Fiber-optic Probes for Mid-infrared Spectrometry

1 Fiber-optic Probes for Mid-infrared SpectrometryPeter J. Melling and Mary ThomsonReproduced from:Handbook of Vibrational SpectroscopyJohn M. Chalmers and Peter R. Griffiths (Editors) John Wiley & Sons Ltd, Chichester, 2002 Fiber-optic Probes for Mid-infrared SpectrometryPeter J. Melling and Mary ThomsonRemspec Corporation, Sturbridge, MA, USA1 INTRODUCTIONC hemical composition sensors in the form of systems incor-porating Mid-infrared Fiber-optic Probes are commerciallyavailable from multiple sources and they provide a widerange of capability. Basically if a technique can be usedin the sample compartment there is a Fiber-optic equiva-lent available. Fiber-optic techniques are quantitative andcan almost always be calibrated. This, combined with theflexibility and ability to measure in situations where takinga sample is not possible, means that fiber-optics provide avery powerful technique to analytical chemists.

2 The Mid-infrared region of the spectrum as defined by chemists is4000 cm 1to 400 cm 1( mto25 m). In that rangeoccur most of the fundamental molecular vibrations andmany of the first overtones and combinations. The bands inthemid-infraredtendtobesharpandhavever yhighabsorp-tivities, with both characteristics being desirable. Becausethe bands are sharp, most small molecules have distinc-tive spectral fingerprints that can be readily identified inmixtures. Also, because individual peaks can often be asso-ciated with individual functional groups, it is possible to seechanges in the spectrum of an individual reagent due to aspecific chemical of the fibers now available covers the fullmid- infrared range. However chalcogenide glasses in theAs Se Te system and chalcohalide glasses in the so-called TeX1system cover most of the fingerprint region(>4000 cm 1to 900 cm 1) and so can be used in chemicalsensing systems.

3 If it is acceptable to use just the O H orN H stretching region or the C H region then it is possibleto consider the glasses in the As S system such as As2S3or the heavy metal fluoride glasses. John Wiley & Sons Ltd, sensing is arguably the first truly commercialapplication of Mid-infrared transmitting optical fibers. Theavailability at an economic cost of chalcogenide glassoptical fiber with losses of less than 1 dB m 1over mostof the spectral range has made possible the developmentof commercially viable sensor systems using mid-infraredfiber-optics. infrared (IR) fibers provide the capability of taking the spectrometer to the sample instead of thetraditional method of taking the sample to the ability to place Probes in reaction flasks or reactors hascreated a whole new paradigm for molecular based on IR fibers are being used on a daily basis inuniversity laboratories, for industrial process development,and in industrial quality control laboratories.

4 They arebeing used because they conveniently provide importantinformation that is not readily available by any can be constructed using any of five basicsensing schemes: transmission, reflection, grazing anglereflection, attenuated total reflection (ATR), and a variant onthe ATR effect known as the fiber evanescent wave OPTICAL DESIGN ISSUESTwo major factors underlie the design of a mid-infraredfiber-optic system: overall optical signal level and themagnitude of the chemical signature ( the strength ofIR absorption by the sample). All other questions comeback to these two fundamental problem we have when working in the mid-infraredis that the sources are weak, the detectors are not verysensitive, and the fiber is not low-loss. The output of a Mid-infrared source is fundamentally limited by the black bodycurve, which in turn limits the amount of improvement thatcan be made to sources.

5 Because IR photons are low in2 Optical Conduits for Vibrational Spectroscopyenergy, it is usually necessary to use liquid nitrogen-cooleddetectors to get high sensitivity orD* (and hence a highsignal-to-noise ratio, S/N). In a typical commercial systema liquid nitrogen-cooled mercury cadmium telluride (MCT)detector where the frequency cutoff has been matched tothe fiber cutoff to give the maximum possibleD* commercial fiber for broad-band applications haslosses in the range of a few tenths to 1 dB m 1, with a6to10dBm 1spike at the hydrogen selenium vibration(arising from residual hydrogen bonded to the seleniumin the glass). While this is usable, losses of less dB m 1over the whole range would enable the designof much more capable S/N of a spectrometer is typically defined as thesignal level of a blank 100% line divided by the noiselevel.

6 While that may give an indication of the performanceof the electronics and optics within a spectrometer, what theuser is really interested in is the signal-to-noise performancewhen the absorbance of the analyte is compared to the noiselevel of the system. In this definition of S/N, the design ofthe probe becomes important as it can define the signalto noise performance of the system. A probe design thatgives three times the signal of another probe has threetimes the sensitivity of that other probe at a given IR spectrometers use a finite source and the colli-mated beam emitted from most commercial interferometerstypically has a beam divergence of about 2 . Chalcogenidefibers typically have a refractive index for the core and a numerical aperture between and the IR beam is focused into a fiber it is impor-tant not to exceed the numerical aperture of the fiber (seeFigure 1); beyond the numerical aperture, the beam sim-ply reflects off the front surface of the fiber and is problem is that the minimum spot size that can beobtained without exceeding the numerical aperture of the 60 40 off axisFigure field numerical aperture of a 300 m chalcogenideglass optical is typically a few millimeters in diameter.

7 This meansthat either large diameter fibers (flexible glass rods) or fiberbundles must be used to obtain acceptable levels of opticalthroughput. Fresnel (reflection) losses are also an issue, butat the present time the volume of systems being produceddoes not justify the routine use of antireflective aberration can be a concern if lenses are usedin the coupling optics, as the refractive index of some IRmaterials changes across the Mid-infrared region. For thatreason reflective optics should be used where only area where the system designer has any realchoice is in the configuration of the fiber and in the opticalcoupling methods used. Given the optical requirements,two basic approaches have been taken. One is to use largediameter fibers to construct a single-fiber system and theother is to use fiber bundles to provide superior throughputand Commercial single-fiber designsThe first reported IR Fiber-optic probe was an ATR probebuilt by Ciba Geigy (Basel, Switzerland) for monitoringdye synthesis first commercially availableIR Fiber-optic probe was produced by Specac Ltd (Orping-ton, UK) and has been available since the early 1990s;4however, regular commercial production has now been dis-continued.

8 The Specac probe designs use large diametersingle fibers to carry the signal via a series of lenses either toan ATR crystal or to a transmission head. Thermo Spectra-Tech Inc. (Shelton, CT, USA) is also producing single-fiberprobes for ATR and specular reflection measurement5,6but their performance is limited by the low throughputof single-fiber designs and the fragility of large diameterfibers. The use of adhesive to retain an optical window thatis in solution contact also limits the application of their bentfiber ATR Commercial bundle designsRemspec Corporation (Sturbridge, MA, USA) has beenmarketing Mid-infrared Probes using bundle technologysince early Probes consist of two bundlesthat are intermingled at the probe end and are separatedinto a spectrometer cable and a detector cable.

9 For astandard probe, they use 500 m core As Se Te fiber with50 m thick glass cladding from Amorphous Materials Inc.(Garland, TX, USA). The input cable has seven fibers andthe common end is 19 fibers (the next hexagonal pack)leaving 12 for the return or detector cable. The commonbundle is housed in a 6 mm diameter stainless steel shaftFiber-optic Probes for Mid-infrared Spectrometry3that is polished at one end and mated to a splitter assemblyat the other end. Flexible armored cables are then used forthe remainder of the length. The use of such large bundlesmakes the Probes more expensive to fabricate than single-fiber Probes but there are two major advantages: increasedsignal throughput that results in lower noise levels in thesystem and the ability to use smaller diameter fibers whichhave superior mechanical properties.

10 The configurationused also allows a uniquely flexible choice of samplingheads for the probe (Figure 2) including ATR, transmission,and grazing angle specular Fiber-optic reflection probe hasalso been developed9which comprises a 19-fiber cable tobring the signal from the spectrometer to the probe headand either another 19-fiber cable to return the signal to aremote detector or a detector is mounted directly on theprobe (Figure 3). In this probe the signal exiting the fibercable is collimated using an off-axis parabola, and directedtowards the sample at 80 from normal. It is then refocusedby another off-axis parabola into the return fiber cable or adetector crystalSealEnd capATRS pacerFiber bundleReflectionWindowMirrorSample volume TransmissionFigure interchangeable heads for a bundle Fiber-optic of the optics for grazing angle Spectroscopic issuesFor those interested in metal carbonyls or the isocyanate orthiocyanate linkage the strong absorbtion by the H Se bondat around 2300 cm 1can be a problem.