So far I think colleagues are dealing nicely with the methods available for improving adhesion of the Au layer. But the question is actually how to determine the adhesion strength . For this the standard qualitative or quantitative methods are available. Peel off tests may appear to be simple but results can be difficult to interpret are not easy to apply. One gets into a circular question about what to attach to the 'outer' surface of the gold layer in order to perform the measurement. Adhesive tape is used of course but is not without problems and one gets data of adhesion strength between the tape and the gold rather than the gold and SiO2. Film thickness h a significant effect on the results. Adhesion strength is also about durability and this is measured with wear tests, using appropriate tip probes calculating or measuring material loss directly or indirectly. Interdiffusion adhesion takes place with evaporation of Au onto Si, and any testing of adhesion strength will need to take into account the fact that there is atomic exchange between the Au and the Si.

I am afraid I haven't given a specific answer to your Question but I hope some of the points I have made here might stimulate further investigation.

I think there is no adhesion (really poor) in this case. :-)

The adhesion is very weak. In worst cases, you could even blow the film away with mouth and not a nitrogen gun. Thats why people put a 5 nm layer of Ti or Cr in between to make the adhesion better.

Yes, I agree with Aman Haque. If deposite Au directly on SiO2, the adhesion is poor. Usually we use Ti/Au structure with thickness of 10nm/whatever thickness you want. The adhesion is good enough for lift-off in ultrasonic bath for 30 minutes.

As Prof. Aman said that the adhesion between Au and SiO2 is very weak because of the oxidation issue. I always use Cr/Ni/Ti as an adhesion layer. 5 nm Cr/Ni/Ti for 50 nm Au deposition.

That's right, adhesion is extremely poor and need a adhesion layer as already mentioned. As said by Aman Haque we are also using 5nm Ti. However pay attention if you are using aqua regia to etch Au as it will not etch Ti, resulting in a short ciruit between line. In lift-off however there are no such issue of course.

One extremely good adhesion promoting material to use between SiO2 and Au that was not mentioned yet is ITO (Indium Tin Oxide).

The adhesion is poor. For my case of coating gold to optical fiber endface, a thin chromium layer was coated before gold because of its high hardness and resistance to corrosion.

I think you can evaporate about 5 nm of Ti between them

The adhesion is bad. Use Cr or Ti layer of several tens of nanometers as an adhesive layer.

Yes, thin Ti layer (of 5 nm thickness) would be enough to improve the adhesion

I agree with all of the above and we use a thin Ti binding layers.

Of possible interest:

There are also molecular layers that have been used to promote adhesion. Have a look at Whitesides group papers with Biebuyck and with Emily Weiss (no relation). If you have trouble finding them, please let me know.

Depending on your pattern transfer process you would use Cr for etch transfer, as it is easy to etch and Ti or Cr for lift off.

I agree with Gullo, one can deposit a thin layer of Cr in between. It actually depends what one wants? For nano composite system one can also anneal at optimum temperature to improve the adhesion between Au and SiO2.

So far I think colleagues are dealing nicely with the methods available for improving adhesion of the Au layer. But the question is actually how to determine the adhesion strength . For this the standard qualitative or quantitative methods are available. Peel off tests may appear to be simple but results can be difficult to interpret are not easy to apply. One gets into a circular question about what to attach to the 'outer' surface of the gold layer in order to perform the measurement. Adhesive tape is used of course but is not without problems and one gets data of adhesion strength between the tape and the gold rather than the gold and SiO2. Film thickness h a significant effect on the results. Adhesion strength is also about durability and this is measured with wear tests, using appropriate tip probes calculating or measuring material loss directly or indirectly. Interdiffusion adhesion takes place with evaporation of Au onto Si, and any testing of adhesion strength will need to take into account the fact that there is atomic exchange between the Au and the Si.

I am afraid I haven't given a specific answer to your Question but I hope some of the points I have made here might stimulate further investigation.

to address mentioned problem by Christian Renaux you can use wet chemical etchant composed of HCl. Indeed, Cl can easily form complex with various transition metals. therefore, it can etch Au and Ti away, simultaneously.

Another way to improve interfacial adhesion of Au and Sio2 is hydrogenation of SiO2 before deposition or evaporation of Au. This is due to the especial state of d orbital of the outermost shell in Au.

Peter Rolfe nicely reminds us that the real question was how to 'measure' the adhesion. The most straightforward answer I can think of is a blister test.

Since I am not doing it; I can offer a perspective. (i) get a thin silicon coated with oxide (ii) on the back side, pattern and etch as through hole (iii) on the front side; do some good quality cleaning and then deposit the gold on the oxide (iv) Now do a HF vapor etch on the backside to have a freestanding gold film anchored by the oxide. (v) put known air pressure from the back side, watch the blister grow from the front with a microscope. The rest is modeling the air-pressure vs. free surface area increase.

If you need professional help - companies like Impulse Technology (https://www.scienceexchange.com/facilities/impulse-technology-llc) can help with this and other thin film related problems

Emad,

attached you can find a paper where an experimental study and a comparison between the adhesion of evaporated gold layers under different conditions are performed.

They explore the adhesion of gold layers on top of Si, on top of SiO2, and on top of layers that improve the adhesion.

I marked the parts of the text where they explain the adhesion in general terms and where they summarize the experimental results. I hope you find it useful.

Cheers!

I get a really good adhesion with sputtering gold on SiO2 at higher temperature, about 200 °C.

colleagues

Thanks ever so much for your helps, I really appreciate them. i am reading your comments and i find them really useful.

I just noticed a lot of obsolete methods reported to improve gold adhesion. Best method known currently (with gold layer falling in cohesion instead of adhesion as result) is the usage of silane coupling agent before gold deposition (in case of gold - mercaptane-functionalized silane coupling agent is strongest)

Adhesion layers commonly used are around 10nm thick, and mostly metallic like Cr, Ti, Ni. Can be deposited by evaporation of sputtering.

Gold is know to diffuse quite easily in materials, when looking at the nanoscale. Diffusion of several nanometer occurs within the under layer (here the substrate) even at room temperature. Therefore, any annealing can help improving the adhesion by Gold diffusion a little bit in the oxide.

Personally, I always use Cr 10nm before any Gold evaporation and it works well.

One important concern is the cleanliness of your substrate before metal deposition. How do you prepare your substrate before deposition ?

You can also increase the roughness on your glass by performing some BHF treatment. Depending on what kind of glass you are using, that is fused silica, quartz or pyrex, the results are different. Pyrex contains metallic species in abundance. So dipping in BHF leads to higher roughness than fused silica. The highest the roughness the best in term of adhesion of your Cr, Ni, Ti on it.

But sometimes, BHF can give a rough surface that is detrimental to the optical properties of your glass. Transparency can be changed. So you can replace this etching by an "oxidizing" treatment like SPM (H2SO4-H2O2) that will probably remove some metallic species while keeping SiO2.

Hope it can help …

If you need numbers for stress values you should measure it (there is a standardized equipment for it). The film has to survive (adhere to the substrate) to yield any data and this is likely to be a problem with 50 nm gold on silica. You may try thinner gold layer but reliability of stress measurements goes down with film thickness quite rapidly. Any technological of gold films is almost certainly going to use one of the metallic intermediate layers discussed here: Cr, Ti , Ni or something similar.

Lately I was involved in the development of adhesion test by buckling of thin films under compression. We managed to get reliable results for Ag/SiO2 interface and estimated its adhesion energy to be ~0.6 J/m2. The details are to appear in JMPS. In principle, the method is rather unique and Au/SiO2 can be tested as well.

Have fun!

It doesn't stick. You have to put a Ti adhesion layer. I did like 1 nm. I think you can get away with 0.5 nm. And it should work.

5 nm is overkill. 1 nm will do it. You can even do less. You should probably do a series to test it. HTH.

I did never try it, but a very simple and inexpensive method which can possibly give quantitative figures consists in using a suction cup with a strain gauge attached to it. A pull test can be performed measuring the force per unit area needed to detach the film.

Dear Emad,

the used procedure to improve the adhesion of Au is evaporate a few nm of Cr ,

Regards,

You will need to adda thin layer of either Ti or Cr in the thickness of 50-100A between the SiO2 and the gold layer for the gold film to be adhere.

You will need to adda thin layer of either Ti or Cr in the thickness of 50-100A between the SiO2 and the gold layer for the gold film to be adhere.

Just now

You can use 0.5 to 1 nm Ge as adhesion layer. I have tried it and the gold film surface is smoother that that using Cr as adhesion layer

A lot of answer discuss the way to improve adhesion between an Au layer and a SiO2 substrate. It is absolutely clear that gold will not stick to the silica layer. A metal will adhere to silicon dioxide if the heat of formation of its oxide is greater than the one of SiO2. And that why people suggest to use Cr, Ti, etc. The other way is to use a layer with wich Au will react, as in case of Ge. BUT: the Emad work is related to the problem to have a good adhesion on the silica substrate, OR to something else? certainly he knows how to improve the gold adhesion, but probably this is not what he wants.

I have an answer which I will get back on this item.

Poor adhesion between Au film and substrate is usually coursed by add-atoms adsorbed on the substrate. Oxygen, Hydrogen atoms are easily attached on a substrate but difficult to get rid of. Add-atoms trapped at the interface during Au film deposition is the main reason of poor adhesion. Back in the mid 1980s, there was an industry problem on the adhesion between Au and man-made diamond for mounting laser (or LED) on a diamond heat sink. Heat generated from LED caused de-lamination of the gold from diamond and causing a burnout of LED. This adhesion problem between Au and diamond was finally solved by sputtering the diamond substrate with Ar ions, and do Au deposition on diamond in-situ without breaking vacuum (U.S. Patent 5068020*). I know that diamond is different from SiO2 substrate, however, I wonder if the problem is the same due to add-atom at the surface of the SiO2. I suggest using the in-situ sputtering cleaning may work.

*also listed in my publication in RG.

There may be some confusion over your question among colleagues but If I understood your question correctly you are looking for a convenient technique which can give you simple understandable "numbers" for adhesive strength of such a film. Try Pull off test with standards like ASTM D4541. There is a low cost instrument, "Positest AT-Manual or Auto" from Defelsko Corporation in NY (I am sure there will be many more companies with similar products). You also need to find a suitable adhesive (Araldite 2012 etc.) which must surpass the adhesive strength of your film for the analysis.

You should evaporate about 5 nm of Titanium between them

In the Au on diamond adhesion case mentioned, Titanium inter layer did not work. W, Ta did not work. Ti, Ta, and W may work on Si, but not on diamond. Diamond is chemically inner comparing to Si or other common substrate. Ar sputtering removed the ad-atoms such as H, O, or N, and in addition, it will mess up the diamond surface such that many carbon dangling bonds existing. Au or other metals deposited in situ on this surface will easily form metal carbides at the interface. This is why the prescribed methods work. I tried to remove the Au from diamond by mechanical means on purpose but failed. The interface bonding is stronger than the diamond-diamond bond. The key to interface integrity is: (1) clean interface, free from ad-atoms, (2) having dangling bonds to promote intimate chemical interaction with the top layer. In-situ Ar sputtering is a simple way to accomplish both conditions.

Dear Emad Esmaeili,

i can't provide you number as a strain or force per area because I've never had a need for my purposes but I can give you a technique which at the end of the process you can get them alone with the opportune measurements.

If you want to have only gold on SiO2 without stack of Cr or Ti or Al as adhesion layer before of gold deposition, try to use primer (i use in the past hexa-methyl-disilazane) into a vacuum chamber. You have to be careful because is a little bit critical and delicate as a process, because immediately after the sample is out of the primer vacuum chamber, you have to insert the sample into the evaporator (if you use the evaporator obvious). The time of the exposition to the primer depends by the thickness of gold that you want to deposit, in the past have been able to evaporate 50 nm of gold above 1.2um native SiO2 with this technique (exposition to the Primer 25 minutes).

My kindest regards and good luck for all.

Andrea Lucibello

Recent replies are mostly about very useful strategies for improving adhesion. Together with measures of adhesion strength there could be nice opportunities to do some collaborative research on this important topic. To test any of the suggested methods for improving adhesion it is vital to have methods for measuring adhesion. It would also be worth while organising a seminar to focus on this at an upcoming conference.

Regards, Peter

I m agree with Dr Naseem Abdel, so we can use Ti or Cr

Basically, Cr, Ti, Ni with thickness of 5nm to 10nm may be suitable.

You cannot deposit Au directly on SiO2 layer. The gold film will de-iminate. You need first to deposit a very thin layer, 50-100A of Cr or Ti as an ahesion layer for the gold.

CCL

thank you for all concern to these valuable information ,

I inquiry , if this technique is applied generally for any thin film or specially for SIO2 ?

When you deposit Au on SiO you will get complex structure comprised in different layers because Au penetrate SiO and create AuSiO layers compound (Au3+, Au2+, Au+) . with increasing Au deposition you will have Au layer . Then the adhesion is AuSiO layers.

Though 50 nm thick Au films with reasonably smooth surface can be deposited on SiO2 layers/ substrates, delamination of Au films from SiO2 poses the main problem. You should deposit a thin film (~ 5nm for ensuring continuous film) of Cr or Ti on SiO2 layers/ substrates prior to deposition of Au films.

Second with Dr. Peter Rolfe. by the way, what is the point of measuring such weak connection between Au and SiOx. adding an intermediate layer appears improving the adhesion. I suggest to maybe give it a bit annealing to strengthen the connections.

Normally in MEMS fabrications they use Titanium (Ti) as adhesion layer between the isolation layer like SiO2 and metals.

As I stated previously.  You need to first deposited a thin layer 50 A of Ti or Chromium before you deposit the gold film.  Gold does not adhere to iO2 directly, and you will need an metallic inter-layer to provide the adhesion.

C.C. Liu

Depending on the fabrication and method of preparation  the  evaporation Au to SiO2 can be accompanied with complexes structure. Therefore, it fabricate complex layers   

Chung Chiun Liu Thanks for your answer... I already tried Au without Cr or Ti after e-beam lithography during field effect fabrication process. What i get is simply dirty contacts and also some of my sample disappear ... 

Did you try to deposit Ti/Cr in between Au and SiO? If yes, did you overcome the problem as mentioned in your last post. This week I am planning to evaporate Au on Cr. Initially, I will sputter Cr on thermally oxidized Si. I am using stainless shield mask with pattern for depositing gold. What method would you suggest to deposit Cr? 

If you do not want any other metal in between you could use some silane to improve the adhesion of Au on Si substrate (actually on its oxide).

With only gold you could get a perfect layer only above 12nm of thickness.

Using these adhesion layers of silane you could get that down to 5nm.

You would have a dielectric of about 1-2nm between Si and Au, though.

http://pubs.acs.org/doi/abs/10.1021/am508681u

Yes I agree with Prof. Aman. The adhesion between Au and SiO2 is very poor. You can deposit 5 nm of Cr/Ti before depositing 50 nm of Au.

Dr. Shuja Khan,

I always use ebeam evaporator to first deposit 5nm of Ti and then 50 nm (or 80 nm) of Au on SiO2. I get pretty good adhesion.

Here is a comerical version as an example:

https://www.phasis.ch/products/bionano-gold