SPI Supplies^® Brand Silicon Nitride Membrane Window Grids for TEM

Try this "ultimate substrate" for the ultimate in TEM support materials

Membrane window thickness of 100 nm Si₃N₄
with surrounding silicon support of 200 µm.
Total dimensions diagonally in figure to the left is 3.0 mm.
Window examples shown above: 0.5 mm and outside square 0.75 mm

Click to see other views of this useful and innovative TEM support sample. Generally speaking, the specimen will be applied to the opposite side of the grid because for AFM work, the cantilever would not fit into the etch pit. However, the view from the opposite side is much less interesting in terms of communicating the details of the product.

New products being developed:
Because of the success of the original line of SPI Supplies Brand Silicon Nitride Membrane window grids, we now have under development other types of membrane window grids for TEM applications including perforated membranes for the growing of nanotubes and other applications.

The original product line: These unique membranes of silicon nitride (Si₃N₄) were made using a highly innovative approach. Normal electronics industry grade silicon waters were obtained and a thin film of Si₃N₄ was deposited to the desired thickness. Presently, we have standardized with seven different thicknesses of the nitride, namely 200, 150, 100, 75, 50, 30, and now 20 nm.

The next step is to back etch through the silicon wafer to the silicon nitride layer, thereby producing the "membrane window" effect. And while the membranes are very thin and fragile, they are sufficiently "robust" to permit one to do their experiments. The product is meant for a single use and we would not recommend trying to clean off old samples for reuse. Discriminating between the two sides with the naked eye is quite easy, the membrane side vs. the etch pit side are quite easy to tell apart. The membrane side is completely smooth and even whereas the back side has the etch pit in it.

At any of these thicknesses, for TEM applications there is not enough material to contribute to a diffraction pattern and the membrane itself is completely structureless and featureless. Hence, anything seen, either by diffraction or in terms of the image itself can be attributed solely to the sample and not to the support structure. Outstanding results are routinely obtained at 200 KV and presently, the membranes are being evaluated for use at 500 KV.

Details of the nitride membrane material:
This has been an area of some amount of confusion and our goal here is to clarify the precise nature of the SPI membrane window grids. Our original technology permitted the manufacturing of what in the industry is called "stoichiometric" Si₃N₄ and for the first year or so offering this particular product line, this is indeed what we were producing. However, stoichiometric or "ST" membranes have a relatively high residual intrinsic stress, and that contributes to a less robust film and one that is more easily ruptured. But with the passing of time, as well as gaining much additional experience, we have perfected the technology to produce "low stress" silicon nitride membranes. These membranes have a Si:N ratio greater than 3:4, that is, Si is greater than 3. And the end result is quite dramatic. At first, we offered the low stress, or LS now sometimes called "SiN", membranes at a premium price relative to the ST products, but we have increased our yields to the point that production costs are comparable, and since we think of the LS products as being superior in every possible way for these kinds of applications, we now automatically are offering LS membranes when one places an order for these products. We will continue to make the ST membranes upon request, however, we are certain that for most users, the LS membranes will give superior results.

Why would anyone ever find the ST membranes, the type we used to make, preferable to the LS membranes we now produce as standard? We are not aware of any reason why the ST membranes would be more attractive for the TEM or x -ray microscopy application. However, if one is taking the membranes, perhaps as entire wafers, and sputtering onto them, something that can be inherently a compressive situation, then the higher stress in the nitride layer could in fact be a real advantage. But in general, we believe that most users will find the new and improved LS membranes to represent an improvement of an already outstanding product.

Just remember that all silicon nitride membranes are of the low stress or SiN type except for the SPI Supplies® Brand Silicon Nitride Membrane Window Grids for TEM are of the LS type except for those with membranes 20 nm in thickness. Of course, quite possibly one might want 20 nm in LS instead of ST nitride, we can make it on special order but the "default" product listed on this website page is ST.

In addition, we can provide some general information about the electrical resistivity values for these membrane window products. And in case you were wondering, the "right" side for the sample is usually the "membrane side" of the grid as opposed to the "etch pit side". However, we have heard of the etch pit side being used in special situations.

Who would find this product useful in their work?

Actually there are a number of instances where the SPI Silicon Nitride Membrane Windows could be useful, some times making possible the otherwise impossible:

An inert substrate that can be used at elevated temperatures to observe dynamically, reactions in situ by either TEM or SEM or in some instances, by AFM.
Use as a durable (e.g. "robust") substrate for doing "matched pairs", of the same identical area, first by TEM and then by SEM.
Use as a durable substrate for matched pairs comparing the AFM vs TEM image.
Use as "wet cells" for both SEM BSE imaging and TEM transmission imaging

Just about anyone studying nanoparticles would find these membrane window grids to be indispensable for their work. Those studying aerogels and/or xerogels, because of the smallness in size of the basic particles, find the SPI silicon nitride membrane window grids especially valuable for their work.

Advantages:

For SEM applications, the background is structureless and featureless.
For x-ray microscopy, there is really no other way for mounting many of the samples one would want to analyze.
For SEM BSE imaging, cells can be grown directly onto the nitride windows, and the volume sealed with a "blank" without membrane for a perfect UHV compatible environmental chamber.

Deeply grooved grain boundary in which parts have reached
the substrate, 400°C. The long low contrast area running
from the upper left to the lower right is the amorphous
silicon nitride membrane onto which the films were deposited
(between the two large central grains).

Film that de-wetted the substrate in the solid state at 615°C.
Both micrographs were taken from publication, In-situ
TEM observations of abnormal grain growth coarsening, and
substrate de-wetting in nanocrystalline Ag thin films.

For high temperature applications, we believe that the nitride membranes are stable up to at least 1000°C. We are happy to report some new information that validates the robust nature of the most recent batches of the membrane windows. We are also happy to report that a growing number of researchers are starting to report some rather interesting results obtained using these membrane window grids. Morphological information can easily be obtained as a function of annealing time without concern about breakage of the membrane (within limits, of course!).

Presently, the SPI Silicon Nitride Membrane Windows are available in standard thicknesses from 500 nm down to 20 nm. In some instances, when a smaller window can be tolerated, it is possible to make membrane thicknesses down to 10 nm. Once made the membranes can be handled as would any TEM grid, with a good pair of high precision tweezers.

Are the membrane windows hydrophilic?
The grids, as manufactured tend to be hydrophobic, so if samples are being applied from an aqueous suspension, this can cause suspended nanoparticles to appear in a nonuniform distribution across the membrane. The membranes can be made hydrophilic by treatment in a plasma etcher, such as the SPI Plasma Prep II Plasma Etcher, however the effect of the treatment is transitory, and while we have not measured it precisely, we expect that it is along the lines of longevity as carbon coated TEM grids that are treated in the same way, also to make them hydrophilic. So it is possible for us to treat the membrane window grids in this way, however we can not guarantee the longevity of the effect. This could be one of those instances where one could make the case for having this kind of etching unit handy, in their own laboratory.

Optical transparency:
We are often times asked about the optical transparency of the membrane windows. We have not yet ourselves made such measurements, although we hope to do that soon. But visually, one can indeed "see through" the windows, for example, if you view them by transmission light microscopy. There has to be some thickness beyond which the optical transparency starts to decline , but even at 200 nm thickness, they are quite transparent in the visible range of the spectrum. When considering optical properties, keep in mind that the absorption edge is just below 13 nm. Hence, at 13 nm, the transmission for a 100 nm membrane is about 44%, but at 12.4 nm, it drops down to about 13%. So one should always keep in mind where they are relative to the absorption edge when trying to estimate membrane optical properties.

Flatness and roughness of the membrane:
We assume that our membranes are as flat as the underlying silicon wafers. But in terms of roughness, our specification for the starting wafer is to be better than 0.5 nm RMS but it is not a stated specification for roughness for the final membrane. We have had reports of customers doing their own roughness measurements, some reporting as good as 0.2 nm RMS and we believe that is probably typical (but without having made exhaustive measurements of our own). Although this roughness is more than good enough for most users, much to our frustration, it is not good enough for everyone.

But we have been led to believe that the flatness is pretty good for at least most users. We are sometimes asked about the flatness on the "well" side of the grid and we assume that the flatness is similar on both sides. However, we can't measure the "well" side flatness because we could not get a SPM probe down into the well. However, since we do believe that the flatness closely mirrors the flatness of the underlying silicon wafer, then the well side can't be any more rough than the opposite side, it could only be smoother.

Window size:
The "standard" window size is 0.5 mm square and this dimension is more than enough for most TEM workers. A larger membrane window is never going to be as robust as a smaller window. So our advice is never to get a window larger than what is needed. But we have had some requests for larger windows for TEM and we can now off at least two standard products that have a 1 mm square window.

But we advise against even considering still larger membranes for the TEM application for two reasons:
First, and the most important, the amount of frame left is so little that the grids could "snap" when being broken out of the wafer into individual grids and secondly, it would difficult indeed to pick up a grid with such a small frame with a pair of tweezers without damaging the window. We can make larger windows but not in a TEM grid product.

Selection of window size:
Those considering larger than the standard (e.g. 0.5 mm) windows seem to be those who want a larger field of view. So for those we offer the 1 mm window products. For those contemplating processing, possibly to the extent that such processing could rupture the window, smaller windows are often times desired. All of the SPI Supplies silicon nitride membrane window grid products can be easily spin coated provided arrays of membranes are being coated.

Ultra thin membrane windows:
We are quite excited about our 20 nm windows. These represent the very latest in production technology for membrane window grids. The membrane itself is "stoichiometric" instead of "low stress" but it does not really matter anyhow since the stress is so low. But any way you look at it, 20 nm is very thin, so thin that we would hesitate to guarantee to anyone that the 20 nm windows could be spin coated or exposed to various other forms of processing without rupturing the membrane. We do know that the membranes are sufficiently robust to perform as a TEM grid, with the smaller size windows being more robust than the larger ones. We cannot guarantee that the 20 nm windows would be completely free of pin-hole defects (but for TEM applications, this should not matter anyhow.

Now here!:
Because of a number of customer requests, we are developing a parallel product line that will have SiO₂ instead of Si(Si₃N₄ membrane windows. And the first silicon oxide membrane products in this new family are now available for immediate shipment from stock. We also have available from stock a perforated silicon oxide membrane window grid.

Our expectation is that we would expect the silicon oxide membrane window grids to be less stable in some ways relative to the nitride product. For example, the oxide is somewhat compressive so it will be less flat. Also, from a chemical standpoint, the nitride is inert just about to everything except free fluorine, but the oxide is reactive to quite a few different chemical species.

Cells adhered to membrane imaged in a routine
tungsten source SEM. Cell processes are clearly visible
at the periphery of the cells. Micrograph width is 30 µm.

Wet Cell Kits now Available
The SPI Supplies Brand of silicon nitride membrane window grids can be used to create a unique environmental chamber or "wet cells" which permit the BSE imaging of cells and other wet samples at resolutions heretofore just not possible.

Cleaning of membranes before use:
The silicon nitride membrane window grids should not need further cleaning before use. There are occasionally small pieces of nitride scattered around from the snap out corners (that make the frames rounded) and the presence of such debris particles can't be prevented, but at the same time, should not pose any problems or interference to the user.

However, if one wants to clean them for some other reason(s), then we suggest using H₂ SO₄ : H₂O₂ (1:1) for organic and H₂O:HCl: H₂O₂ (5:3:3) for metals. The membranes are in general not compatible with ultrasonic treatments to assist with the cleaning, as the windows have a tendency to shatter.

Adhesion of coatings to the membrane surface:
All the commonly used coating methods can be used on the membrane but there can be adhesion problems, not necessarily because they are membranes, but just because they are normal surfaces with the properties of normal surfaces. One can use vacuum evaporation, sputter coating, or spin coating without undo problems. In the case of gold coating, we normally apply a thin layer of chromium to enhance adhesion.

Holey membrane window grids
We have also successfully made our first holey membranes using micro fabrication techniques. Such grids make possible the depositing of nanofibers or other samples across the holes so that data can be taken without any contribution from the support being present in the final data. Putting holes into the membrane increases the price typically by 2-3 times, depending on the number of holes needed.

All orders for the SPI Silicon Nitride Membrane Windows are shipped in an SPI Slide-A-Grid storage box and is included in the price.

Membrane windows for spin coating:
Very often, as part of the larger protocol for using these membrane window grids, the membrane side has to be spin coated, using for example, the widely popular table-top spin coater offered by SPI Supplies, the Model KW-4A. For those doing spin-coating, we can offer these same membranes in the form of a "multi-frame array". Hence a large number of the membranes can be coated at one time and then each frame snapped out individually when time for its further processing and use.

Thickness of the "frames" (e.g. silicon support):
The Silicon support structure, made from a silicon wafer, is 200 µm standard. Silicon support thickness 381µm is available on special request and at slightly increased cost. We have also started producing a 20 nm window but the nitride is stoichiometric (ST) instead of low stress (LS) nitride for technical reasons. Also, We are now starting to offer these grids on thinner wafers, that would be only 100 µm thick, mainly for use in some of the newest TEMs that can not take a 200 µm thick grid. We have also introduced our UltraThin frame thickness, only 50 µm, designed for use with the newest HRTEMs.

Refractive index (RI) values of the nitride membrane:
Our target is for an RI of approximately 2.15. We do not guarantee the refractive index for the TEM grids since in most instances, refractive index has no real significance anyhow. But if you believe you require tighter control over RI, we can of course exert better control, but it will require a substantially larger minimum purchase and also, as a result of the tighter controls, it will require a higher pricing level as well.

Availability of silicon nitride coated wafers:
We are now able to offer to our customers the same silicon nitride coated wafers used by SPI Supplies for the production of the membrane window product line.

We can also custom make entire wafers containing windows of special custom made design.

What other customers order when ordering silicon nitride membrane window grids:
A substantial number of membrane grid customer order tweezers (to safely pick them up). and additional grid storage boxes for storing the prepared grids.

Silicon
Silicon [CAS #7440-21-3] 200 µm Thick Frames (Window Size: 0.5 mm): Packs of 10
Thickness of membrane window	SPI #	Each Pack	Each Pack, 10+	In Stock
500 nm	4109SN-BA	$143.91	$129.52	Yes
200 nm	4120SN-BA	143.91	129.52	Yes
150 nm	4121SN-BA	143.91	129.52	Yes
100 nm	4122SN-BA	143.91	129.52	Yes
75 nm	4123SN-BA	179.62	161.66	Yes
50 nm	4124SN-BA	215.67	194.10	Yes
30 nm	4125SN-BA	287.57	258.81	Yes
20 nm	4159SN-BA	432.34	389.11	Yes

200 µm Thick Frames (Window size: 1.0 mm): Packs of 10
150 nm	4161SN-BA	143.91	129.52	No
100 nm	4112SN-BA	143.91	129.52	Yes
50 nm	4135SN-BA	192.37	173.13	Yes
30 nm	4162SN-BA	304.78	274.30	Yes

100 µm Thick Frames (Window Size: 0.5 mm): Packs of 10
100 nm	4131SN-BA	182.85	164.57	Yes
50 nm	4132SN-BA	274.28	246.85	Yes
30 nm	4192SN-BA	399.13		Yes

50 µm Thick (UltraThin) Frames (Window Size: 0.5 mm): Pack of 100
100 nm	4160SN-MB	2418.71	2176.84	No

200 µm Thick Frames (Window Size: 0.25 mm): Packs of 10
Thickness of membrane window	SPI #	Each Pack	Each Pack, 10+	In Stock
500 nm	4098SN-BA	$143.91	$129.52	Yes
200 nm	4099SN-BA	143.91	129.52	Yes
150 nm	4100SN-BA	143.91	129.52	Yes
100 nm	4101SN-BA	143.91	129.52	Yes
75 nm	4102SN-BA	215.88	194.29	Yes
50 nm	4103SN-BA	215.88	194.29	Yes
30 nm	4104SN-BA	287.86	259.07	Yes
20 nm	4105SN-BA	432.34	389.11	Yes

200 µm Thick Frames (Window Size: 0.10 mm): Packs of 10
Thickness of membrane window	SPI #	Each Pack	Each Pack, 10+	In Stock
500 nm	4091SN-BA	$ 143.91	$ 129.52	Yes
200 nm	4092SN-BA	143.91	129.52	Yes
150 nm	4093SN-BA	143.91	129.52	Yes
100 nm	4094SN-BA	143.91	129.52	Yes
75 nm	4095SN-BA	228.55	205.70	Yes
50 nm	4096SN-BA	228.55	205.70	Yes
30 nm	4097SN-BA	304.78	274.30	Yes
20 nm	4107SN-BA	304.78	274.30	Yes

200 µm Thick Frames (Window Size: 0.05 mm/50 µm): Packs of 10
Thickness of membrane window	SPI #	Each Pack	Each Pack, 10+	In Stock
30 nm	4090SN-BA	$ 304.78	$ 274.30	Yes
20 nm	4163SN-BA	304.78	274.30	Yes
100 nm	4088SN-BA	304.78	274.30	No

Note: One pack of ten gives ten membrane windows, ten packs of ten give a total of 100 membrane windows.

To Ask a Question or Make a Comment

To Place an Order or Request a Quote

Return to:

Wednesday May 14, 2008 © Copyright 1998 - 2008. By Structure Probe, Inc.
Contacting SPI Supplies and Structure Probe, Inc.
All rights reserved.
All trademarks and trade names are the property of their respective owners.
Privacy Policy

Worldwide Distributors, Representatives, and Agents Flag logo