Gas chromatograph DIY - Done!

[ChristofferB]

EDIT MAY 2019: Sorry for reviving a dead thread but the alternative was making a new thread with the exact same title. Not sure what best practice is then.

I've completed the next iteration of my gas chromatograph and made a brief video about it! Hope you'll enjoy!



Some of you will have seen my previous posts on different elements of this project but now that a working prototype has been finished, I thought I'd make a thread describing
the project in more details.

So here it is, my home-built gas chromatograph with thermal conductivity detector (GC-TCD).





It consists of the following major components:

 - Carrier gas supply (helium)
 - Carrier pressure regulator
 - cold on-column injector
 - column
 - detector
 - detector amplifier

I know there aren't many chemists here, and I know the project is a bit of a mashup of electronics, chemistry, mechanical engineering and so on, but bear with me.

For the non-chemists, here's a very TL;DR description of what this instrument does:

->a flow of gas is established through column; a long tube packed with a fine mineral or polymer powder with a large surface area, that can be either coated with a film of liqid or uncoated.
->A sample of a mixture of gasses is injected into the gas stream before the column, and as the molecules travel across the large surface in the column, they'll interact more or less strongly with it.
->this results in different components effectively moving through the column at different speeds and thus a separation has been performed.
->A variety of detectors is then used to sense the change in gas flow, indicating when each of the gas components comes through. detector readout as a function of time is recorded.
->These systems can analyze gasses, why the instruments are typically built in ovens, capable of reaching 400+*C

that's it, that's all. but it's one of the most robust tools in analytical chemistry, along with its liquid brother HPLc.


These machines usually costs 20,000+ USD, which gave the inspiration to this project: cheap, off the shelf and robust.

I wanted to design a high-capacity instrument for basic everyday lab use, following reactions and the like, that could be put together without (many) custom components.

Design considerations

Carrier gas:

In my mind, if I had to buy/rent a cylinder of high purity lab-grade gas, this project would already be so much over budget it wouldn't be worth it, so The carrier gas was chosen to be either welding argon or balloon helium.
Both of these can be obtained in low-pressure single use cylinders. Hydrogen is common also, and while it is handy that you can just have an electrolysis rig supplying your instruments, and not storing the gas, I still deemed it too
dangerous for me. I simply did not want to have to worry about it. Balloon helium was chosen over the grounds that it's cheap and helium is a very commonly used carrier gas, making comparisons with litterature easier.

Mechanical components:

Every part of the pressure part of the system is made from stainless steel tubing and compression fittings, both cheap and chinese from ebay. The proper fittings to use are the Swage-lok line of process fittings, but at 40 times
the cheapest price i found from china, i went with cheap, and hoped they wouldn't be leaking. The system only operates at max 60 PSI, so i wasn't too worried about fitting failure, more that the compression ferrules would be so rough they wouldn't seal tight.
Luckily, they did, a spray-down with soapy water is a good indicator, and giving the compression nuts 1/4 turn more usually solves any small leaks.

The inlet is made so that a rubber disk, known as a septum, can be placed under a compression nut on a T fitting and tightened down, to introduce samples via syringe. it's made longer than need be to give some volume where the gas
can sit until it enters the column. In the future this assembly will be outside the oven, heated separately.

The pressure regulator is a Brooks 8601 regulator scavenged from an old gas chromatograph, it's a panel mount reduction valve, they show up on ebay at reasonable prices.

Detector:

The detector design came from a paper in the Journal of Chemical Education, and simply consists of an 1n4148 in a short lenght of thin-walled 1/8 tubing, cathode spot welded to the wall, anode connected to a 150 ma current supply,
that heats the diode. it'll reach some equilibrium in the stream of helium, and when the gas momentarily changes, so will the heating of the diode, giving a change in voltage over the diode.

The diode voltage is fed into an instrumentation amplifier with a fairly large amplification set by a 10-turn pot, and compared to a "stable" voltage source (AA battery). The reference voltage can be changed with a potmeter to set the zero.

and the amplified signal is passed out to a BNC, where any datalogger can be used to record the data. I use an arduino with some Vbasic script i found that plots time and voltage in excel.

Column:

The column i use is also of 1/8" stainless tubing, around 50 cm packed with aluminum oxide (alumina, pottery store grade), which has been wetted with approximately 14% by weight liquid paraffin. this is mixed until the powder is reasonably dry.
the column (straightened) is fitted with a cotton plug in the bottom end, to which a small vacuum pump is attached, and a funnel is mounted on top, in which small amounts of the stationary phase is added.
The pump is turned on and the column tube is tapped with a screwdriver (ugh..) for a long.. long... time.. until no more alumina sinks in. at this point it is plugged with cotton (use glass wool if it needs to be heated!), the column is coiled up,
and ferrules and compression nuts are mounted.

test:

Here's the first analytical separation: a propane/butane mix from a blow-torch. 150 ul is drawn with a syringe and injected through the septum with a helium head pressure at 20 psi, giving around 18 ml/min flow through the column.



The sample is injected around 20s in. it is seen that there is a tiny wide peak, which is residual air, and then two large peaks with a minute inbetween. I'm fairly certain the first one is propane and the 2nd one is butane.
(propane is more polar than butane, thus more similar to air, thus less column retention, since the column packing is coated in oil).



I believe this is the first internet-available results of a homebuilt gas chromatograph!



Further work:

-better detector: I want a full wheatstone bridge thermal conductivity detector that compares the column output with the clean helium in the carrier line. this way any changes in flow or pressure will self-correct, it'll also benefit sensitivity a lot.

-heat: the column will be mounted in a toaster oven, power resistor heaters and thermostats on both injector and detector will allow for analysis of liquids and solids with a boiling point up to around 200 *C.

-temperature programming: being able to ramp temperature over a set time, make temperature plateaus and so on will MASSIVELY increase resolution.

-better columns: finding better coating liquids, making columns longer and more uniform for better performance.


Anyway I hope you found this interesting, I know I have! I'd also like to thank all of you who gave advice on the project in its different phases over the last year and a half.

--Chris



 

[LukeW]

Wow, it actually works! Well done!
How much did it cost all up? What was the most expensive part?

[BradC]

Thanks for the write up. Both awesome and fascinating.

[ChristofferB]

Wow, it actually works! Well done!
How much did it cost all up? What was the most expensive part?

Thanks!

The most expensive part was definately the regulator, as I bought almost ½ of an old HP 5710 gas chromatograph. I've since found them individually on ebay much cheaper, though. here's an example. eBay auction: #https://www.ebay.com/itm/BROOKS-PRESSURE-REGULATOR-8601/132623659801?hash=item1ee0fc7319:g:9XIAAOSwKoFa-0hE

These regs don't go as high on the input side as the disposable balloon helium tanks, so i omitted a high-pressure regulator for just bleeding the tank down to an acceptable pressure. It doesn't impact analysis cost much, but it still feels kinda terrible to just let out the gas you just purchased.

[ChristofferB]

well.. still too long to read but looks interesting. anyway as always the advice is... i hope you dont blow yourself up...

Yeah I guess it's a bit wordy. Nothing to blow up though! The entire systems runs on around 20 psi of helium in great part for my own peace of mind. The total volume of compressed gas in the system is also tiny, so even if a fitting shoots apart, the pressure will die out almost instantly.

[vindoline]

Congratulations on a working prototype!  I'm impressed with the separation and resolution you've achieved with your home made column.

[babysitter]

Just to inject (pun intended) one of EEVBLOGS most favourited voltage references into your gas pipeline:

* Get a LM399.
* dremel or drill away the top of its cap so that you have some lip left to solder it to
* solder the LM399 into the gas pathway
* Apply heating voltage, measure heater current.

The Heater tries to hold the die on a constant temperature, sinking different currents when the cooling gas changes.

[RoGeorge]

That's very interesting, congrats! 

If you test other gas mixtures, I would be very curious to see the results.

[mikeselectricstuff]

I wonder if a heated thermistor might be more sensitive.

[ChristofferB]

I wonder if a heated thermistor might be more sensitive.

It definitely would be. The proffessional thermal conductivity detectors consists of a bridge of hot wire filaments of a tungsten/rhenium alloy.
That's the next big upgrade I'll get working on after the oven and heat control, either with filaments or thermistors.

[Tom45]

Congratulations. Very nicely done.

Back in high school (55 years ago) 3 of us built a gas chromatograph with the guidance of our high school chemistry teacher. Ours was considerably more primitive than yours.

The column was 8 feet of copper tubing packed with Tide laundry detergent. Did that in a stairwell. Then coiled the tubing around a fire extinguisher body to make it compact enough to fit in a reasonable size homemade oven. We used helium too. Pressure was adjusted manually while watching a Hg manometer.

The detector was a number 47 bulb with the glass removed. It was in one arm of a 6 volt Whetstone bridge. Results were graphed on a strip chart recorder that we also built. We couldn't apply power to the bridge until helium had reached the column's output. Otherwise, the bulb filament would burn up. Held a flame up to the column's outlet and when it went out, we were OK to apply power to the bridge and inject the sample.

I hadn't really thought about since until seeing your project.

[ChristofferB]

Congratulations. Very nicely done.

Back in high school (55 years ago) 3 of us built a gas chromatograph with the guidance of our high school chemistry teacher. Ours was considerably more primitive than yours.

The column was 8 feet of copper tubing packed with Tide laundry detergent. Did that in a stairwell. Then coiled the tubing around a fire extinguisher body to make it compact enough to fit in a reasonable size homemade oven. We used helium too. Pressure was adjusted manually while watching a Hg manometer.

The detector was a number 47 bulb with the glass removed. It was in one arm of a 6 volt Whetstone bridge. Results were graphed on a strip chart recorder that we also built. We couldn't apply power to the bridge until helium had reached the column's output. Otherwise, the bulb filament would burn up. Held a flame up to the column's outlet and when it went out, we were OK to apply power to the bridge and inject the sample.

I hadn't really thought about since until seeing your project.

That's actually a quite good idea for column material! Detergents contains sulfonates (did in the old days atleast) and i'm pretty sure you can get commercial columns that are chemically similar, with sulfonic acid groups on vinyl resin beads. I might just borrow that.


I'm considering doing a small series of youtube videos, each detailing a major component in theory, design and construction, I'll put links in the first post if i ever get around to it.

[mikeselectricstuff]

Keep an eye on The Haackaday prize - as & when they do a category that this would fall into it should have a good chance!

[ChristofferB]

Sorry for reviving a dead thread but the alternative was making a new thread with the exact same title. Not sure what best practice is then.

I've completed the next iteration of my gas chromatograph and made a brief video about it! Hope you'll enjoy!

[T3sl4co1l]

Cool!

Tim

[DaJMasta]

I know next to nothing about the subject, but it's a very neat project!

What kind of temperature are you looking to heat to?  I'd imagine something like 60C or so wouldn't be too much of a problem, just taking a control circuit from a OCXO and beefing up the drive transistors for the larger chamber, and it looks like you already have insulation that could be suitable for that sort of temperature.

[MagicSmoker]

Ambitious project with a number of clever hacks! A couple of minor points/suggestions:

Helium for balloons often contains up to 20% air to reduce the likelihood of asphyxiation. He for welding should be sufficiently pure, but it will also cost a whole lot more than Ar.

Temperature change in the oven needs to happen as evenly as possible, and depending on how hot you really need to get it (assuming much less than 400C) then a suitably modified hair dryer might do the trick. I'd mount it pointing down in the center of the column so that the resulting air flow is turbulent and wraps around the chamber sides before hitting the column circumference. Or you could use a fan from a table-top convection oven and nichrome wire/ribbon as the heater, but still avoid pointing heated air directly at the column.

A TCD is a workhorse of a detector, but in the words of Elzar the Chef, break out the spice weasel and kick it up a notch by making a mass spectrometer. As I recall from working on a similar device years ago (for laying down patterns of ions on a flat target), you "just" need to ionize this gas flow from the column, capture the electrons, then use a precisely controlled electromagnet to deflect the ion beam onto a negatively charged plate using the amount of current and gas flow rate to infer the mass of the ions being deflected. The CC source for the magnet likely needs sub-mA resolution and must be extremely stable, but at least it doesn't need to be terribly fast so a chopper-stabilized op-amp driving a BJT should do the trick.

[KE5FX]

Very nice, and surprisingly stable.  I've never seen anything that makes an Arduino talk to Excel before, but it seems like it'd be great for prototyping.  Any links to more details on that?

Also, any plans to work on the spin-resonance spectroscopy idea on your Tumblr?

[ChristofferB]

I know next to nothing about the subject, but it's a very neat project!

What kind of temperature are you looking to heat to?  I'd imagine something like 60C or so wouldn't be too much of a problem, just taking a control circuit from a OCXO and beefing up the drive transistors for the larger chamber, and it looks like you already have insulation that could be suitable for that sort of temperature.

Thanks! You could absolutely do it like an OCXO controller, just an old school thermistor/op amp based analog heating control, but you can get cheap ready-made 7-segment display modules where you can just set the temp and then it toggles a relay to a heater, and honestly they're so cheap I don't think I'll bother doing it myself. I wouldn't trust them at mains voltage at 10$ from china, but for a 12V 50W heater? I think It's ok.

Ambitious project with a number of clever hacks! A couple of minor points/suggestions:

Helium for balloons often contains up to 20% air to reduce the likelihood of asphyxiation. He for welding should be sufficiently pure, but it will also cost a whole lot more than Ar.

Temperature change in the oven needs to happen as evenly as possible, and depending on how hot you really need to get it (assuming much less than 400C) then a suitably modified hair dryer might do the trick. I'd mount it pointing down in the center of the column so that the resulting air flow is turbulent and wraps around the chamber sides before hitting the column circumference. Or you could use a fan from a table-top convection oven and nichrome wire/ribbon as the heater, but still avoid pointing heated air directly at the column.

A TCD is a workhorse of a detector, but in the words of Elzar the Chef, break out the spice weasel and kick it up a notch by making a mass spectrometer. As I recall from working on a similar device years ago (for laying down patterns of ions on a flat target), you "just" need to ionize this gas flow from the column, capture the electrons, then use a precisely controlled electromagnet to deflect the ion beam onto a negatively charged plate using the amount of current and gas flow rate to infer the mass of the ions being deflected. The CC source for the magnet likely needs sub-mA resolution and must be extremely stable, but at least it doesn't need to be terribly fast so a chopper-stabilized op-amp driving a BJT should do the trick.

Thanks for the kind words! My plan at the moment is to get a cheap thermocouple thermostat with relay output and then have a few heatsinked power resistors in the oven. A small fan on the inside might help, but since I don't plan on temperature programming, I can afford to wait for a stable temperature just by convection.

Ideally I'd like a few degrees resolution between room temp. and 200 degrees C.

So far any air in the helium (I need a gas chromatograph to analyze the composition properly oh the irony) doesn't bother me. As long as it don't burn out the hot filaments or oxidize the column material or analyte I should be OK. I've looked at those little disposable welding argon bottles too, they'd be a good alternative. Eventually I'll add an adsorbent filter on the gas lines with some active carbon and drierite to remove organic impurities and moisture.

Mass spectrometry is more involved than that, most concerning is the cost of the vacuum setup needed (2 stage rotary vane pump + turbomolecular pump). Using permanent magnets to deflect ion beams are very VERY old school these days, having a linear quadrupole (4 parallel rods connected to an RF supply) to make the ions spiral and delay proportionately to mass is the modern way to go. There are just so many offspring projects in that; ultra high vacuum, high voltage gen. for ionization, RF electronics, quadrupole, dynode detector, software etc. it's a very elaborate project in itself. Maybe some day.

Right now i'm working on two new detectors: a flame ionization detector (using propane or hydrogen, more research needed) and a corona discharge ionization detector: a flow cell with two electrodes with 8kV between them, the corona then becomes more conductive when a compound gets ionized in it. See attached!

Very nice, and surprisingly stable.  I've never seen anything that makes an Arduino talk to Excel before, but it seems like it'd be great for prototyping.  Any links to more details on that?

Also, any plans to work on the spin-resonance spectroscopy idea on your Tumblr?

Thanks!

It's an excel macro thing called plx daq by parallax: https://www.parallax.com/downloads/plx-daq. Lots of tutorials for making an arduino talk to it! works great.

I haven't gotten much further on the electron spin /electron paramagnetic resonance spectroscopy, still collecting X-band waveguide hardware. It's kinda on hold for the time being.

[ChristofferB]

There is an instrument called a residual gas analyder,  it's basically a compact mass spectrometer used to analyze high vacuum systems, such as electron microscopes and sputtering chambers. It can operate at a poorer vacuum than a conventional MS, is mUchida cheaper and usually pops up on ebay. I'm considering getting one and modifying it as a GC detector, I suspect it will be as cheap as making one myself and much much better.

--chris

[eslavko]

Hello...
Interesting device. I just wonder if it's possible to build one at home. As my friend has a farm with gassifier I like to know if this can be used for measuring various parameters.

At least measure the ratio of CH4 / CO2 / H2S from digestor
And measure the CH4 / O2 from exhaust gas of engine.

For column I think to use PTFE tube 2mm internal and 4mm external diameter somewhere 1 to 2m long. For packing material I think the zeolite can work well to and what about activated carbon?

For carrier gas I wonder if air can be used. I don't know what real demand is but maybe hydrogen (electrolysis on demand) is option too.

For detector I see that you use full bridge with 4 elements. I think that should start with halfbridge. If understand correctly The best is to have input and output thermaly coupled. I think that I don't need the oven but I need to prevent temperature fluctuations.
Some other thoughts?

[Kleinstein]

A clean carrier gas helps to also detect small effects. Just environmental air would give quite some background from natural fluctuation. It may be still good enough for a start or if only large gas concentrations like the CH4, CO2, H2S gas is measured with relatively high concentrations (e.g. % level).
Reading absolute concentration is still tricky and would need good control over the injection and gas flow and some calibration measurements.

Hydrogen from electrolysis is used in some systems. It may be more suitable for an even thinner column and relatively low flow.
Reasonable pure nitrogen from a bottle is usually also not so expensive. It depends on the local supply.