[OT] Scanning electron microscope display

This post inspired by recent posts by Jeff Miller.

My questions include

Does anyone have good links to descriptions of:
a. scanning electron microscopes in general?
    I've found Google searches for "SEM" or "scanning electron
    microscope" to be rather non-productive.
b. SEM detection and display schemes?

My only explicit qualification to ask questions is that
 >30 years ago, I was an instrument tech in a university setting.

I think my most basic question is why a fast 16 bit A/D?

When you have a CRT display, is more than 8 or 9 bits really useful?

A secondary question revolves around using scintillation(sp?) followed 
by a photomultiplier.

Why not use a "channel electron multiplier" in count mode?

Also the problem of "image tearing" was mentioned in one post.
Would not that be alleviated by synchronizing display and offsetting of 
scan?

I assume I've demonstrated my naivety ;]

Richard Owlett wrote:

> This post inspired by recent posts by Jeff Miller.
> 
> My questions include
> 
> Does anyone have good links to descriptions of:
> a. scanning electron microscopes in general?
>    I've found Google searches for "SEM" or "scanning electron
>    microscope" to be rather non-productive.

Try google, it's too bad I didn't catch this question last night: I 
found some pretty good ones.

> b. SEM detection and display schemes?

Most common detection is scintillator/PMT for secondary electrons. Two 
factors affect the contrast of a secondary electron image: topography 
(shadow) but also secondary emmision ratio varies depending on material: 
from about .5 to 2 secondarys per incident electron. Perhaps next most 
common is X-ray spectrometry, which is very good for distinguishing 
materials based on atomic number. Usually requires LN2 to reduce noise 
to point of getting a usefull signal. Often used for false colorizing. 
Back-scattered electron (BSE) detectors can also distinguish materials I 
think. Typically silicon. There are a few others.

Traditionally, the images are displayed on a CRT with a long-decay 
phosphor. At high mags, refresh time can be 2 seconds or more: very 
annoying. Typically 2 CRT's are provided, one with a camera mount: 
ultimatley you don't _really_ get the "big picture" until your film is 
developed.

It's critical to understand that in all cases, the CRT deflection coils 
are driven in sync with the SEM scanning coils: practically in parallel.

> 
> My only explicit qualification to ask questions is that
>  >30 years ago, I was an instrument tech in a university setting.
> 
> I think my most basic question is why a fast 16 bit A/D?

I've posted to this group and cross-posted to several others asking 
about S/N of various media, we'll see what turns up. There's always the 
rationale that the extra information can be used for post-aquisition 
processing, which would bring up details that wouldn't otherwise be 
visible. What's that called: histogramtic edge enhancement, for example.
> 
> When you have a CRT display, is more than 8 or 9 bits really useful?
> 
> A secondary question revolves around using scintillation(sp?) followed 
> by a photomultiplier.
> 
> Why not use a "channel electron multiplier" in count mode?

Contamination. Typical vacuum in the chamber is 10-5 to 10-6, and 
traditionally it is very dirty with diffusion pump oil. This not only 
coats the surfaces of the EM but polymerizes on exposure to electron 
bombardment. Nevertheless, I have a strong background in high vaccum and 
am confident I can get a clean (ie: dominated by water vapor) vacuum of 
10-8, and will be trying this very thing. That's one of the reasons I 
want a 16 bit A/D: having heard that the analog front ends of SEM's are 
generally accepted as having 72 dB SNR, I'd like some headroom because I 
want to figure out where the limitation is with expereiments like this.
> 
> Also the problem of "image tearing" was mentioned in one post.
> Would not that be alleviated by synchronizing display and offsetting of 
> scan?

Traditionally this is no good because the refresh rate is so slow. But 
ultimatley there's no (built in) way to know the sample has moved, as 
the sample is moved mechanically.

> 
> I assume I've demonstrated my naivety ;]
> 
> 

Also look for information about homebrew STM, and try to find the 
company in Germany selling kits for about $1K euros. I'll definitely be 
building one "soon".

I suppose this discussion belongs in sci.microscopy. They're lonely over 
there.

-Jeff

jeff miller wrote:

   ...

> ultimatley you don't _really_ get the "big picture" until your film is 
> developed.

Polaroid?

> It's critical to understand that in all cases, the CRT deflection coils 
> are driven in sync with the SEM scanning coils: practically in parallel.

   ...

> ... There's always the 
> rationale that the extra information can be used for post-aquisition 
> processing, which would bring up details that wouldn't otherwise be 
> visible. What's that called: histogramtic edge enhancement, for example.

YES!

>> When you have a CRT display, is more than 8 or 9 bits really useful?

I doubt it.

>  ... Typical vacuum in the chamber is 10-5 to 10-6, and 
> traditionally it is very dirty with diffusion pump oil. 

Good Lord! Don't SEMs use cold caps and nitrogen traps? Even Freon coils 
from an old freezer is a great deal better than nothing.

   ...

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

jeff miller wrote:
> Richard Owlett wrote:
> 
>> This post inspired by recent posts by Jeff Miller.
>>
>> My questions include
>>
>> Does anyone have good links to descriptions of:
>> a. scanning electron microscopes in general?
>>    I've found Google searches for "SEM" or "scanning electron
>>    microscope" to be rather non-productive.
> 
> 
> Try google, it's too bad I didn't catch this question last night: I 
> found some pretty good ones.

When I Googled I got many hits. It turned somewhat [while mangling a 
metaphor] into "not seeing forest for the leaves, let alone the trees".

I suspect browsing sci.microscopy, as you suggested, may answer my 
underlying questions for which I don't have background to frame



> 
>> b. SEM detection and display schemes?
> 
> [ snip description of what incident electrons produce ]
> 
> Traditionally, the images are displayed on a CRT with a long-decay 
> phosphor. At high mags, refresh time can be 2 seconds or more: very 
> annoying. Typically 2 CRT's are provided, one with a camera mount: 
> ultimatley you don't _really_ get the "big picture" until your film is 
> developed.
> 

Which is the driving force behind going digital. You can see large 
high-resolution images in seconds rather than hours.

> It's critical to understand that in all cases, the CRT deflection coils 
> are driven in sync with the SEM scanning coils: practically in parallel.
> 

Understood.

>>
>> My only explicit qualification to ask questions is that
>>  >30 years ago, I was an instrument tech in a university setting.
>>
>> I think my most basic question is why a fast 16 bit A/D?
> 
> 
> I've posted to this group and cross-posted to several others asking 
> about S/N of various media, we'll see what turns up. There's always the 
> rationale that the extra information can be used for post-aquisition 
> processing, which would bring up details that wouldn't otherwise be 
> visible. What's that called: histogramtic edge enhancement, for example.
> 
>>
>> When you have a CRT display, is more than 8 or 9 bits really useful?

I'm following your thread on this topic.

>>
>> A secondary question revolves around using scintillation(sp?) followed 
>> by a photomultiplier.
>>
>> Why not use a "channel electron multiplier" in count mode?
> 
> 
> Contamination. Typical vacuum in the chamber is 10-5 to 10-6, and 
> traditionally it is very dirty with diffusion pump oil. 

Brings back memories. We had a couple of old mass specs that my boss had 
retrofitted with ion pumps. The diffusion pumps were used essentially as 
roughing pumps. They had LN2 traps. But I don't recall if they were 
always used.



> [snip channel multiplier problems]
>> Also the problem of "image tearing" was mentioned in one post.
>> Would not that be alleviated by synchronizing display and offsetting 
>> of scan?
> 
> 
> Traditionally this is no good because the refresh rate is so slow. But 
> ultimatley there's no (built in) way to know the sample has moved, as 
> the sample is moved mechanically.

Is that still a problem in current designs?
I can think of a couple of ways to detect "when"/"how far" sample holder 
has moved, unaffected by platform being move mechanically with attendant 
"slop".


> Also look for information about homebrew STM, and try to find the 
> company in Germany selling kits for about $1K euros. 

That sounds *CHEAP*!
How much "infrastructure" ( pumps, electronics, etc ) does that include? 
That could be affordable to tech oriented high schools. In the 60's, 
when Varian introduced 60 MHz NMRs with permanent magnets, I thought 
that had impressive possibilities.

> I'll definitely be building one "soon".

Not I. I've purchased a 100+ year old house which has undergone strange 
improvements.


> 
> I suppose this discussion belongs in sci.microscopy. They're lonely over 
> there.
> 

I may keep them company ;)


Thanks for your reply.

Jerry Avins wrote:

>>  ... Typical vacuum in the chamber is 10-5 to 10-6, and traditionally 
>> it is very dirty with diffusion pump oil. 
> 
> 
> Good Lord! Don't SEMs use cold caps and nitrogen traps? Even Freon coils 
> from an old freezer is a great deal better than nothing.
> 

How about trap chilled by cold tap water.
Actually a somewhat serious question as I recall the heater of an oil 
diffusion pump is rather "warm" ;]

[ may I claim out of the mouths of ....  ? ]

Richard Owlett wrote:
> Jerry Avins wrote:
> 
>>>  ... Typical vacuum in the chamber is 10-5 to 10-6, and traditionally 
>>> it is very dirty with diffusion pump oil. 
>>
>>
>>
>> Good Lord! Don't SEMs use cold caps and nitrogen traps? Even Freon 
>> coils from an old freezer is a great deal better than nothing.
>>
> 
> How about trap chilled by cold tap water.
> Actually a somewhat serious question as I recall the heater of an oil 
> diffusion pump is rather "warm" ;]
> 
> [ may I claim out of the mouths of ....  ? ]

I used to keep a water-cooled cold cap right over a 10-inch (no typo) 
diffusion pump to keep heater radiation from quickly boiling off the 
nitrogen in the chevron-baffle oil trap. That was some pump! It could 
handle 10^-3 without choking, and it pulled a 20"-diameter, 24"-high 
working volume from 10^-4 to 2x10-^7 in about 4 minutes. The pressure 
never went above 3x10^5 when we were evaporating gold, tin, nickel, or 
even lead and chrome. It took 12 hours to get down to 10^-8, though.

To answer your question, I suppose that ice water would reduce oil 
contamination in the work chamber. but it won't keep it low enough for 
many applications. Vapor pressure goes with something like the 4th to 
7th power of absolute temperature. Liquid nitrogen is great. The lowest 
one can reach with commercial refrigerants is helpful.

I used cold brine in my home-brew mirror-coating system. I can't say 
that it helped. "Why do you wear that amulet?" "To keep tigers away" 
"But there no tigers anywhere on this continent!" "See? it works!"

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

jeff miller wrote:

> 
> I suppose this discussion belongs in sci.microscopy. They're lonely over 
> there.
> 
> -Jeff
> 

I couldn't find that group ;{

Jerry Avins wrote:
> jeff miller wrote:
> 
>   ...
> 
>> ultimatley you don't _really_ get the "big picture" until your film is 
>> developed.
> 
> 
> Polaroid?
> 
>> It's critical to understand that in all cases, the CRT deflection 
>> coils are driven in sync with the SEM scanning coils: practically in 
>> parallel.
> 
> 
>   ...
> 
>> ... There's always the rationale that the extra information can be 
>> used for post-aquisition processing, which would bring up details that 
>> wouldn't otherwise be visible. What's that called: histogramtic edge 
>> enhancement, for example.
> 
> 
> YES!
> 
>>> When you have a CRT display, is more than 8 or 9 bits really useful?
> 
> 
> I doubt it.
> 
>>  ... Typical vacuum in the chamber is 10-5 to 10-6, and traditionally 
>> it is very dirty with diffusion pump oil. 
> 
> 
> Good Lord! Don't SEMs use cold caps and nitrogen traps? Even Freon coils 
> from an old freezer is a great deal better than nothing.
> 
>   ...
> 
> Jerry

Many use turbos these days, though vibration can be a problem. I've 
never seen a trap on an SEM, but I'm sure it's done for special 
purposes. The real problem is they never bake out.

But it seems the vacuum just isn't a problem. All the rage these days 
are scopes that operate at 10-1 to 10-3 range. They have at least one 
pump on the column near the filament, often several pumps one between 
each aperture, so that the vacuum is good near the cathode (filament) 
which is all that really matters.

-Jeff

Richard Owlett wrote:

> jeff miller wrote:
> 
>> Richard Owlett wrote:
>>
>>> This post inspired by recent posts by Jeff Miller.
>>>
>>> My questions include
>>>
>>> Does anyone have good links to descriptions of:
>>> a. scanning electron microscopes in general?
>>>    I've found Google searches for "SEM" or "scanning electron
>>>    microscope" to be rather non-productive.
>>
>>
>>
>> Try google, it's too bad I didn't catch this question last night: I 
>> found some pretty good ones.
> 
> 
> When I Googled I got many hits. It turned somewhat [while mangling a 
> metaphor] into "not seeing forest for the leaves, let alone the trees".
> 
> I suspect browsing sci.microscopy, as you suggested, may answer my 
> underlying questions for which I don't have background to frame
> 
> 
> 
>>
>>> b. SEM detection and display schemes?
>>
>>
>> [ snip description of what incident electrons produce ]
>>
>> Traditionally, the images are displayed on a CRT with a long-decay 
>> phosphor. At high mags, refresh time can be 2 seconds or more: very 
>> annoying. Typically 2 CRT's are provided, one with a camera mount: 
>> ultimatley you don't _really_ get the "big picture" until your film is 
>> developed.
>>
> 
> Which is the driving force behind going digital. You can see large 
> high-resolution images in seconds rather than hours.
> 
>> It's critical to understand that in all cases, the CRT deflection 
>> coils are driven in sync with the SEM scanning coils: practically in 
>> parallel.
>>
> 
> Understood.
> 
>>>
>>> My only explicit qualification to ask questions is that
>>>  >30 years ago, I was an instrument tech in a university setting.
>>>
>>> I think my most basic question is why a fast 16 bit A/D?
>>
>>
>>
>> I've posted to this group and cross-posted to several others asking 
>> about S/N of various media, we'll see what turns up. There's always 
>> the rationale that the extra information can be used for 
>> post-aquisition processing, which would bring up details that wouldn't 
>> otherwise be visible. What's that called: histogramtic edge 
>> enhancement, for example.
>>
>>>
>>> When you have a CRT display, is more than 8 or 9 bits really useful?
> 
> 
> I'm following your thread on this topic.
> 
>>>
>>> A secondary question revolves around using scintillation(sp?) 
>>> followed by a photomultiplier.
>>>
>>> Why not use a "channel electron multiplier" in count mode?
>>
>>
>>
>> Contamination. Typical vacuum in the chamber is 10-5 to 10-6, and 
>> traditionally it is very dirty with diffusion pump oil. 
> 
> 
> Brings back memories. We had a couple of old mass specs that my boss had 
> retrofitted with ion pumps. The diffusion pumps were used essentially as 
> roughing pumps. They had LN2 traps. But I don't recall if they were 
> always used.
> 
> 
> 
>> [snip channel multiplier problems]
>>
>>> Also the problem of "image tearing" was mentioned in one post.
>>> Would not that be alleviated by synchronizing display and offsetting 
>>> of scan?
>>
>>
>>
>> Traditionally this is no good because the refresh rate is so slow. But 
>> ultimatley there's no (built in) way to know the sample has moved, as 
>> the sample is moved mechanically.
> 
> 
> Is that still a problem in current designs?
> I can think of a couple of ways to detect "when"/"how far" sample holder 
> has moved, unaffected by platform being move mechanically with attendant 
> "slop".
> 
> 
>> Also look for information about homebrew STM, and try to find the 
>> company in Germany selling kits for about $1K euros. 
> 
> 
> That sounds *CHEAP*!
> How much "infrastructure" ( pumps, electronics, etc ) does that include? 
> That could be affordable to tech oriented high schools. In the 60's, 
> when Varian introduced 60 MHz NMRs with permanent magnets, I thought 
> that had impressive possibilities.
> 
>> I'll definitely be building one "soon".
> 
> 
> Not I. I've purchased a 100+ year old house which has undergone strange 
> improvements.
> 
> 
>>
>> I suppose this discussion belongs in sci.microscopy. They're lonely 
>> over there.
>>
> 
> I may keep them company ;)
> 
> 
> Thanks for your reply.

Beauty part is you don't need any infrastructure for an atomic force 
microscope. They operate mechanically, and at atmospheric pressrue.

I've got a house like that, 1884. It deos suffer for my hobbies.

-Jeff

jeff miller wrote:

   ...

> Many use turbos these days, though vibration can be a problem. I've 
> never seen a trap on an SEM, but I'm sure it's done for special 
> purposes. The real problem is they never bake out.

My set-up was a stack: pump, cold cap, nitrogen cold-trap, slide valve, 
working chamber. With the valve closed and steam in the cold trap, it 
baked out just fine. I only needed to do that when it had gotten exposed 
to air. Even when the chevrons were oily enough to drip at room 
temperature, there was no problem with oil vapor coming off them when 
cold. Of course, we tried hard to avoid ever going higher than roughing 
vacuum in the pump stack. I had valves at both ends of it. We vented the 
working chamber with helium before opening it to the air. That was 
supposed to minimize adsorption of nasties.

> But it seems the vacuum just isn't a problem. All the rage these days 
> are scopes that operate at 10-1 to 10-3 range. They have at least one 
> pump on the column near the filament, often several pumps one between 
> each aperture, so that the vacuum is good near the cathode (filament) 
> which is all that really matters.

A lot of processes get by on minimum vacuum. Before I knew better, I 
made a "vacuum" "tube" without an envelope. We had an electric hot plate 
with a nichrome coil heater in an open ceramic labyrinth. The top of the 
coil nearly touched the bottom of the pot. (Can you imagine such a thing 
on the market today? We never had trouble with it.) I laid a piece of 
brass screen wire over it for a grid (so that's where the name comes 
from!), some glass glass beads on that to make an insulating gap, and a 
metal pie plate (so that's where the name comes from!) with water in it 
so it wouldn't warp, and with 700 volts on the plate, made a Hartley 
oscillator work. An eighth inch at one atmosphere is much more than an 
electron's mean free path, but that's evidently not the whole story.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������