Using the Internet as an analytical instrument

TRAC 2696 17-5-01

internet columnUsing the Internet as an analytical instrumentKimberly L. Glenn, Dana Rodgers, Robert A. Lodder, Cynthia L. Banyon*Department of Chemistry, University of Kentucky, Lexington, KY 40506-0055, USA

Analytical chemists are presented daily with a vari-ety of real-world questions in which the chemicalidenti¢cation and quanti¢cation of a sample is thesimplest part of the problem. In the past decade,the Internet has moved from the realm of computerscience to all areas of life. The power of the Internetto augment traditional chemical analysis is begin-ning to be tapped to move the instrumental analysiscourse beyond simple chemical identi¢cation andquanti¢cation of ersatz samples to complete solu-tion of problems. Complete solution of a problem isachieved when the analyst determines not just whatan unknown sample is, but what it is used for andwhy. In contrast to analysis of the same laboratorymixtures year after year, complete solution of realproblems intensi¢es student interest in laboratoryexperimentation, especially when a solution to aseemingly dif¢cult problem is achieved in one after-noon. The result suggests that instrumental analy-sis laboratories should install Internet computers inthe lab beside the usual instruments and teachInternet techniques along with instrumental theoryand practice. z2001 Published by Elsevier Sci-ence B.V.

Keywords: X-ray £uorescence; Computer; Text searching;Image searching

1. Introduction

One of the most interesting aspects of analyticalchemistry research at a university is the diversearray of problems presented not by the usual fed-eral agencies, but by others. These analytical prob-lems can be used to stimulate undergraduate inter-est in chemical analysis. Periodically the statepolice brings forensic work. Occasionally insur-ance companies investigating fraud or seeking to

identify stolen goods come with evidence. Atother times, attorneys with product-tamperingcases bring in samples. Occasionally, someonewill even present evidence someone claims tohave obtained from an extraterrestrial spacecraft.There is a nearly automatic bias against spendingmuch time in the study of such samples. The per-ception always exists that these studies are a wasteof time and resources, or worse, attract people withpsychiatric problems or will simply result in anaccumulation of hazardous waste in the lab. Toavoid such problems, a prudent chemist examinesonly materials from cases previously screened byresearchers connected with recognized organiza-tions. Science demands some sort of mechanismfor analyzing such samples, even if they are rare.After all, if scientists automatically dismissed everyabsurd idea without doing a single experiment,where would science be?

Psychologists or other professionals sometimesconduct interviews of `contactees' (people whoclaim or believe they have had contact with extra-terrestrial technology or biological entities ). Thispaper reports the results of the examination oftwo hard metal objects submitted by such aresearcher. One object was left on each of two sep-arate occasions by someone who claimed to be acontactee. Were the items really `souvenirs' pickedup as proof of the visit while aboard a spacecraft? Ormore likely, were the objects completely terrestrialin origin? These questions proved irresistible for theinstrumental analysis students.

The shape of the two objects was arguably similarto reported shapes of unidenti¢ed spacecraft ( seeFig. 1). The metal was polished, smooth and veryhard ( the objects could not be scratched by a metal¢le). The objects had a maximum diameter of 0.400inch, and a maximum thickness of 0.320 inch. Thehardness and shape suggested the objects weresomething comparable to tapered roller bearings.The objective of this exercise was to test the hypoth-esis that the objects were bearings by (1) searching

0165-9936/01/$ ^ see front matter ß 2001 Published by Elsevier Science B.V.PII: S 0 1 6 5 - 9 9 3 6 ( 0 1 ) 0 0 0 5 7 - 7

*Corresponding author.E-mail: [email protected]

trends in analytical chemistry, vol. 20, no. 5, 2001 219

TRAC 2696 17-5-01

for manufacturers of similar bearings, and (2) byexperimentally determining whether the chemicalcomposition of the objects was similar to that ofcommercially available bearings using scanningelectron microscopy (SEM) and X-ray £uorescence(XRF) spectrometry.

XRF spectrometry is a bulk characterization tech-nique used for the rapid, simultaneous, and non-destructive detection of all elements heavier than£uorine [ 1 ]. In XRF the sample is irradiated with X-rays and re-emits X-rays with wavelengths that arecharacteristic of the sample composition. In con-trast to £uorescence emission of visible light,which arises from valence (or outer ) electronicshell transitions, XRF is X-ray photon emission bysamples from inner shell electronic transitions. Thinlayers of contamination, especially heavy metals,can often be detected with XRF. The advantagesof using XRF spectrometry for samples include thefact that XRF is non-destructive, and provides quan-titative analysis of bulk elemental composition aswell as trace analysis ( sensitivity ) to parts-per-mil-lion levels. Common applications of XRF includemetal alloy analysis, detection of metallic contami-nation on and in plastics and polymers, and deter-mination of pharmaceutical trace metals.

2. Procedure

2.1. Instrumentation

A Hitachi 3200N variable-pressure scanning elec-tron microscope with energy-dispersive X-rayspectroscopy and 3.5 nm spatial resolution was

used to collect spectra from the metal samples.The mass of the samples was determined usinga top-loading balance. A Dell Pentium computerwith Netscape was used for research on the Inter-net.

2.2. Internet strategy

The search for similar bearings took two routes:(1) a search of the Internet for companies that man-ufacture a similar part, using both text and imagesearches, and (2) the posting of pictures of theobjects to an Internet newsgroup (Usenet ), whereothers could help to identify the objects.

A search engine, AltaVista [ 2 ], was used to seekmanufacturers of bearings. AltaVista permits theresults of searches for keywords and images to bere¢ned by (1) performing a frequency analysis ofother words on the pages with hits from a search forbearings, and (2) including or excluding pageswith these new keywords from the frequency anal-ysis in a Boolean fashion. An Internet newsgroupindex service, DejaNews [ 3 ], was used to identifyan appropriate news group on which to post photosof the objects. A search of the newsgroups for bear-ings using DejaNews suggested that a group calledrec.crafts.metalworking would be very appropriatefor posting the pictures, because many postingscontaining the word `bearings' or similar wordswere found in that newsgroup. However, no post-ing of binary ¢les ( including programs and pic-tures) is permitted on rec.crafts.metalworking.Binary ¢les are posted on a web server at www.metalworking.com using an automated `dropbox'. Users of the newsgroup with binary ¢les topost put the ¢les in the drop box, and then post amessage to the newsgroup giving the ¢le name inthe drop box. Other users of the newsgroup canFig. 1. The metal sample on the right was brass-plated.

Fig. 2. The ballcone design combines the burnishing abilitiesof balls and cones into one carefully proportioned shape.

220 trends in analytical chemistry, vol. 20, no. 5, 2001

TRAC 2696 17-5-01

Fig. 3. A: XRF spectrum of smooth metal of unplated ballcone. B: XRF spectrum of erosion spot of unplated ballcone. C: XRFspectrum of smooth metal of plated ballcone. D: XRF spectrum of erosion spot of plated ballcone.


TRAC 2696 17-5-01

Fig. 3 ( continued ).


TRAC 2696 17-5-01

then access both the text posting and the binary ¢le.Analysts should check the FAQ (a frequently askedquestions list with answers) of each newsgroup forany special posting rules before making a post [ 4 ].The FAQ or a pointer to it is usually posted to anewsgroup every week or two.

In this experiment the Internet functioned in partas a `telephone party line' that connected the stu-dents to other humans who recognized the imagesof the ballcones. Ideally it would be possible to ¢ndthe ballcones using only an image (e.g. Fig. 1) andan image similarity search engine. AltaVista is oneof the few search engines that currently has a similarimage search function. This hyperlink, whenpresent on an image returned by a search, gener-ates a results page containing images that are visu-ally similar to the selected image as well as pageswith content that match the initial search. Similarityis based on visual characteristics such as dominantcolors, shapes and textures. The actual subject mat-ter returned may be the same or it may be com-pletely unrelated. Unfortunately, there is no wayto input an image to the search engine. The indexhas to already contain the image to input, and thatimage must be found using text keywords. Onlythen can images similar to the original image befound. The visual search for ballcones failedquickly because the ballcone GIF image (see Fig.2) was not in the AltaVista index.

3. Determination of the unknownobjects

Using the Internet in this manner, it was deter-mined that the objects were most likely 1/4 inchballcones (see Fig. 2). Ballcones are a tumblingmedium used for burnishing metal parts and stones.Burnishing media like ballcones come in an assort-ment of shapes in order to reach into the recesses ofwork pieces. The AbbottBall Company in WestHartford, CT, USA, is one manufacturer of ballconesthat are apparently identical to the unknownobjects [ 5 ].

To con¢rm the identity of the parts as ballcones,several tests were performed.

1. The objects were determined to be ferromag-netic using a laboratory magnet.

2. The mean and standard deviation of the massof each ballcone was measured (n = 6 repli-cates). The unplated ballcone had a mass of

2.2640 þ 0.0010 g, while the brass-plated ball-cone had a mass of 2.2702 þ 0.0013 g. Thedimensions of the objects were measured asreported above.

3. SEM was performed with XRF spectrometry todetermine the elemental composition of theballcones. The SEM images were unremark-able and were not retained. The parts weresmooth except for microscopic oxygen- andcarbon-containing `islands' that appeared sim-ilar to corrosion. XRF spectrometry wasemployed to identify the elements found inthe objects. The chemical composition wascompared with the composition of manufac-turers of commercial ballcones to con¢rmidentity. The shape matches (down to theexact dimensions of 1/4 inch ballcones)those of at least one commercial manufac-turer, AbbottBall Company in West Hartford,CT, USA. Fig. 3 shows the X-ray spectra of theplated and unplated ballcone and the oxygen-and carbon-containing areas on each ball-cone. The elemental analysis results are simi-lar to the composition of commercial ball-cones made by AbbottBall. Carbon steelcontains small amounts of carbon, as well asmanganese, phosphorous, and sulfur. Thebrass plating is a mixture of copper and zinc.Detailed speci¢cations of AbbottBall's variousproducts are given on the company web site[ 5 ].

4. Conclusion

While the identi¢cation of the items as burnishingmedia with a terrestrial use does not logically ruleout the possibility that spacecraft could include pol-ishing equipment, the Law of Parsimony suggeststhat the ballcones came from a terrestrial shop. Thisconclusion leaves the psychologists and physiciansto discover why the ballcones were represented inthe way that they were. More important, the epi-sode reveals the power the Internet provides ana-lytical chemists in reaching a solution to a seem-ingly tedious problem ( identifying the actual useof an unknown object ) in one afternoon. Instru-mental analysis laboratories might do well to installInternet computers in the lab alongside the HPLC,FTIR and NMR, and teach image searching, the Use-net, drop boxes and `netiquette' along with instru-mental theory and practice. This in turn will enable


TRAC 2696 17-5-01

more interesting laboratory exercises, and providean impressive demonstration of chemical analysisfor undergraduates.

References

[ 1 ] X-ray Fluorescence, http: / /hendrix.pharm.uky.edu /che626/XRF /xrf.html.

[ 2 ] AltaVista, http: / /www.altavista.com/.[ 3 ] DejaNews, http: / /www.dejanews.com/.[ 4 ] The Net: User Guidelines and Netiquette ^ Index,

http: / /www.fau.edu /netiquette /net /.[ 5 ] AbbottBall Company, http: / /www.abbottball.com/.


Documents

Using the Internet as an analytical instrument