Sciencemadness Discussion Board

Building a Frankenstein HPLC system

tandpasta - 13-10-2014 at 10:41

I want to build small home laboratory with a HPLC system. Complete HPLC systems are quite expensive, both new and second hand.

But I can get second hand components for low prices:

Binary pump (Spectra system P2000): €300
Injector €150? (haven't found a price yet)
Column (NUCLEODUR® 100-3 C18 ec (new)) €400
Guard column (new) €180
UV detector (Spectroflow 757 or 783) €150
Total: ~€1200

As I understand it, I can hook up an UV detector to a PC via the RS-232 interface. There seems to be open source software to read the results. Is there anything that would prevent components from different manufacturers from working together? As long as I use proper sized tubing, pressure and such, it should be alright, or not?

€1200 is quite affordable. I expect to spend at least €1000 on auxiliary things like solvents, safety equipment, glassware etc etc.

So a HPLC setup can be had for €2200 or not much more. Does that sound right? Am I missing something?

And am I correct in saying that it's impossible to do proper quantitative analyses with a HPLC without reference samples with known purities? If that's true, could I nonetheless differentiate between a sample with 20% purity and one with 80% purity?

I will mostly use it for testing health supplements. I will have a suspicion of what should be in my samples, but I may come across unknown substances from time to time. I've read that a mass spectrometer can detect the presence of more types of compounds. Would a MS be a better investment than a UV-S?

Thanks for reading!

[Edited on 13-10-2014 by tandpasta]

Little_Ghost_again - 13-10-2014 at 11:18

I dont know but I have a hplc column for sale and a column temperature control system (heat and chill).
I dont know much about it all, I know it works and I think the column is almost new, I will get you column type etc if you want.
I wont want much for it. if that helps

Dr.Bob - 13-10-2014 at 12:24

It is possible to build an HPLC from parts, not easy to make it work and then keep it working. Reading the data may be tougher than appears, and the pumps are a pain, but we have 15 year old systems working here, so it is possible.

A MS is also helpful, but even harder to make work - they require very high vacuum, which normally takes a mechanical pump and then another stage, now normally a "turbo-molecular" pump (often used to be diffusion pumps, even more fun). Good pumps are pricey. And the electronics are not simple, but I would assume that you buy an older used MS with those parts, as those are hard to make at home.

Best is an LC-MS, but that requires 2 instruments to be up and working at once, and that requires some skill and money. The HPLC would be useful for seeing how many compounds are in your material, not really quantitative (due to different UV absorbance for every compound) for varying compounds, but can be calibrated with some work. TLC on a good plate may be almost as good as mediocre HPLC, if done carefully. Plus you can use many stains, iodine, and UV to look for peaks.

tandpasta - 13-10-2014 at 14:54

>not easy to make it work and then keep it working

How come?
Leaks, pressure problems and inconsistent measurements seems to be simple to prevent, at least according to a book I'm reading right now (HPLC: A Practical User's Guide)

>Best is an LC-MS, but that requires 2 instruments to be up and working at once, and that requires some skill and money.

Would MS be more suitable for approximate quantification without reference samples? And what kind of price difference are talking about?

>TLC on a good plate may be almost as good as mediocre HPLC

What makes a HPLC system mediocre? Column quality? Spectrometer quality? Pump quality?

Chemosynthesis - 13-10-2014 at 21:29

The HPLC specialists I knew were constantly fiddling with the plumbing. I forget the interesting names they had for little gaskets and parts, but whenever I spoke, they were ordering parts. I only used them on occasion.

Dr.Bob - 14-10-2014 at 05:05

LC generates a set of peaks which allow some quantification of area. If the absorbance of the compounds is similar, the peak area is proportional to the percentage of the material. This is similar to GC with a simple thermal detector. Similar compounds is based on absorbance, which is often similar in related molecules. I am speaking of UV peak detection, there are other ways which are more or less sensitive, like Refractive index and ELSD. Not all compounds absorb UV light.

MS generates a set of MW masses, which are proportional to the MW of the material and its ability to ionize at a set flow, temp, charge, etc. Very similar looking molecules can have ionization potentials which are many MAGNITUDES different. So identifying what components are in a mixture by MS is doable (if they ionize at all), but quantitating them is not trivial. An LC-MS has the advantage that each peak can be identified, each MS slice if for only one peak, so easier to decypher, as well as that you have at least two measures of the peak area, via UV and MS (you can integrate the MS signal, but much harder to do well, as very quick data generation, so you need good DSP hardware and software.)

Mediocre LC is what you get with older equipment, non-perfect or non-optimized columns, imperfect solvent systems, poor UV detectors, etc. It takes a lot of work and money to maintain an HPLC perfectly, determine the best solvent gradient and columns, buy the best equipment, software, etc. WHat older LCs provide (like the ones that I use) is often mediocre results. And the cost to improve the quality goes up exponentially as you want better quality. And the best solvents for LC are often expensive and it uses a fair amount of solvent. If you want to purify compounds you can go to flask chrom, but that is more for preparative separations than analysis.

Since TLC has pre-made, consistent plates available; the solvent systems are usually simple and involve only small amounts of solvent; and the detectors are simple stains and UV light, the cost for a "professional" result is pretty low.

phlogiston - 14-10-2014 at 06:42

Software problems can also be a major nuisance, especially with old systems. Don’t underestimate that.

Leaks, clogs, broken injection needle/robot, pump seal leaks, contamination, the list of problems is endless. Although often simple to resolve in principle, it can consume vast amounts of time which is annoying if you are trying to get something done and instead find yourself fixing the instrument all the time.
There are big differences in this respect between manufacturers/models. Some systems are very robust while other systems require constant effort to keep them in working order.

MS is much more expensive than photometric methods. To purchase, but also to operate (gasses, energy, pump wear). And it is much more sensitive to (inevitable) contamination. Without a reference sample of known concentration (a standard) you cannot use MS for absolute quantification. I would say get LC setup first using a simpler type of detector. You can always add MS later if you so desire/need.

The main advantages of MS over other detection methods are that you (1) can be more confident that you are measuring the right components (2) are less dependent on good chromatographic separation (you have the additional mass dimension) and (3) can get structural information about unknown compounds.

DrMario - 16-10-2014 at 14:29

Quote: Originally posted by tandpasta  

>Best is an LC-MS, but that requires 2 instruments to be up and working at once, and that requires some skill and money.

Would MS be more suitable for approximate quantification without reference samples? And what kind of price difference are talking about?


MS systems are very expensive, and as someone else mentioned, you need a rotating vane pump and a turbopump. MS need to be calibrated often, and are extremely susceptible to contamination. Today we did some calibration of an ion trap, and after a dozen measurements, I was surprised to see one of the peaks from the commercial calibrating compound, still visible. If you don't know what you're doing, your MS spectra will be less than useless.

Someone else mentioned simple thin layer chromatography as a cheap and relatively good option. I agree. Even though in our division we have many LC, LC-MS, GC-MS and separate MS devices (we have 8 MS's in the MS lab alone), we still use TLC for a lot of our research.

tandpasta - 23-10-2014 at 05:59

Thank you all for your insights. Right now I'm starting to lean towards TLC, it seems a lot cheaper and can do a lot of things that HPLC can do.

I don't quite understand it yet, though. I think that there are 2 ways to do TLC?

1. Fully manual: blot samples on the plate; put the plate in the development chamber; wait; look at the results, with uv light if necessary; consult handbook to interpret results
2. Fully automated (HPTLC): put samples in machine; wait; results

Is this correct? Is there no "in between" method? Where you can scan the TLC plate after developing, but without the costs of an automated sampler and developer?

DrMario - 23-10-2014 at 09:35

Quote: Originally posted by tandpasta  
Thank you all for your insights. Right now I'm starting to lean towards TLC, it seems a lot cheaper and can do a lot of things that HPLC can do.

I don't quite understand it yet, though. I think that there are 2 ways to do TLC?

1. Fully manual: blot samples on the plate; put the plate in the development chamber; wait; look at the results, with uv light if necessary; consult handbook to interpret results
2. Fully automated (HPTLC): put samples in machine; wait; results

Is this correct? Is there no "in between" method? Where you can scan the TLC plate after developing, but without the costs of an automated sampler and developer?


I am not at all sure of what you mean by "scan the TLC plate", but maybe you mean "analyse".

Anyway, there is nothing between good-old manual TLC and HPLC.

However, TLC can be fully automated (including sample preparation) and the spectrum analysed automatically, by robots that are commonly called "grad students". These are surprisingly economical robots, that require only a small monthly fee. :P

/ramblings of a post-doc

Dr.Bob - 23-10-2014 at 12:11

Quote: Originally posted by tandpasta  
Thank you all for your insights. Right now I'm starting to lean towards TLC, it seems a lot cheaper and can do a lot of things that HPLC can do.

I don't quite understand it yet, though. I think that there are 2 ways to do TLC?

1. Fully manual: blot samples on the plate; put the plate in the development chamber; wait; look at the results, with uv light if necessary; consult handbook to interpret results
2. Fully automated (HPTLC): put samples in machine; wait; results

Is this correct? Is there no "in between" method? Where you can scan the TLC plate after developing, but without the costs of an automated sampler and developer?


I have tried "automated" TLC and such, it is a waste of time and LOTS of money. As stated, grad students are the cheapest robots made, and even paid people can run most TLC faster and cheaper than any fancy system. If you have the money to buy an automated TLC system, then buy an LC-MS.

If you are on a limited budget, as I am assuming you are, then TLC is cheap. You buy some TLC slides (price varies but is not terrible), and then find a chromatography chamber or a simple jar and a few mls of solvent. Once run, the plate can be developed or scanned by your eyes and a UV lamp, dye, stain, or by simply heating the plate to char the organics (which method works depends on the substrate.)

The cost of doing a few TLCs can be as little as $50 for a whole bunch of plates to run 100's of samples (You can often get 5-12 samples per small plate-2.5 x 5 or 5 x 10 cm in size). An LC or MS starts at $1000's used, with many other costs and then goes up to $10,000+ to $100K for LC-MS. If you just have a few samples to test, you will be much better off to find some one at a local testing lab or university to analyze the for you.

tandpasta - 24-10-2014 at 10:13

Aha.

DrMario, I've read about slit scanning densitometers being used with TLC, that's what I meant by scanning.

Is it hard to learn how to analyze plates? I suspect that it takes quite some skill. Or is the Rf value the only thing to worry about?

The TLC plates aren't expensive, you're right. €500 should do the trick for a basic setup. I can't wait to eat/drink stuff, then pee in a cup to analyze the stuff. It probably won't be easy, but it will be fun.

DrMario - 24-10-2014 at 12:14

Quote: Originally posted by tandpasta  
Aha.

DrMario, I've read about slit scanning densitometers being used with TLC, that's what I meant by scanning.

Is it hard to learn how to analyze plates? I suspect that it takes quite some skill. Or is the Rf value the only thing to worry about?

The TLC plates aren't expensive, you're right. €500 should do the trick for a basic setup. I can't wait to eat/drink stuff, then pee in a cup to analyze the stuff. It probably won't be easy, but it will be fun.


€500? Where did you get that figure? It's closer to €50 (an order of magnitude less):
On eBay you can find Merck TLC plates of various types at very reasonable prices. Best choice for a researcher on a budget is to buy the aluminium TLC plates, as they can very easily be cut with scissors into narrower strips. That way, a single 20 cm x 20 cm plate can be used many times. Five of these, with shipping, will cost you €30. These plates have silica gel with F254 UV indicator.

Then you need some solvents. A great and cheap non-polar solvent is mineral turpentine (also known as mineral spirit) which can be bought in any hardware store across the EU.
For a polar solvent your choices are a bit more limited, but 1-propanol (isopropanol) can be bought in pharmacies, and for a more economical alternative, Sinol or Marinol - basically, denaturated alcohols (mostly ethanol and a bit of 1-propanol) can be bought in hardware stores, and even petrol stations.

Lastly, you need a glass jar - an old jelly jar will do great! Or if you want to be fancy, use a beaker with a watch glass on top.

The total price of the above is somewhere between €40 and €70, depending on quality and quantity of solvents, and whether you just use an old jelly jar or a borosilicate glass beaker. But really, the jelly jar will probably work much better than the beaker, especially if you still have the original lid of the jar.

For an extra (bonus points), you could get a handheld LED UV lamp or a nail polish UV lamp. Either of them can be had for about €12 including shipping from China.

tandpasta - 25-10-2014 at 01:37

The €500 figure is for the the whole setup. So plates, solvents, micro-pipettes, sample containers, disposal containers, safety equipment etc etc. So basically everything that's needed to work in a reproducible fashion.

I think I'll go with HPTLC plates, €142 for 25 plates of 10x10cm. Getting lab grade materials is not a problem. CarlRoth has every solvent I could possible need.

smaerd - 25-10-2014 at 04:24

The only real advantage of HPLC would be that it's largely reverse phase making solvents cheaper and maybe easier to dispose of in a lot of cases. The disadvantages are obviously cost and maintenance as stated above.

HPTLC is nice for really tough separations and analysis or whatever but again, for most of what you will need to do at home there's really no point. Especially if you're still learning how to run TLC's. Why ruin an HPTLC plate learning how to spot when you can ruin one worth 1/4 its cost.

Also you probably don't need micropipettes. Granted they can be 'handy' for lots of tasks, but for basic TLC, you can get away with capillary tubes (TLC "Spotters"). Developing containers don't need to be as elaborate as the companies who sell you glass-ware want you to think. Granted I did buy some second hand TLC chambers and am very happy with them (less likely to drop a plate, less solvent etc), but flat bottom beakers and a watch glass work fine.

Solvents are important but it really depends on what you work with. For me personally, EtOAc and Hexanes work great for damn near anything I use. Sometimes a drop of glacial acetic acid or triethylamine helps with tailing. I largely work with polar molecules though. Anyways welcome to the world of chromatography.

benzylchloride1 - 26-10-2014 at 20:03

If I was buying a HPLC system on the cheap, I would go for a either a rack mounted Waters 600 system or two 510 pumps with a 484 detector. Avoid buying an autosampler unless you are running lots of samples, way too many things to break. The 510 pumps are very robust, you will need to probably replace the seals and possibly check valves if the system does not hold pressure. Use an old HP integrator, this will give you the chromatogram and peak areas for quantitation, these can be had very cheaply on Ebay. An older CRT 600 based controller, pump, rack and Ralph cable will cost under $500, the detector will be under $250. Rheodyne injector valves are about $100, you will probably need to replace the rotor seal. You will need some parts such as seals, check valves, D2 lamps, this will probably run under $400 if you are shrewd. The only issue with the 600 series is the low pressure mixing valve, the solvents will need to be throughly degassed, you will need a helium cylinder for this. A good system can be easily built for around $1000 if done slowly, looking for the best deals. Find a local QC lab that does HPLC work and see if they can give get some out of specification columns, often these columns are still usable for research work, but will need to be evaluated for your specific purpose. Good luck with venturing into this area. LC/MS is a totally different beast. You cannot get away with this without a suitable data system. I have ventured into this area to an extent, it is an absolute money pit. I have two single quads, the HP 5970 GC/MS is almost operational. I was almost able to autotune, but the manifold heater regulator has malfunctioned and I am looking for a new one. I also have an HP 5989 MS engine with extended mass range, 3 different LC-MS interfaces including ESI, particle beam and thermospray as well as EI and CI GC MS The system was $400 K new, and the cat's meow of single quads 25 years ago. I bought a second 5989 for parts, I am very close to putting the system together, but need to buy several roughing pumps for the LC MS interfaces. I have about 4 K in this system, probably will have to spend another 4 K to get everything running. There are ways of putting the data systems together, the HP HPIB cards are cheap, $40 on Ebay, HP 486 vectra computers are about $150 and old G1034 chemstation software can be legally bought for around $300.

[Edited on 27-10-2014 by benzylchloride1]

[Edited on 27-10-2014 by benzylchloride1]