t take our word for it – let decades of research by field experts speak for itself! Below are a few example papers describing research performed by field experts toward characterizing the exact benefits of ion assisted deposition:

Sites et. al. investigated the value of ion beam pre-clean on the substrates using Argon ions and lower current densities[2]. The results of their experiments showed much higher substrate adhesion of the coating – in fact “[b]oth the single-layer and three-layer coatings exceeded the limit of our adherence tester (10,000 psi or 6×107 N/m2).” Under adhesion testing, their substrates usually failed before the coating.

In their paper, Martin, et. al. tested several common films including TiO2 and SiO2 with and without ion assist in their deposition process[3]. Their results showed “very large changes in the spectral transmittance curves” when the substrates were exposed to air with a 50% relative humidity after reactive evaporation coating. The same coatings showed an increase in refractive index of well over 2%. Substrates deposited with IAD using Argon showed high absorption, but substrates deposited with oxygen IAD showed “much lower absorption losses … and a very large reduction in moisture absorption, indicating very high packing density.” The spectral transmittance of these substrates also changed less than the sensitivity of their instrumentation (less than 0.2%) after exposure to moisture. Multilayer coatings showed similar reductions in absorption and spectral transmittance shift.

Similarly, Al-Jumaily et. al. tested spectral shift, index of refraction, and extinction coefficient for evaporated and IAD-deposited films[4]. Their results showed consistently better extinction coefficients under IAD, and when the samples were bombarded with Oxygen instead of Argon, better transmittance / absorption as well. They also explain some of the basic physics behind why IAD is best when using Oxygen as the ionized gas. Finally, their results showed a 12% improvement in the packing density of the IAD-deposited coatings.

Research performed by Demiryont et. al. demonstrated the importance of reaching the ideal stoichiometry of thin films to optimize your index of refraction and obtain the minimum absorption coefficient[5]. Fully-oxidized (stoichiometric) films had significantly better n-values, and as much as 3 orders of magnitude lower k-values than cermet or suboxide films. In a field where a 1% variation can make or break an optic, the importance of 1000 times lower absorption cannot be overstated.

These publications demonstrate the value of appropriate ion-assisted deposition to the overall performance of an optical coating. Clearly adding energy (and energetic oxygen) to a coating during the deposition process is critical for minimizing defects, eliminating spectral shift, and optimizing stoichiometry.

Let us assist you with your assist selection!

[1] McNally, J.J., “Ion Assisted Deposition”, Handbook of Plasma Processing Technology, S.M. Rossnagel, J.J. Cuomo, W.D. Westwood, Ch. 20, pp 466-482
[2] Sites, J.R., Gilstrap, P., Rujkorakarn, R., “Ion Beam Sputter Deposition of Optical Coatings”, Optical Engineering Vol. 22 Issue 4, pp. 447-449 (1983)
[3] Martin, P.J., Macleod, H.A., Netterfield, R.P., Pacey, C.G., Sainty, W.G., “Ion-Beam-Assisted Deposition of Thin Films”, Applied Optics, Vol. 22 No. 1 (1983)
[4] Al-Jumaily, G.A., Edlou, S.M., “Optical Properties of Tantalum Pentoxide Coatings Deposited Using Ion Beam Process”, Thin Solid Films, Vol. 209 pp. 223-229 (1992)
[5] Demiryont, H., Sites, J.R., Geib, K., “Effects of Oxygen Content on the Optical Properties of Tantalum Oxide Films Deposited by Ion-Beam Sputtering”, Applied Optics, Vol. 24 No. 4 (1985)

QCM Sensitivity with IBAD

  As a deposition system manufacturer, we have the opportunity to investigate what causes process problems, and how best to fix or at least address them. When you invest in a Techne IBAD system from Plasma Process Group, you also benefit from our insight into these problems that have plagued deposition processes since their inception. Here is a piece of that insight:

Have you ever observed the QCM deposition rate dipping below zero at the beginning of a coating (see the figure below)? Is the QCM losing mass or is the measurement sensitivity just too low? The answer is neither. The QCM measures mass by differences in its resonant frequency. However, that value is also vulnerable to thermal changes. When the QCM is allowed to heat up, it records a negative mass change. This has the secondary effect of creating an offset to your recorded QCM thicknesses, which in turn can throw off the entire layer, particularly if the layer is short!

QCM Thickness Error
QCM thickness measurement during one process layer

The good news is that this effect can be calibrated out. Because the deposition rate during a layer is (very nearly) constant, the QCM measurements will make a straight line, offset by the initial (negative) effect of thermal exposure – the 1st order equation of a line.

      Thickness = (Linear Rate)x(Time) + Offset

The slope of the line comes from your existing data starting when the measured deposition rate becomes constant (derivative of the qcm measurements becomes constant). Using the data above, this is illustrated graphically in the chart below. The slope should be close to your expected
deposition rate, and will be the true deposition rate calculated by the QCM.

Calculating the Slope
Getting the slope from your QCM measurements

After calculating the slope, you can determine the offset (see chart below). This offset is how much overshoot you would have had if you ignored the thermal effects to the thickness measurement.

Measurement Offset
QCM Measurement Offset

Graphically, shifting the line upward until it passes through (0,0) will allow you to see how long the layer should have been run to meet your desired thickness (chart below). Alternatively, the layer could be terminated early by the offset thickness.

Corrected Measurement
Corrected Measurement

A QCM is capable of fantastic levels of precision. By correcting for the impact of thermal variation, your process can begin to take better advantage of the true capacity of your QCM.

QCM Control

If you run quartz crystal microbalance (QCM) controlled processes, you probably have noticed that a quartz crystal can seem very unstable, particularly on short layers. Since you rely on the accuracy of this device for the repeatability of your coatings, instabilities and noise can be a major headache and lead to inaccuracies. Selection of the appropriate crystal material makes a significant difference.

     Measurements taken on a Techne IBAD system have shown a significant improvement in crystal stability when using Alloy-coated crystals instead of Gold-coated crystals. Examine the two graphs below for a comparison.

Gold-coated Crystal
Gold coated crystal

Alloy-coated Crystal
Alloy coated crystal

Both crystals in the graphs are new and running the same process, but the gold crystal shows significantly higher instabilities and noise. The reasons for this are many; however, the primary reason lies in thermal stability. QCMs measure thickness based on resonance frequency, which is strongly effected by temperature. Each crystal-coating combination will carry a different temperature at which the thermal effects are weakest, and operating your QCM at that temperature will give the most stable and highest precision readings. At standard IBAD process temperatures between 70 to 100°C, the Alloy-coated crystal has a much better response than a standard Gold-coated crystal.

RF source care – antenna

We do a lot of cleaning and repair service at Plasma Process Group and we let a good chance to see your hardware throughout its life cycle. For the same reason, we also get to see what causes that life cycle to shorten. Here are a few quick things you can do or keep in mind to get the most out of your hardware:

A fresh RF antenna is coated with silver that improves the RF conductivity, and therefore efficiency, of the RF ion source. Over time, this silver coating will tarnish (oxidize) or erode. While the antenna will continue to work, the performance will decline slightly. Additionally, the antenna can gather debris on its surface that needs to be cleaned to prevent arcing. About once every 2 – 3 months, during a vent cycle, take a Kimwipe or other gentle cleaning cloth and wipe the coils of the antenna clean. When doing this, it is important to not push the debris into contact with the insulators holding the antenna. If the antenna has lost all of its silver coating, showing the bare copper underneath, it can be replaced entirely to maintain the performance of the ion source.

The insulators which hold the RF antenna in place may need occasional cleaning. New insulators are either pure white, or have a slight yellow tint to them. As they are exposed to debris, heat, and small amounts of plasma, they will discolor, turning copper brown or black. To clean them, remove one pair at a time and media blast them, followed by an ultrasonic bath and an isopropyl alcohol wipedown. Then reinstall that pair before removing the next. This will prevent inadvertent changes to the antenna positioning.

Ion Energy with Assist Applications

     Ion energy, measured in electron volts [eV], is an important detail in the application of ion beam technology. At certain energies, an ion striking a surface will have little or no effect on it, while at higher energies, the ion may erode or damage the surface. Balancing the appropriate ion energy for the process desired is critical to achieving a high quality coating.

Sputtering vs. Densifying

     The ion energy distribution (number of ions at a given energy) of an ion source is an important parameter that will determine what the ion achieves when it strikes the surface. For a given ion specie and substrate material combination, the energy at which sputtering starts to occur is roughly constant and does not start near 0 eV. There will be a minimum energy at which sputtering occurs, and a “probability” of sputtering that increases with increasing ion energy.

Figure 1

Figure 1: Sputter rate of Argon ions striking Silicon. Calculated from Yamamura, Y and Tawara, H, ADNDT 62 149-253 (1996)

In Figure 1, the sputter rate dependence of ion energy is shown. While these particular data are specific to the combination of Argon ion and Silicon target, the curve shape is typical of most ion / target combinations. In particular, the interest is in the low end of ion energies, where Ion Assist applications usually occur. Figure 2 is a closer look at this ion energy range.

Figure 2

Figure 2: Sputter rate of Argon ions striking a Silicon surface: focus on low energy.

From Figure 2, sputtering only begins above about 50 eV. Indeed, up to about 90 eV, only two out of every 100 ions that strike the surface will cause a sputtered atom to come off the surface (essentially, a 2% chance of sputtering). In contrast, an ion with an energy of 200 eV has a 13% chance of sputtering – much more likely. Sputtering causes the newly coated surface to wear away, loosens adhesion, and can introduce defects into the structure of the coating. Thus, for Ion Assist applications, minimizing sputtering is ideal.

Ion Energy Distributions

     To minimize sputtering, control of ion energies is important. The ion energy distribution of an ion source shows both the range of energies emitted ions can have, and the probability that a given ion will (or fraction of ions which) have a given energy. As an example, Figure 3 shows the ion energy distribution for a typical gridless Hall-Effect source.

Figure 3

Figure 3: Ion Energy Distribution for a Mark-2 ion source. Nominal voltage = 100V. Zhurin, Surf. Eng. 27 5 311-319 (2011)

From Figure 3, it is apparent that setting the emission voltage of a gridless ion source to 100V does not mean that the emitted ions have 100 eV of energy. In fact, only 75% of the ions fall between about 40 eV and 140 eV, and 25% are outside this range, and can be as low as 10 eV. About 1/3 of the ions are being emitted with very low voltage (<35 eV), and are therefore doing almost nothing for the process. This can be offset by requesting a higher voltage - for example, requesting 140V brings the minimum voltage up to around 50 eV. However, this also brings the maximum energy up to over 180 eV. Referring back to Figure 2, at this energy a significant fraction of the ions would now have a more than 10% chance of sputtering. About 60% of the ions would now have over a 5% chance of sputtering. This means many ions would now be damaging the coating instead of improving it.

In contrast, a typical ion energy distribution for a gridded source is shown in Figure 4.

Figure 4

Figure 4: Ion Energy Distribution for a gridded ion source. Nominal voltage = 150V. Rubin, et. Al. RSI 80 103506 (2009). Edited with permission.

From Figure 4, the entire energy range of the ion source is 20 eV. While not exact, if 125 eV ions are requested from a gridded ion source, the ions emitted will be between 115 eV and 135 eV. As such, to minimize sputtering a request of 90 eV could be made, with 100% of ions still contributing to the Ion Assist application, and 98% of the ions having a 3% or less chance of sputtering.

Conclusion

     Based on the above ion energy distributions, it is clear that a gridded ion source presents the most efficient and effective method of executing an Ion Assist application, with a minimum of sputtering damage. Furthermore, it is clear that a gridded ion source presents far more consistent, reliable, and predictable sputtering rates due to the narrower Ion Energy Distribution. Variation in ion energy can and does present
significant challenges in ion applications, and only gridded ion sources can offer the control needed to produce the highest quality coatings, repeatedly and with reliability.

RF source care – discharge chamber

We do a lot of cleaning and repair service at Plasma Process Group and we get a good chance to see your hardware throughout its life cycle. For the same reason, we also get to see what causes that life cycle to shorten. Here are a few quick things you can do or keep in mind to get the most out of your hardware:

The discharge chamber inside your RF ion source is made of quartz and can crack easily. We’ve all damaged a discharge chamber at one time or another, usually when the boss is standing right next to us. But if you don’t break it, can you just pretend nothing happened? Unfortunately, the answer is no. Tiny cracks are huge holes to process gas and plasma.

When process gas or plasma leak behind the discharge chamber, it increases the risk that a secondary discharge ignites. An uncontrolled discharge near the RF antenna and beaded leads may cause the source to extinguish and sputter coat the ceramic insulators with copper. Sputter coating damage can occur to the gas isolator and RF feedthru. A significant discharge may require a complete source refurbishment to recover insulator performance.

Quartz discharge chambers can be cleaned using the same process to clean molybdenum grids. Media blast pressure should be 30 psi or less. Just in case, we recommend keeping a spare discharge chamber on hand and change them out between process when necessary.

Grid cleaning

  We do a lot of cleaning and repair service at Plasma Process Group and we get a good chance to see your hardware throughout its life cycle. For the same reason, we also get to see what causes that life cycle to shorten. Here are a few quick things you can do or keep in mind to get the most out of your hardware:

  Remove the hardware from your grids before they are cleaned. The stainless steel hats are damaged very quickly if they are not removed. They will trap your cleaning media underneath and contaminate your process. Additionally, when the hats become damaged they warp and change the spacing between the grids. This may lead to arcing between the grids and cause the ion beam to wear away the grids much more quickly.

  Monitor the accelerator current. If the accelerator current looks too good to be true, it probably is. During normal operation, the accelerator current will typically be between 3 to 5% of the beam current. If it is 0 to 2 mA, the accelerator grid might not be connected and is becoming damaged. Check the accelerator electrical connections for continuity.

  These items can significantly reduce the life of the grids. For grid maintenance, send them to us for cleaning and we’ll do all the hard work. We will make sure the grids are cleaned, and correctly aligned. We might even catch a problem before more damage is done.

RFN help

  We do a lot of cleaning and repair service at Plasma Process Group and we get a good chance to see your hardware throughout its life cycle. For the same reason, we also get to see what causes that life cycle to shorten. Here are a few quick things you can do or keep in mind to get the most out of your hardware:

  RFNs are extremely vulnerable to oxygen and water. The life of the RFN collector diminishes when it is exposed to atmosphere when they are still hot. After running a process, the RFN should be cooled with Argon flowing continuously for 30 minutes prior to venting.

  For the same reason, purging your gas line before starting your process is important. Every time you vent, air and water vapor can get into the gas line and thus contaminate the RFN. We recommend flowing Argon to the RFN for 5 minutes prior to starting process. We also recommend using electropolished stainless steel gas lines from the gas bottle to the RFN. Other types of gas lines may contaminate the collector.

  Placement of your RFN makes a difference. Too far from your ion source, and the source may struggle to ignite. In the path of the deposition plume, and it will both interfere with your process and require much more frequent cleaning. Too close to your ion source, and it may be sputter eroded by the ion beam itself. The ideal location for an RFN is 8 – 10 inches away from the ion source, and on the opposite side of the ion source from the substrate fixtures.

  The RFN is not water cooled. Instead, it radiates its heat out into the chamber. While this heat has no effect on the process, improperly shielding the RFN can lead to unexpected negative effects on the RFN itself. Many people put protective foil over the RFN in an attempt to keep it from getting coated during process run. However, if the foil blocks the RFNs ability to radiate heat, it is unable to cool and will eventually overheat, which can damage the antenna and gas isolator. If you are looking to protect your RFN from coatings, consider a piece of stainless steel sheet parallel with the RFN, mounted (or otherwise attached) to the chamber wall, and two to three inches off to the side. It should extend no further than the front of the RFN Keeper so that the emission from the RFN has a clear path to the ion source. The sheet also does not need to wrap 360&deg around the RFN – just shield the exposed side to the coating.

  When the time comes for a refurbishment, send it to us and we’ll make it good as new without the expense of purchasing a new unit. Many of our refurbishment customers get over 2000 hours of RFN use between maintenance cycles.