Sciencemadness Discussion Board

Gas Chromatography - Plasma in a Tube

aliced25 - 3-12-2010 at 00:39

These three articles from the same pair of Author(s) should have got more interest here than they have. They describe the use of a fused silica capillary tube, with electrodes fitted over it (cut from a 14G Needle, Terumo) and a block separating them with a radial outlet for optical fiber. The power for the unit came from a 12V Battery.

It was first described in the article Guchardi, Hauser, 'A Capacitively Coupled Microplasma in a Fused Silica Capillary' J. J. Anal. At. Spectrom. Vol.18, 2003 pp.1056–1059. It was followed up with Guchardi, Hauser, 'Determination of Organic Compounds by Gas Chromatography using a New Capacitively Coupled Microplasma Detector' Analyst Vol.129. 2004 pp347-351 and also by Guchardi, Hauser, 'Determination of Non-Metals in Organic Compounds by GasChromatography with a Miniature Capacitively coupled Plasma Emission Detector' J. At. Spectrom. Vol.19 2004 945-949.

That is it, the whole system less the spectrometer, for the sensitive detection & determination of compounds by Gas Chromatography. There is even a move toward Plasma Source: Mass Spectroscopy, which will presumably use the pre-ionized stream from the plasma and watch its flight path in a magnetic field. As we are talking about a decent field strength magnetic field in another thread, plus electromagnetic fields as well, it might be something to consider.

Wonder what that would make it? No longer GC/MS, but maybe GC/AE/MS? The Gold Standard just got moved again, and for once we may have the capacity to follow it.

[Edited on 3-12-2010 by aliced25]

not_important - 3-12-2010 at 11:38

Plasma source MS isn't new, ICP-MS dates from at least the early 1990s and can be used for elemental analysis as it rips molecules into atoms (which may reform into simple very stable ions), while chemical ionisation goes back to the 1960s - plasma production of the ions used to ionise the sample - and is useful for determination of molecular mass.

Currently used mass spectrometers are often different that the earlier simple sector types that you refer to. Time of flight, various forms of quadrupole, ion cyclotron resonance, and Orbitrap, are all in use; the last two use Fourier transform techniques to analyse the ion masses. I like the Orbitrap as it doesn't require a magnet but does have lower resolving power as m/z increases.

The GC front end is used to resolve mixtures, it adds complexity so if all you're after is the mass or elements analysis of reasonably pure compounds you might not use the GC from end. Remember that you need rather pure carrier gas, and that it can interfere with detection of some peaks in certain uses.

aliced25 - 3-12-2010 at 20:40

I was thinking of something like a small version of a lecture bottle with its own regulator. Packed inside a box with the GC/MS.

Plasma source isn't new, but the generator is fairly simple and that is new. The magnet in the contemplated variant would be the same magnet, so 1T as under discussion elsewhere. Time of flight, Quadrupole Filtered CCD Based Mass Spectrometers look like what we are after.

Designs like this one which utilise miniature, high-speed CCD/CMOS Sensors to collect a huge number of frames per second and then use that to determine the parabolic curve that the mass is descending in. I'm assuming the descending ions are still giving off an afterglow from ionization, which means the computer sees bright spots and where they are in an x,y axis and based upon multiple frames, determines the z axis of each bright spot. Obviously the sensor has to be put out of the line of ion-trajectory or it would be damaged (see this)

Now, provided a pure sample is at hand, it could be injected directly into the ionization/plasma chamber, no? If an impure/suspect sample was all that was at hand, it could be injected prior to the column, heated till it evaporates and put through a thin column, to separate the components? That would allow the emission spectra to be be of single components, each of which would have a unique Rf through the column?

In any event, post-plasma ionization, it is put through the Mass Spectrometer, where it's trajectory will be imaged with the CMOS sensor, probably mounted on the opposite wall, or even looking down into the well from above. Here is a 1 Tesla portable Mass Spectrometer, and here is a Quadrupole system based upon an atmospheric argon based plasma ionization scheme.

Looks interesting, the charged-particles are given forward momentum by the gass-flow, the trajectory is altered by the weight of the ion as the forward-impulsion of the gas-flow stops being able (due to the increase in area I suppose) to propel the gas forward. The variances in the trajectories enable the distinction between the various charged particles, which are obviously still emitting.

Ok so a useful Instrument would have the capacity for the sample to be injected at either of two ports, before or after the packed column. It would utilize an onboard gas-cylinder that could be changed easily. It would utilize a CCD/CMOS Fast-Frame Rate Camera (monochrome & lumogen coated), that was out of the way of the charged particles, but still able to take multiple 2D images which would allow the 2D to be computed into a 3D Trajectory? It would also be coupled to a UV-NIR Spectrometer for the Emission spectra (at ionization/plasma) to be recorded & mapped?

not_important - 4-12-2010 at 16:48

Quote:
I was thinking of something like a small version of a lecture bottle with its own regulator. Packed inside a box with the GC/MS.


Do you know the needed flow rates? And you still need high purity gas.
Quote:
...I'm assuming the descending ions are still giving off an afterglow from ionization, which means the computer sees bright spots and where they are in an x,y axis and based upon multiple frames, determines the z axis of each bright spot.


The paper states
Quote:
The photofragmentation is performed at 193 nm, using a focused ArF excimer laser with a pulse energy of approximately 0.5mJ. The photolysis products from CS2 are ionized using resonantly enhanced multi photon ionization (REMPI). For dimethyldisul¯de fragmentation and ionization occur via a number of multiphoton channels at 193 nm [23].
The resulting ions are velocity mapped onto a detector consisting of a pair of 75mm microchannel plates with a 32 ¹m pore spacing, and a P47 phosphor screen (with a decay time constant of ¼ 40 ns). The ion lenses are tuned such that the different masses arrive at the detector spaced in time by at least 250 ns, in order to prevent overlap between consecutive
images on the phosphor screen.


So
a) photofragmentation
b) photoionisation
c) microchannel plates for apmplification
d) a fast phosphore for conversion to light
Quote:
Now, provided a pure sample is at hand, it could be injected directly into the ionization/plasma chamber, no?


What type of ionisation? There are several that use plasma, both CI and ICP do. At this point I just hear buzzwords in your description.
Quote:
If an impure/suspect sample was all that was at hand, it could be injected prior to the column, heated till it evaporates and put through a thin column, to separate the components? That would allow the emission spectra to be be of single components, each of which would have a unique Rf through the column?

Useful only if you know the RFs of all possible compounds for the column & conditions. Otherwise you make assumptions, repeated runs with added amounts of knowns to see which peaks get bigger, or appear before or after a given known.

The emission spectrum? And then mass spec? I'm confused as to why. The emission gives many elements, MS can do almost all.

aliced25 - 5-12-2010 at 00:59

At this point I am describing something that would utilize the pre-ionised particles coming from the plasma in the column (the one that gives off the glow). The flow rate through that is in the paper (NB can't be asked, too busy learning codeigniter). If the initial port was pre-column, it would be heated and passed through a packed column (shit, they ARE on ebay), separating the suspect sample. Yes, there would, presumably, be a bunch of Rf's, but whatever came through the column at the same time would be ionized at the same time by the plasma, giving off an Emission Spectra, which would presumably be different to the other components of the impure sample (when they get there). So you have a built-in detector/ionizer to determine the Rf's (when the glow changes there is something other than argon/helium in there).

For a sample known to be pure, inject it into the system after the column, but prior to the plasma generator/ionizer, the suspect sample will also get to the same point & be ionized by the plasma. Either way, it is ready for the Mass Spectrometer is it not? With the information gained by the use of all three methods in one unit, it should be a little easier to work out what is what, (everything in the mass spectrum for a given period is going to be from the one Rf isn't it?)

Yeah, you are right, get the Rf, then Emission Spectrum, then Mass Spectrum - that should really get the data to do it all shouldn't it? Built into one little package, wouldn't be cheap but, the columns run at horrible prices. Plus you'd have to have software to compute the connection along the z axis of the x,y bright spots in multiple frames, with a whole bunch of other bright spots moving through at the same time.

That's saying nothing whatever about the spectrometer needed to cover the VUV-NIR regions, that is less than trivial if you want any sort of accuracy (as you pointed out elsewhere). But the bottom line is that it can be done, provided being converted to plasma ionizes the sample (like it ionizes anything else), provided the high-frame rate camera can be sorted out & provided the emission spectra can be arranged, nothing to it, well, apart from software really (maybe a teensy weensy bit of hardware as well). As for knowing the Rf's of all possibilities prior to the emission/mass spectrum being to hand, why should you? An Rf compared to them would have to give a pretty good indication of what was in the sample I'd have thought.


By the way, doesn't this look like fun, Gigavolt to set off a tankful of Deuterium.....:o Just generating electricity honey... Hmmmm, I need some lithium with the Hydrogen plasma (neutron + lithium = tritium), down to the battery store I go:P The energetics crowd would love that, how to get deuterium to burn, triethylborane would seem to be a good choice (can you deuterated boranes?).

Anyway, I'm going to have to try and get hold of a lecture bottle of hydrogen (that should be fun with my history), to try out a simpler way of making the D-Bulbs (they are currently in quartz envelopes). If I can make an arc in Hydrogen, it will also work for Deuterium is the theory. I'm enrolled now, so getting access to the lab is no drama in the near future, getting someone to let me in when I want to send kV's of electricity through hydrogen might be a bugger but, but at least they'll have a decent vacuum system and probably hydrogen on tap. They won't have D<sub>2</sub> on tap, but I can always raid the piggy bank and order it, especially for a non-dodgy patent application.

[Edited on 5-12-2010 by aliced25]

DDTea - 5-12-2010 at 02:50

Maybe I'm misunderstanding what you want to accomplish, but Atomic Emission + Mass Spec seems highly redundant.

Mass Spec has much higher sensitivity than AES. Also, AES is quite finicky when you look at the physics behind it (mainly, the need for extremely tight temperature control) and really more suitable to inorganic samples than to organic samples. However, from the second paper by Hauser that you cited, he's simply using the emission spectrum of carbon to see when something is coming off the column. You could just as well use some cruder means as a column detector.

For actual structure determination of an organic sample, though, plasma atomization is not going to be very good for organics!

not_important - 5-12-2010 at 12:08

Quote: Originally posted by aliced25  
At this point I am describing something that would utilize the pre-ionised particles coming from the plasma in the column (the one that gives off the glow). The flow rate through that is in the paper (NB can't be asked, too busy learning codeigniter).


I'd consider being bothered at bit more. You missed a significant portion of that design, even the block diagram included everything I mentioned in my previous posts.
Quote:

If the initial port was pre-column, it would be heated and passed through a packed column (shit, they ARE on ebay), separating the suspect sample. Yes, there would, presumably, be a bunch of Rf's, but whatever came through the column at the same time would be ionized at the same time by the plasma, giving off an Emission Spectra, which would presumably be different to the other components of the impure sample (when they get there). So you have a built-in detector/ionizer to determine the Rf's (when the glow changes there is something other than argon/helium in there).


Added complexity for no good reason. How do you know that the sample is a pure substance? Perhaps by running it through a GC system, yes-no?

How will the emission spectra of 1-bromo-2-phenyl-ethane and 2-bromo-2-phenyl-ethane differ? What about the amides of C8, c10, and C12 fatty acids?
Quote:
...
provided the high-frame rate camera can be sorted out & provided the emission spectra can be arranged, nothing to it, well, apart from software really (maybe a teensy weensy bit of hardware as well). As for knowing the Rf's of all possibilities prior to the emission/mass spectrum being to hand, why should you? An Rf compared to them would have to give a pretty good indication of what was in the sample I'd have thought.


That high frame rate camera uses a CCD that shifts an image into storage cells on the chip, so it grabs a number of frames as analogue data before sending them off to be digitised. Not really common hardware, likely to be pricey.

And then there's the microchannel plates, such as carried here
http://www.sales.hamamatsu.com/en/products/electron-tube-div...

also not really cheap.

I'd worry about getting a UV detector that does what you need before messing about making your own deuterium lamp. It's not as simply as slapping a coat of phosphor on the top of a CCD package; consider the trig for 10 um spaced detector cells with a 1 mm thick glass window above them, at 100 times the cell spacing that's going to mean a 5 um spot will excite a lot of cells (phosphor emits in a full sphere).

You might to browse the site http://www.eureca.de/neu-german/optoelectronic/sensors.html or http://www.eureca.de/neu-english/optoelectronic/sensors.html to see the types of sensors normally used for instrumentation; also at one time they listed some of the shortcomings of consumer CCDs (high dark current and noise, limited dynamic range,...) .

Note that MS takes high vacuum, neither cheap nor really low power (although there are spiffy micro-turbopumps) Also that the GC needs heater power, again really good insulation can help as can microchannel heat exchangers to capture heat from the vented gases.

aliced25 - 5-12-2010 at 15:30

I've requested and got a number of articles on the application of the phosphor to the sensor, there are a number of such articles about, they aren't that easy to get. But from what I've read so far, that is the least of the problems.

Why would I need a multi-channel plate? The ionization takes place in a quartz capillary, and while they are not cheap, they're not too bad considering. The fiber from the plasma generator to the spectrometer is exxie as all hell (Deep UV, Solarization Resistant Fiber ain't cheap).

Small digital cameras, such as the high-frame-rate direct digital output low-cost webcams are what I'm looking at. The resolution sucks at the higher frame rates, but that ain't so important here, they are designed to be built into systems where the image is shifted off the ccd/cmos sensor as a binary blob (or series thereof) directly up a USB cable, extra storage on chip really isn't required or a good idea, it would be hell easy to get them out of sequence.

I don't know how the two you cited will differ, or the fatty acids, quite probably there are textbooks on the subject that will allow them to be separated, after all, there will be 3 separate lots of data from which to do so. As to determining whether or not a substance is pure, that is why the GC is there.

not_important - 5-12-2010 at 21:54

Multichannel plate == high gain amplifier for 2D distributions of ions, electrons, and stuff that can form them; in effect an array of electron multipliers. Note needed for the ionisation detector, as that does directly generate light; however the mass spec is a different story. The MCP is what turns an ion in the MS into enough electrons to make a bright enough spot via the phosphor. Gains of several orders of magnitude, a 2 plate system might have an overall gain of 10^8.

I seriously doubt that camera you are looking at is anywhere near fast enough. The reason they store images within the chip is that reasonable A/D converters are not available or prohibitively expensive. Look at the frame rates they are doing:
Quote:
The camera used for this work is a programmable ultra-fast frame transfer CCD camera (Dalsa inc.), which can be clocked to the arrival times of the different masses. The camera is capable of recording and storing on chip up to 16 consecutive images with user determined
exposure lengths after an initial trigger, with a time resolution of 5 ns. The minimum and maximum exposure times for each image are 10 ns and 1.275 us, respectively. After each acquisition cycle, the stored images are transferred from the chip to a PC at standard data rates for processing and storage.


In case you missed that, the __slowest__ frame rate works out to around 25,7 Gbits/sec assuming 8 bit pixels; well past USB-3 rates much less USB-2. The camera is the analogue memory (NOT digital) for the images, conventional ADC and digital data transfer happens at the standard snail's pace long after the frame capture sequence. 10s or 100s of microseconds later. As for sequence issues, that's no problem provider the circuit designer isn't too busy doing something else to bother to read the specifications. FCOL, reset a counter at the 'start' trigger, increment every frame, repeat to extract the frames or allow random access to the stored frames and shoot the programmer if they screw up coding the access sequencing; Digital Design 101.

The point on the example compounds is that while their Rfs will differ, at the ionisation detector they will look quite similar; without reference runs you may have no clue as to which is being detected. BTW - retention times on various columns can vary widely, affected by packing, temperature, flow rates, and so on; the best you can hope for in many cases is obtaining the relative order of several compounds.

aliced25 - 6-12-2010 at 15:56

Yes, the Rf's will differ that is kind of the point with Chromatography of any nature isn't it? But you are contradicting each other, one is saying that it is overkill, the other that some compounds will require all three techniques in order to separate the compounds in question.

As to the camera, I'm basing my projection on the trend which is being driven by the enormous Automotive and Mobile Markets, for High-Frame Rate, CMOS-Image Sensors, with digital output in the >1,000 FPS Range (some in the >10,000FPS Range (or 100 nanosecond)). Yeah, they are expensive at the moment, but as we all know competition of the kind that is going on at the present time will see these under $100, then under $50 each within 12 months.

Those sorts of speeds will outstrip the capacity of the USB, then we will have to look at the multi-gigabit per second Ethernet-type connections, quite possibly more than one, one for the camera on the x,z axis and the other for another camera on the x,y axis (the combination of cameras would allow for much easier computation of traces along the x,? axes. A lot of the computation could presumably be done prior to sending the data to the computer as well. The DSP's that are out now are quite capable of separating everything but the pixels of interest (anything with an intensity over 10) per frame, thus reducing the number of frames being sent per second to the PC/Laptop.

Integrating the system is not going to be as simple as I facetiously made it appear earlier in this topic, that was kind of tongue in cheek if you have a look.

not_important - 7-12-2010 at 11:58

10000 FPS = 10^5 FPS, that camera they used is doing 10^6 to 10^8 FPS. Not a lot of demand for those, and not real likely to have any near term drop in price. Best to stick to stuff that is in the falling part of the price curve right now, rather than projecting what will happen to something still up on the high end - too many have been surprised when a predicted product price drop or introduction failed to happen.

USB 3 tops out at 4800 Mbits/sec. How fast is 1000BaseT Ethernet? (Hint, what is the numeric portion)

All three -techniques - GC to separate, or if you will prove that something is pure, MS will do better at elemental analysis than plasma-atomic emission. On the other hand, that plasma detector is pretty simple, it would work well as the general detector for the GC with the bonus that certain elements can be detected even if not always accurately enough for quant work; but certainly well enough to distinguish between precursor and product where one has a halogen and the other doesn't - handy when you don't have reference samples of both to determine what a given peak is.

The point on the Rfs differing is that you can not easily compare the Rfs from one piece of equipment with those from another, and even from the same piece of equipment if run with different conditions or even over time. It's not uncommon to run a GC, then immediately again with a reference compound or 3 added to the sample.

Best to read some of the other papers on that MS to get a solid understanding of the method, including why they do what they do. There aren't 3 dimensions being looked at in the paper you've referenced.


not_important - 8-12-2010 at 17:23

Argh - too late to edit my last post.

As for the GS-plasma set:

Bypassing the column means adding valving and plumbing, and possibly another injector; all of these must be extremely clean and free of anything that can vapourise to even a small extent at the operation temperatures. It would be simpler to always run the sample through the column, which also gives proof of purity.

Note the purities involved - a half ppm of N2 in the argon used is responsible for the detection of carbon via the line from CN.

Given that out of the elements more common to organic chemistry, only sulphur might need observation of lines below 350 nm, I'd omit the VUV aspect and concentrate on a good resolution and sensitivity over the 350-1000 nm range or thereabouts. VUV can be really delightful to try to get properly functioning...

unionised - 9-12-2010 at 12:35

Am I missing something here?
Way back in the late 80s our lab could do GC FTIR in real time. The major advantage of this was that it let us brag that we had the fastest computer in the building.
The bandwidth required for GC UV wouldn't be much different- it's still a wiggly line every tenth of a second or so.

Computers are a whale of a lot better than they were back then.

An expensive way to use the idea posted above would be to get an old photodiode array spectrometer or LC detector and put the glowing quartz tube in the place of the sample cell.
Something like this
http://cgi.ebay.co.uk/WORKING-WATERS-996-PHOTODIODE-ARRAY-DE...

The easy way would be to use a polychromator and a few detectors- one for each element you wanted to monitor.

A handful of detectors feeding signals with bandwidths of a few tens of hertz wouldn't tax a crap old computer, never mind a new shiny one.
The detection limits wouldn't be great, but you would still have an interesting toy.


The mass spec side would be a bit more tricky

Incidentally, re.
"The point on the Rfs differing is that you can not easily compare the Rfs from one piece of equipment with those from another, and even from the same piece of equipment if run with different conditions or even over time. It's not uncommon to run a GC, then immediately again with a reference compound or 3 added to the sample"

Old timers like me have heard of this trick.
http://en.wikipedia.org/wiki/Kovats_retention_index


not_important - 9-12-2010 at 13:46

Bandwidth for the plasma detector + spectroscope is low, and not the issue. The high bandwidth would be for the MS method he was talking about, an entirely different question.

The plasma detector he pointed out has some advantages in simplicity over many other GC detection methods, and certainly both simplicity and cost are better than with FTIR unless you already happen to have one lying about. I think that for a homebrew GC the plasma detector is well worth considering, especially if you avoid trying to mess with the deep UV spectrum for specific element lines.

Kovats takes a supply of fairly pure alkanes to generate the relative times, something that may be difficult for the amateur to obtain. Tossing in other at-hand knowns may be less generally useful, but certainly can be used to ID peaks from reagents used. And while there are databases of these, there's still often a number of compounds with similar or same Kovats values, so they're not a panacea to identifying an unknown peak (that's one aspect of the plasma detector + spectra that can help - the easily detected elements are common in organic chem, so some possible hits can be eliminated or bolstered). I don't know how extensive the databases are, what the likelihood of finding your reagents and products in them are; I know that a significant amount of the older lit I've read where GC was given didn't use Kovats making corrilations to a different machine's results tougher (hell, some didn't give details on the column and conditions either)

unionised - 12-12-2010 at 08:43

If I were looking to buy a detector to put on the end of a GC I would go with a mass spec; no question. For what it's worth GC/MS systems have been around for long enough that the computing power available when they were new looks like a toy compared to today's PCs. The interfacing is tricky but the point still stands. The bandwidth of a mass spec isn't very different from a UV or IR. In all cases you just get a wiggly line each time you sample.
The thing about using this technique is that you could get a reasonable measurement of, for example, the PCBs in an oil sample even without any clean-up. If you only look for a chlorine emission line then all the hydrocarbons and esters just don't show up. I know that a mass spec can sort of do that too but the ionisation is dependent on what else is present in the ion source at the time. With a stupidly high temperature plasma you can hope that everything is atomised and ionised .
There will be problems with ion suppression and with variation of emission conditions. Neither detector is (anything like) perfect but you can compare the two answers.
The clever bit is that if both your detectors agree then you are a bit more certain of your results.

A lot of the difficulty with getting a pure source of each alkane goes away if you have a decent GC and a few alkanes to use as markers. You fill in the ones in between the markers using something like diesel oil. It's pure after you GC it.
Having said that literature is pretty thin on Kovats index data and I'd not report a result on just retention time without some other confirmation.
Having two confirmatory techniques would be even better.

not_important - 12-12-2010 at 10:38

Agree, however a borosilicateUV-Vis spectrograph is much easier to DIY than a mass soec, and DIY seems to be the goal. If wishes were MS machines, then I'd go for both plasma ionisation for the atoms and CI for MW and favored fragments. I'd also have a number of nubile lab assistants wearing skimpy swimwear under their lab coats, doing all the scut work for me...

I'll take your word on the way to get sufficient reference peaks for Kovats, with minimal pure refs. I've not done enough to have any real feel for that. Sounds handy if it does generally work.