Help me determine the chemical makeup of...well...everything!
November 16, 2009 7:12 PM   Subscribe

What type of equipment is necessary to determine the chemical makeup of a given substance, is such equipment available to the general public, and if so at what cost?

I have always been fascinated by shows such as CSI where someone is given sample evidence, a scraping is taken of this evidence, put into a solution, and then a device is used to determine what was actually in the sample.

Is this pure fiction, or do such things exist? What equipment would be necessary to, for instance, tell you what that smear on your shoe is, how much caffeine is in that cup of unknown liquid, or what makes up that pressed pill you bought from the dude on your street corner? Is this the sort of thing that only exists in a television writer's dream, or can someone purchase a device or devices that will replicate this process?
posted by anonymous to Science & Nature (22 answers total) 2 users marked this as a favorite
You can do some of this with mass spectrometry. I don't watch TV, but I imagine what you see there is unrealistically fast, accurate, and portable.
posted by alms at 7:18 PM on November 16, 2009

When it comes to the pressed pill you bought from that dude on the street corner, some variants of this exist. See DanceSafe.
posted by mykescipark at 7:19 PM on November 16, 2009 [2 favorites]

Some of what you see on TV is exaggerated, some of it isn't. alms it right about mass spectrometry; thanks to established databases of chemical signatures, mass spec is able to identify most compounds in a relatively low amount of time and effort. However, scaling this up for complex mixtures is a task that quickly becomes too complicated, as signal overlap can hide the characteristic peaks of particular compounds. What would be more valid would be a series of separation and purification steps prior to analysis to divide up the mixture into smaller and more manageable portions. I'm not that familiar with the show, nor am I a forensic scientist, but my guess would be that investigators usually have a set of given chemicals they might want to look for and stick to that. There are many methods by which chemists and biochemists are able to determine chemical structures and composition. Mass spectrometry, Nuclear Magnetic Resonance, X-ray Crystallography are the primary tools. Per your question of ease of access, the answer is no, the average person doesn't have the funds to own this equipment. It isn't pure fiction, but you're more likely to be better off in the short term to look for specific chemical assays for any particular test you're interested in. It won't be determining structure and composition de novo, but you don't need to rediscover caffeine to recognize it.
posted by LifeEngineer at 7:36 PM on November 16, 2009

I'm a biologist, so I interact with these techniques only tangentially, but here's one take.

Like alms said, one technique is mass spectrometry. This is sufficient when knowing the molecular mass of something is sufficient to distinguish it from similar objects. The technique, for example, is surprisingly effective at identifying different proteins.

Another technique people use is NMR spectroscopy, which is feasible only for identifying smaller things. Essentially the configuration of atoms in a given molecule gives rise to a relatively unique "spectrum" of absorption when it is placed in a magnetic field. This is how organic chemists identify substances, I believe, and is probably the magic technology they are normally showing on CSI.

There are, of course, a number of much more old school techniques for identifying things. (E.g. lots of chemical tests that can detect presence or absence of some particular chemical property.) But these are probably more specialized.

Both mass spectrometry and NMR spectroscopy require expensive equipment and are difficult to interpret computationally, so it's not the kind of thing you're likely going to build in your garage, unfortunately.
posted by dubitoergosum at 7:41 PM on November 16, 2009

There are s ton of different instruments that can help you determine different constituents of samples. Different classes of compounds will work better in different types of instruments. Some things won't work at all in some instruments, and depending on how complex the sample is it can get pretty complicated to analyze. There are plenty of cases where you could never resolve all the components of a sample just by bunging it through a machine. If you have a pretty good idea of what a relatively pure compound is you could in many cases very quickly verify whether the compound was what you thought it was. Usually you have to do some separating out of the constituents by bench chemistry sorts of means (test tubes, bunsen burners, filters, you know) first. For example, that mystery pill would probably contain a lot of filler that would just gum up your instrumentation results, you would try to isolate the active compound first.

When I did mystery compound stuff in chemistry labs in college I did a lot of bench chemistry tests, i.e. how does it react in some particular chemical reaction. Something as simple as litmus paper and similar tests (where a chemical reaction creates a measurable reaction e.g. a change in color in a standard sample). In this case the cost could be simple glassware and reagents costing less than a hundred dollars and yes, commercially available drug verification tests are available although I personally wouldn't, ah, trust my life on them.

High level analytical instrumentation can cost from thousands of dollars to easily hundreds of thousands of dollars.

Like everything else in CSI the analytical science is largely super-simplified bunk.
posted by nanojath at 7:44 PM on November 16, 2009 [3 favorites]

It really depends on what you're looking for. For something smallish (organic molecules) that's a known compound, there's an old-school technique that could be done at home (with access to the right chemicals):

You do a couple tests to see what kind of compound it is (acid, alcohol, aldehyde, etc).

Then you take a melting point and look in a big book of compounds, which are conveniently organized by melting point. You pick the five or six closest to your compound as likely suspects.

Then, you modify the chemical synthetically into one of various derivatives, and take a new melting point. The book conveniently has the melting points of those derivatives listed on the same page as the original.

If the melting point of the derivative matches one of the originals and not the others, you have your compound. If not, it's either unknown or you screwed something up along the way.
posted by Dr.Enormous at 7:49 PM on November 16, 2009

There is also X-ray crystallography which can be used to determine the actual structure of molecules. I don't think CSI would use this very often but this is a very powerful technology. Regardless of the method it is never as easy as they depict it on CSI. bbbbbbb
posted by Procloeon at 8:00 PM on November 16, 2009

When a client submits a urine specimen that is positive for a class of drug (benzos, opiate, THC, etc.) to drug court they are given the opportunity to have the specimen retested to confirm whether it was a true or false positive. If they want a retest and the specimen is less than 30 days old we'll have it shipped to a lab where a gas chromatography/mass spectronomy test is done that provides a highly specific result that yes, could totally tell you exactly what the active ingredients in a pill you bought on a street corner are. The test costs $12 and if the specimen is confirmed positive the client pays for it, which is a disincentive for clients to constantly bog the system down with retest requests if they know they used. The reality of having this test done means going through a bunch of different bureaucratic channels. I have to notify my supervisor of the client's request, he in turn has to contact the person in the parole and probation department that handles all those requests, that person has to contact parole's lab to have the specimen pulled and then parole's lab needs to send the specimen to the lab where they do the GCMS. Then the result gets communicated back through all these same channels in reverse and I get the result usually 2-3 weeks after the initial request.
posted by The Straightener at 9:08 PM on November 16, 2009

Is mass spec offered as a service by anyone? I walk in with a sample, and walk out with a printout?
posted by intermod at 9:09 PM on November 16, 2009

I happen to be at a conference for drug court professionals right now so I'll see if I can find that out tomorrow, I think there's a bunch of Varian reps around.
posted by The Straightener at 9:15 PM on November 16, 2009

I think that toxicology and forensic labs actually typically use a GC-MS, rather than just a mass spectrometer. GC stands for "gas chromatograph," which helps separate the different compounds in a sample. These are then fed into a mass spectrometer for further characterization by molecular weight. GC-MS is useful for a wide range of assays, but they aren't cheap to acquire or operate.

There may be cheaper ways to identify and measure certain compounds, by, for example, seeing how much light is absorbed at various wavelengths. This can be facilitated further with special dyes or other indicators that behave in a certain way depending on the presence of your target substance.

Really though, figuring this stuff out is a big part of what it is to be a chemist.
posted by Good Brain at 9:16 PM on November 16, 2009

This is what the field of analytical chemistry exists for. Some of the techniques are very cheap while others are very expensive.
posted by grouse at 9:34 PM on November 16, 2009

I'm no chemist so correct me if I'm wrong, but to put it in TV terms I think that the real process of identifying a substance is going to be much more like the process of identifying a disease as depicted on the show House - there are tens of thousands of specific tests you could run so you have to apply broad principles to narrow it down to a few possibilities. And there's clever intellectual banter and drama in the course of that.
posted by XMLicious at 9:57 PM on November 16, 2009

And there's HPLC. At the soap factory where I work, this is the primary method we use to determine the %whatever in our products before release. Some products have multiple active ingredients, and we can assay them all at once, getting identity* and concentration of each. Those ads for Coca-Cola touting their "secret formula"? -- um, no, not secret. Not in the age of HPLC. You probably wouldn't find one at a garage sale, though. We have twelve at the soap factory, and they go for about $40K each, plus the computer to run it, plus the columns and chemicals.

*Identity is not that simple, as posters above point out. If you have a pure sample (a standard, or positive control) of what you think should be in your mystery substance, you can see if you have a match, and that's what experimenters always like to do if they can.

In school, I used to have to identify a mystery compound (from a list) based only on an IR or an NMR spectrum, dozens of them. This is actually really fun. Seriously, try it: don't worry about the theory, just learn the rules and try to solve one. If you like Sudoku, you'd love NMR spectra. Of course, in real life, the NMR software solves the puzzle for you.

Lab equipment is so fun to play with, it is a shame you can't have it at home. When it turned out that my sister had a hereditary deletion mutation associated with breast cancer, and she was wondering whether I possibly had it too, it really frosted me to contemplate that I could look up the sequence, design some primers, isolate the DNA from my blood, PCR it, run it out on a gel, and see for myself -- but no, I've got to go to the fucking doctor, and sit there being anxious while his lab techs have all the fun.
posted by Methylviolet at 10:19 PM on November 16, 2009 [4 favorites]

One big problem with GC-MS is preparing samples - GC requires a gas sample, (HPLC, which Methylviolet mentioned, requires a liquid sample) so if you have a relatively solid sample (e.g. that smear of dried mud on your shoe), you have to convert it into a liquid. (The GC itself will have the capability to vaporize that liquid on injection, to get the gas sample it needs). So a lot of the work is in finding a way to prepare your sample that keeps contaminants out of the equipment but retains whatever compound you're interested in - this is much easier to do when you know what compound you're interested in. Additionally, as LifeEngineer mentioned, many compounds can break down into similar pieces that will give confusing results, so you also need to know that there is nothing present that might give similar result.

There are labs that will do analytical chemistry for you, for money, although if you just give them a mystery sample and want them to tell what's in it, it could run to a lot of money. I used an outside lab for a relatively simple test for fats, oil and grease present in a solid sample, and the cost was $32/test.

Methylviolet - when I was working in lab, I used to daydream about how awesome it would be to have analytical chemistry fingers - Interested in ammonia? No problem! I have an HPLC finger! Elemental analysis? Don't worry, I have an AAS finger! (My right index finger would be a pH meter, of course).
posted by periscope at 10:29 PM on November 16, 2009 [1 favorite]

Aside from what's been mentioned (old-school analytical chemistry, gas/liquid chromatography, NMR spectroscopy, and mass spectroscopy) some techniques that I've heard of are FTIR spectroscopy (eg) and Raman spectroscopy (eg). As Good Brain says, these techniques can be used in combination— use solvents or pyrolysis or something to break a sample into pieces, chromatography to separate out the components, followed by something else to identify each component.

Googling on any of these terms + "analytical" will get you a bucketload of companies willing to analyze samples for you (usually in specialized fields of course).
posted by hattifattener at 10:59 PM on November 16, 2009

The problem with using melting point to identify compounds is that you have to have a pure crystal first. This is relatively straight-forward when you're synthesizing something in the lab. But obtaining pure crystals in other circumstances is a much messier problem. X-ray crystallography is also limited to pure crystals.

There are many different types of spectroscopy, but the most widely used is NMR. Gas chromatography is also a very common technique. Neither need pure crystals; they use solutions, which allows for more leeway in terms of purity. The latter can handle relatively complex mixtures if it's combined with liquid chromatography (and it often is).

Mass spectroscopy can be used to identify either chemical compounds or biological samples, whereas NMR and GC are generally limited to the former. MS also requires very little sample, and combined with in-line liquid chromatography, it can handle a fairly complex mixture. The biggest limits are time and expense. Yes, you can find labs that will do MS work for you. No, you can't afford it. Some of the older or more basic machines might be getting closer to affordability in terms of run-time, but anything fancy could easily reach into the three-or-four-figures-per-hour vicinity. These machines cost more than your house and are not cheap to maintain either.
posted by dephlogisticated at 11:00 PM on November 16, 2009

Miniaturized chemical sensors are getting smaller and cheaper all the time, but are only useful for very specific questions (e.g. ammonia concentration in the air). There are also lots of one-off quantitative tests for specific compounds under specific conditions (e.g. glucose in blood using diabetic test strips, lead paint testing strips, urine drug testing kits, etc.

Also note that if you are planning on figuring out how pure your stash is by sending it to a service chemistry company, if they detect Schedule I compounds in your sample, I believe they are forbidden to tell you the amount, only whether it is present or absent, and you will probably get a visit from some government employees very soon. The supposed goal is to prevent analytical chemistry labs from serving as QC for drug dealers (the Dancesafe link above has a much longer discussion of the harm reduction debate).
posted by benzenedream at 11:11 PM on November 16, 2009

NMR spectroscopy, which is feasible only for identifying smaller things.
Depends what you mean by smaller; complexes of DNA with chemicals or small peptide chains can be analyzed via NMR. No one's about to characterize the the spliceosome via NMR, but the upper size limit's been growing larger and larger.

Here's a detailed answer covering almost all of the various methods used to ID chemicals that I wrote in response to a similar question a while ago. Most of these methods are not things you can do at home: they require complicated, insanely expensive machines. Data analysis can also be difficult: analyzing NMR spectra can be hard for a single large but pure chemical, but analyzing it for a mixture of crap is much harder, and analyzing multidimensional NMR spectra is an art.

The methods you might be able to do at home are old-fashioned analytical chemistry. Take the melting point, note the flame color when you burn it, taste or smell it (yes, they used to do this!), observe how it reacts or doesn't react with a bunch of common acids, bases, solvents, follow a whole tree of "if it reacts with A, then try B, otherwise do C," etc. A fairly time- and resource-consuming process, and not necessarily possible unless you have an extraordinarily well-stocked home lab! It's easier when you know what sort of chemical you are looking for, because you can tailor the tests to see if that class of chemicals is there or not, and avoid unrelated tests.

However, if you're truly trying to ID a mystery sample (e.g. the gunk on your shoe), no matter what methods you're using, you're going to get signals (or reactions) from all of the random crap in your sample, making it hard to tell what's actually there. You'll need to isolate and purify the different compounds as much as possible, then, probably using chromatography (likely HPLC for the small amounts of compound we're talking about), possibly filtering, distillation, recrystallization, etc. You'll then need to characterize the compounds you could purify separately, using NMR, IR, GC/MS, MALDI, analytical HPLC, etc. - if you have enough of the compound left! You'll still probably be able to get some answers in the end, but only after a lot of work.

So yes, the technology necessary to ID a mystery compound exists, but no, it's nowhere as easy as it looks on CSI, and you're unlikely to be able to do it at home any time soon.
posted by ubersturm at 5:24 AM on November 17, 2009

Many FTIR machines come with a database of known compounds. In some cases, this allows near-immediate identification of a substance. The same is true for mass spectrometry and the various types of chromatography (there are published tables of retention times/Rf values).

If the substance is not in a database of spectra, several techniques are generally used in combination. A mass spectrum will show the masses of various fragments of the molecule. For polypeptides, which consists of only about 20 different types of amino-acids joined in various ways, a computer can easily find the complete structure the mass spectrum alone; in other cases, it is much harder (in the '60s, an AI program called Dendral was designed to try doing this, and it has many descendants). Incidentally, Mass spectrometer have truly exquisite sensitivity: at one tab I toured, I was invited to tap a glass rod against the side of a pot that had once contained cocaine, but now appeared clean, and then did it into 1ml of solvent - the concentration of cocaine in the solvent was sufficient not only for identification, but also to saturate the instrument.

Infra-red spectroscopy looks for particular types of bonds, and can indicate whether a molecule is, say, a ketone or an alcohol. Essentially, the vibrations of different bonds are associated with different energies, and hence absorb different frequencies of light. It can also do fancy things like measure the concentration of several (known) components of a mixture simultaneously. The price of an FT-IR instrument is on the order of tens of thousands of pounds.

An NMR spectrum looks at atoms of a specific isotope (1H, or 13C, commonly), and determines how many different `chemical environments' it exists in. NMR instruments are big, and very expensive. As well as having a high purchase cost, they have a high operating cost, as they are cooled with liquid nitrogen and helium, and generally require the attention of trained operators.

These three techniques, together with sufficient thought, can go a long way.
posted by James Scott-Brown at 12:28 PM on November 17, 2009

Is mass spec offered as a service by anyone? I walk in with a sample, and walk out with a printout?

If you could get someone to do this, you would also have to have someone interpret it for you. If you look at published papers using mass spec and GC-MS and MALDI-TOF as identification techniques, you will see why. It's a bunch of "peaks" and you need to have standards in order to know which is which. Or, you can separate an interesting fraction from your sample and (if it's protein or nucleic acid) take it to be sequenced.
posted by Knowyournuts at 12:39 PM on November 17, 2009

For polypeptides... a computer can easily find the complete structure the mass spectrum alone

Bingo. MS is becoming the workhorse of proteome investigation. Shoot in a biological sample and the MS spits out a long list of peptide masses. The software then queries this against online databases of sequenced proteins and returns with a list of all the identified proteins in your sample. The accuracy is quite good. The main limitation is that the technique is biased towards highly-abundant proteins; the low-abundance proteins tend to get lost in the statistical noise.
posted by dephlogisticated at 8:21 PM on November 17, 2009

« Older Can I negotiate rental agency tenant-changeover...   |   Where do I find a high quality card to... Newer »
This thread is closed to new comments.