INFRARED SPECTROSCOPY

Spectroscopy Europe 15/6 (2003) ©Spectroscopy Europe 2003 17

IntroductionInfrared spectroscopy (IR) is one of

the techniques for investigating matterthat has a very long history of use inboth research and applied sciences.

In the early 1980s Fourier transforminfrared (FT-IR) spectrometers werecoupled with a microscope,1 sincewhen the success of IR microanalysishas increased rapidly. Nowadays, IR-microscopy has gained wide acceptancein various scientific communities as aqualitative microanalytical tool foridentifying the chemical make-up ofisolated contaminants or characterisinglocalised areas of much larger samples.The FT-IR microscopes now commer-cially available include many of the fea-tures of research-grade optical micro-scopes: polarisation, Nomarski Differ-ence Interference Contrast, fluores-cence, as well as sophisticated softwarefor generating chemical images andanalysis. Perhaps one indicator of thegrowing acceptance is the recognitionby biologists of the importance of IRmicrospectroscopy as a potential toolfor biodiagnosis.

Present day FT-IR spectrometers andmicroscopes are well matched anddeliver high performance. However,signal-to-noise ratio (S/N) require-ments often limit the smallest practicalspot size to ~20 µm diameter. This lim-itation is not the consequence of dif-fraction or aberration in the optical sys-tem of the IR microscope, but ratherdue to the low intrinsic brightness(defined as the photon flux or poweremitted per source area and solid angle)of the thermal infrared source.

The diagnostic potential of IR-microspectrometric data strongly de-pends on the quality of the spectraacquired.

As with many of today’s analyticaltechniques, FT-IR spectroscopy isincreasingly moving into imaging tech-nology enabled by detector develop-ments that were originally availableonly to the military, for example, formissile guidance technology. Thesedetectors have improved dramaticallyduring the past decade to a point thatmercury-cadmium-telluride IR focalplane array detectors are now routinelyused with infrared spectroscopy.

Higher signal-to-noise values areoften considered as a fundamental mea-sure for achieving a better detection andspectral quality. For most infrareddetectors, the Noise Equivalent Power(NEP) is proportional to the square rootof their active area.2 Small area detec-tors are now routinely employed in IRmicroscopy. The quality of the spectradepends also on the infrared source.The commonly employed source is ablackbody heated to ~1400–2000 K.To increase further the IR microspec-troscopy capability, brighter sources inthe IR region are required.

For thermal sources, one has littlecontrol over the brightness apart fromraising the temperature or changing theemissivity of the photon source.

A particle, such as an electron, emitssynchrotron radiation when acceleratedor (decelerated). The radiation is emit-ted along the direction of motion of theparticles. A number of synchrotronaccelerator facilities exist around theworld specifically to provide an intensesource of radiation. They have beenexploited, for spectroscopy, in theVUV, far-UV and soft X-ray regions.Synchrotron radiation has been usedsuccessfully in both the mid-IR and far-IR regions for about ten years, and hasan equivalent blackbody temperature inexcess of 10,000 K! It is particularly

attractive for IR microspectroscopy, asexemplified in this article.

IR micro-spectroscopySynchrotron radiation: abright source of infraredphotons

Electron storage rings (or synchro-trons) use magnetic fields to bend elec-trons into a closed orbit. Synchrotronradiation is produced at each of thebending magnets. Infrared radiation isgenerated by electrons travelling at rel-ativistic velocities, either inside a curvedpath through a constant magnetic field(bending magnet radiation)3 or by lon-gitudinal acceleration or decelerationwhen leaving or entering a magneticsection (edge radiation).4 The emittedradiation spans an extremely broadspectral range, extending from the X-ray to the very far-infrared region.

The effective synchrotron source sizeis quite small, and can be on the orderof ~100 µm. Furthermore, the light isemitted into a narrow range of angles.

The intensity distribution, for a par-ticular wavelength, is dependent on themode of emission, and has to be con-sidered carefully, in order to optimisethe extraction geometry and to coupleefficiently all collected photons into themicroscope.

This can be illustrated by consideringthe intensity profile of the beam at twocharacteristic wavelengths, 10 and100 µm, and for a storage ring operatingat an energy of 2.75 GeV, see Figure1(a, b). In Figure 1, two extractiongeometries have been considered

A bright source for infraredmicrospectroscopy: synchrotronradiation

Paul Dumasa and Mark J.Tobinb

aLURE, centre Universitaire Paris Sud, BP 34, F-91898 Orsay Cédex, FrancebCCLRC Daresbury Laboratories,Warrington, Cheshire WA4 4AD, UK

(opening angle: horizontal × vertical):10 × 10 mrad and 20 × 20 mrad.Clearly, the intensity of the emittedsource is not distributed uniformly. Foredge radiation, rings of maximum emis-sion are produced, with increasing sizefor longer wavelengths. At 2.75 GeV(recent third generation synchrotronfacilities have energy in the 2–3 GeVrange), a 10 × 10 mrad (e.g. 1.5 ×1.5 cm2 aperture positioned at 1.5 mfrom the source point) would be appro-priate to collect most of the infraredphotons at and below 10 µm. However,the far-infrared wavelength region willnot be optimally collected [Figure 1(a)b]: for such a purpose, an aperture of20 × 20 mrad is preferable [Figure 1(a)d], but may affect the properties of thestored electron beam. These engineer-ing issues are carefully considered insynchrotron facilities. An edge radiationsource is radially polarised. For the con-stant field emission (bending magnet),the emission is distributed longitudinal-ly, but is a minimum at the centre.Clearly, the collection is not yet opti-mised at 10 × 10 mrad for a 10 µmwavelength [Figure 1(b) e], while a20 × 20 mrad aperture is appropriate forsuch a wavelength [Figure 1(b) g]. Itwould be necessary to increase the col-lection angles to collect, efficiently, thelong wavelength radiation at 100 µm. Itis important to note that by enlargingthe horizontal aperture (they are usual-ly rectangular in existing synchrotronIR beamlines) more photons can becollected. Such sources are highlypolarised (along the longitudinal sec-tion).

The two types of synchrotron IRsources, edge radiation and bendingmagnet radiation, have equivalent fluxand brightness. Depending upon engi-neering constraints of the particularsynchrotron, one or the other is chosen.The synchrotron source has a brightnessadvantage about between two and threeorders of magnitude compared to ablackbody source. At this point, it isparticularly important to note that thebrightness advantage does not mean thatthe synchrotron produces more power,but simply that the emitted photons canbe coupled more effectively into aninstrument such as a microscope. Inspite of the increased infrared through-put of synchrotron microscopes, noinduced damage has ever been detected,even on fragile biological samples.5

Another feature of synchrotron lightis that it is pulsed, i.e. the electrons donot form a continuous distributionaround the orbit, but instead they trav-el in “bunches”. Depending upon thestorage ring, these bunches have lengthson the order of 1–10 cm, and thisresults in pulses of light on a sub-nanosecond time-scale, with whichtime-resolved studies can be undertak-en.

Infrared microscopy with aSynchrotron source

FT-IR microscopes are availablecommercially from a number of instru-ment manufacturers worldwide. Veryfew modifications are needed in orderto adapt a commercial instrument foruse with a synchrotron infrared source.

The synchrotron beam can be efficient-ly collimated and is directed into theinterferometer, which modulates theinfrared light, and then is directedtowards the IR microscope. Since thesynchrotron source produces highthroughput at small aperture sizes, thediffraction limit is achieved when themicroscope’s apertures define a regionwith dimensions equal to the wave-length of interest. However, the use ofa confocal optical arrangement leads to~30% improvement in the spatial reso-lution, in agreement with diffractiontheory.6

Someapplications

Several synchrotron infraredmicroscopy beamlines are operatingaround the world, and several others areunder construction or planned.Applications are multidisciplinary, andthe number continues to increase asthese beamlines become accessible to anincreasing number of users. High spatialresolution, excellent spectral quality(signal-to-noise), and fast data acquisi-tion are the essential features of thetechnique.

Polymer films studiesIt is well known that vibrational

microscopy is a powerful tool for char-acterising polymeric materials, provid-ing both compositional and structuralinformation. Most new polymericmaterials are multiphase systems. Theyinclude homopolymers and copoly-

INFRARED SPECTROSCOPY

18 Spectroscopy Europe 15/6 (2003)

(a) (b)

Figure 1. Calculated intensity profile for infrared emission at 10 µm and 100 µm, either from an edge radia-tion or a bending magnet radiation source (see text). The calculations have been carried out using the SRWcode (O. Chubar and P. Elleaume) using the following parameters: electron energy = 2.75 GeV, electron cur-rent = 500 mA, magnetic field = 1.56 T, straight section length = 6 m, distance to point source = 1.5 m.Profiles have been calculated for two extraction geometries (horizontal × vertical opening angles): 10 × 10and 20 × 20 mrad. Colour code: Minimum = black, Maximum yellow. (For convenience, the profiles calculat-ed for the bending magnet are shown in 2D displays.) This allows the operator to define parameters accu-rately for an optimum collection of infrared photons.

INFRARED SPECTROSCOPY

20 Spectroscopy Europe 15/6 (2003)

mers, polymer blends and compositematerials, in which the physical andmechanical properties of the system aregoverned by the relative amounts ofeach component and/or their morphol-ogy (crystalline or amorphous), andtheir intrinsic properties. However, thetype, size and distribution of the differ-ent domains in these materials, and therelationship or interaction between thedifferent phases are also fundamental. Inthis respect, in addition to the obviousdifferences in the chemical structure of,for example, two polymers in a blend,these materials can present differenttypes of heterogeneity: compositional,structural and morphological.

G. Ellis et al. have studied the poly-morphism of isotactic polypropylene(iPP), using synchrotron infraredmicroscopy.7 In Figure 2 an example isshown which corresponds to a LCP(Liquid Crystalline Polymer) fibre-rein-forced iPP film where the fibre has beensheared at the isothermal crystallisationtemperature. Different crystalline habitscan be observed in the visible image,and infrared spectra were obtainedusing a 3 × 3 µm aperture over the areaindicated. The spectroscopic identifica-tion of the different types of crystallinemorphology is described elsewhere.7

However, in this case in order to differ-entiate between the various regions ofthe iPP film a statistical treatment of thespectra was undertaken by hierarchicalcluster imaging using Cytospec software

(see http://www.cytospec.com/ftir.html). High S/N spectra were obtainedeven with the very small aperture. For amore optimum classification the first-derivative of each spectra was analysedover the 700–1350 cm–1 region. Subtledifferences were observed in the char-acteristic first-derivative spectra, andfrom these three clearly distinguishablefeatures could be located in the hierar-chical cluster images of the sampledarea, which could be associated withdifferent crystalline morphologies.

Human tissues investigationsIR spectroscopy is being employed

increasingly in the study of biomedicalconditions, where it has been shown tobe capable of detecting subtle biochem-ical changes within tissues. The cou-pling of a microscope to a Fourier trans-form infrared spectrometer, comple-mented by the use of a synchrotronsource has brought the potential toexamine tissues at cellular and sub-cellular resolution.

The applicability of micro-spec-troscopy, and hence imaging, in partic-ular to biological and pathologicalproblems relies on the informationbeing obtained at high lateral spatial res-olution. Spectroscopic imaging pro-vides diagnostic information in a visualform, a prospect appealing to physiciansand biologists. Image methods can pro-vide potentially far more information tonon-specialists than their non-imaging

counterparts. However, the analysis anddiagnostic potential of IR imagingstrongly depends on the quality of thespectra acquired. Clearly, a parametersuch as the signal-to-noise ratio, strong-ly affects the image quality.

Hereafter, we illustrate how highspectral quality obtained at diffraction-limited spot size sheds new light intothe biochemical composition and pro-tein secondary structure of human tis-sues: in this case, hair.

Human hair sections have been studiedintensively using synchrotron infraredmicrospectroscopy.8 They are50–100 µm in diameter and their cross-section reveals three major identifiableregions. The medulla is the centre-mostportion of the hair. It is 5–10 µm inthickness and composed of looselypacked, keratinised cells that distributemoisture and nutrients to the hair strand.The medulla can be either continuous ordiscontinuous along the hair length, andoften it is completely absent. The cortexmakes up the bulk of a hair fibre anddetermines the strength of a hair. It is45–90 µm in diameter and composed oflong embedded cortical cells. It also con-tains the hair pigment, melanin. The out-ermost layer of a hair strand is the cuticle,which is less than 5 µm in thickness. It isa dense layer of flat, keratinised cells,which protects the hair fibre. Infraredspectra were recorded across hair fibreswith an aperture size of 3 × 3 µm2.

The chemical images of the lipids(height of peak at 2920 cm–1), displayedin Figure 3(a), show that lipids are pre-dominantly located inside the outercuticle layer, as well as inside themedulla (orange and red regions). Thehigh quality of the recorded spectra hasallowed subtle difference in the peakposition and lineshape of the stretchmodes of CH2 to be observed betweenlipids located in the medulla or in thecuticle. Fuzzy-C means-clusteringimages have been generated in the2800–3000 cm–1 region, and associatedimages of the two main classes of CHxstretch modes are shown in Figures 3(b)and 3(c), which reveal the differentnature of the lipids in this two regions.These differences can be related to therelative chain lengths (CH2/CH3 ratio)of the prevalent lipids in each region.

IR spectroscopy can also reveal thedifferent composition of the protein sec-ondary structures. For hair sections, theAmide I band contour is differentbetween spectra recorded from insidethe cortex and from inside the cuticle,see Figure 3(e). The red spectrum hasbeen recorded inside the cuticle, with anaperture of 3 × 3 µm2. Clearly, the peakposition exhibits a downward wave-number shift, which indicates a higherrelative concentration of β sheet to α

Figure 2. Infrared imaging of various polymorphic forms in a LCP fibre-reinforced iPP film where the fibre has been sheared at the isothermalcrystallisation temperature. The optical image displays, within the bluesquare, the mapped area, with marked steps of 3 µm. From the clusteranalysis of the spectra recorded, fuzzy-C means-clustering imageshave been generated, and each of the image clusters corresponds todifferent crystalline morphologies. The characteristic spectra of eachcluster are also displayed, and subtle differences in spectra can beseen.

helix inside the cuticle. This relativeconcentration is shown in Figure 3(d).

Synchrotron infrared microspec-troscopy appears to be extremely wellsuited to the study of hair biochemicalcomposition, and its variation with age,ethnic origin, sex etc., as well as forstudying the interaction of externalagents (for example, cosmetics) on haircomposition and structure, and is thesubject of several current investigations.

Single cell studiesUsing synchrotron infrared micros-

copy, Jamin et al. were the first todemonstrate that the chemical compo-nents of single living cells, e.g. proteins,lipids, and nucleic acids, can be imagedwith high signal-to-noise at a diffrac-tion-limited spatial resolution.9 Thiswork was pursued further by theextended study of programmed celldeath, i.e. apoptosis, using a combina-tion of fluorescence microscopy andsynchrotron infrared microscopy.10

Detailing the bio-composition andprotein secondary structures inside onesingle cell is a subject of paramountimportance, in order to understand thechanges of the cell cycle, as well as anyabnormality that would ultimately lead

to a disease.To illustrate the sub-cellular capabili-

ty studies of synchrotron infraredmicroscopy, we have studied thebehaviour of single HL60 cells duringdifferentiation. HL60 cells are oftenstudied as an attractive model for differ-entiation, and are used specifically forhuman myeloid cell differentiation.Among several agents used for differen-tiation, phorbol myristate acetate(PMA) induces HL60 cells to differenti-ate into mature monocytes/macro-phages in vitro.

Figure 4 illustrates the main observa-tions. Figure 4(a) shows the opticalimage of one HL60 cell before inducingthe differentiation process. Chemicalimaging of the lipid profile has beengenerated using the peak height at2920 cm–1 [Figure 4(b)]. They areclearly distributed around the nucleus.

After 48 hours of differentiationinduction time, a clear change in theAmide I lineshape is observed [Figure4(c)]. It is suggested that followingPMA treatment, the α-helix content, indifferentiated cell membrane protein,increases. These changes in the mem-brane protein secondary structures areprobably related to the mode of interac-

tion with the inducing agent. By usinga statistical analysis in the Amide I andAmide II band frequency region(1485–1710 cm–1) of the high S/Nspectra obtained with a 3 × 3 µm2 aper-ture, it is possible to separate out thedifferent relative composition of theprotein secondary structure, inside thenucleus and in the cytoplasm. Figure4(d) shows the two cluster images andthe average spectrum of each.

Only the use of a synchrotron sourcecan produce such high contrast chemi-cal images inside individual cells.

Future andperspectives

It is clear that synchrotron radiationprovides a high brightness infraredsource, which is being exploited suc-cessfully in microspectroscopy for highsignal-to-noise spectra and for fast dataacquisition, because of the readilyachievable diffraction-limited spatialresolution. Several infrared microscopesare now implemented at synchrotronfacilities around the world, althoughmany applications are still in their infan-cy. An attractive feature of the synchro-

Spectroscopy Europe 15/6 (2003) 21

INFRARED SPECTROSCOPY

INFRARED SPECTROSCOPY

22 Spectroscopy Europe 15/6 (2003)

tron environment is the capability ofachieving complementary characterisa-tion using other available synchrotron-based microanalytical tools. Amongthem, X-ray microscopy is one of thebest-suited complementary approachesto infrared microscopy.

Several improvements are now beingproposed for synchrotron infraredmicroscopy. Nowadays, focal planearrays detectors are being implementedin microscopes that use a globar source,and the performance of these detectorshas not yet been exploited with a syn-chrotron source. Clearly, the size of thedetector array has to be adapted tomatch the projected size of the syn-chrotron beam, in order to keep thebrightness advantage of this source.These small detector arrays might wellbe available soon, and will allow fasteracquisition data, and a slightly improvedlateral resolution using point-spreadfunction (PSF) deconvolution.

Interesting new attempts have beeninitiated with a synchrotron infraredsource in order to extend measurementsto below the diffraction limit11 by

employing PhotoThermal IR Micro-Spectroscopy (PTMS), see Figure 5. Inthis technique, an AFM-based thermalprobe is used as the infrared absorptionsensor, and promising results havealready been obtained.

All together, the bright infraredsource provided by synchrotron radia-tion has a tremendous future, inmicroscopy and near-field microscopy,and such facilities are available to exper-imenters who are invited to contact thevarious synchrotron infrared beamlinerepresentatives. There are currentlyfour operating beamlines in Europe(Daresbury, UK; ANKA, Germany;BessyII, Germany and LURE, France)and several others are under construc-tion and will be available soon (Maxlab,Sweden; SLS, Switzerland; Elettra,Italy; ESRF, France; SOLEIL, France;Diamond, UK).

It is important to note that applica-tions of synchrotron infrared spec-

troscopy are not restricted to microscopybut also exhibit a tremendous prospect inother fields, such as far-infrared spec-troscopy, time-resolved spectroscopyand two-colour experiments.

AcknowledgementsThe authors wish to thank G.P.

Williams, G.L. Carr, L.M. Miller, G.A.Ellis, H. Pollock, A. Hammiche andM.A. Chesters for their long-term col-laboration and support, as well as fortheir very fruitful discussions.

References1. A.J. Sommer, in Handbook of

Vibrational Spectroscopy, Ed by J.M.Chalmers and P.R. Griffiths. JohnWiley and Sons, Chichester, pp.1369–1385 (2002).

2. E. Theocharous and J.R. Birch, inHandbook of Vibrational Spectroscopy,

Figure 3. (a) Chemical image of lipids (peak heightat 2920 cm–1) across a human hair section. The IRmap has been recorded with a 23 × 3 µm2 aperture.Using hierarchical cluster imaging, two types oflipids are recognised, and located either in the cuti-cle (b) or in the medulla (c). The amide I lineshapeacross the hair section is not homogeneous, anddownward (wavenumber) displacement of the maxi-mum is observed inside the cuticle. The image ofthe ββ sheet/αα helix ratio is displayed in (d) while thecorresponding Amide I bands are shown in (e).

Figure 4. (a) Visible image of a single HL60 cell, ofabout 18 µm diameter. (b) Chemical image of thelipids based on peak height at 2920 cm–1, across thesingle cell shown in (a) and before differentiationinduction. (c) Amide I and II frequency range for cellbefore and after 48 hours of differentiation induc-tion, recorded with an aperture of 3 × 3 µm2. Onecan note the narrower lineshape of the Amide I bandat 1655 cm–1 after differentiation induction. (d) Anhierarchical cluster analysis, followed by fuzzy C-means clustering on a cell after differentiationreveals two slightly different protein secondarystructure, inside the nucleus and in the cytoplasm.Average spectra of these two clusters are also dis-played. All IR spectra have been recorded with a3 × 3 µm2 aperture.

Ed by J.M. Chalmers and P.R.Griffiths. John Wiley and Sons,Chichester, pp. 349–367 (2002).

3. W.D. Duncan and G.P. Williams,Appl. Opt. 22, 2914 (1983).

4. R.A. Bosch, Nucl. Instrum. Meth.Phys. Res. A 386, 525 (1997).

5. M.C. Martin, N.M. Tsvetkova,J.H. Crowe and W.R. McKinney,Appl. Spectrosc. 55, 111–113 (2001).

6. G.L. Carr, Rev. Sci. Instr. 72, 1–7(2001).

7. G. Ellis, C. Marco and M.A.Gomez, J. Macromol. Sci. - Phys., inpress.

8. P. Dumas and L.M. Miller,Vib.Spectrosc., in press.

9. N. Jamin, P. Dumas, J. Moncuit,W.-H. Fridman, J.L. Teillaud, G.L.Carr and G.P. Williams, Proc. Natl.Acad. Sci. USA 95, 4837–4840(1998).

10. L. Miller, P. Dumas, N. Jamin, J.L.Teillaud, J. Miklossi and L. Forro,Rev. Sci. Instrum. 73, 1357–1360(2002).

11. L. Bozec, A. Hammiche, M.J.Tobin, J.M. Chalmers, N.J. Everalland H.M. Pollock, Meas. Sci.Technol. 13, 1217–1222 (2002).

Spectroscopy Europe 15/6 (2003) 23

INFRARED SPECTROSCOPY

Figure 5. (a) PhotoThermal IR Micro-Spectroscopy. A microthermalprobe, based on an atomic force microscope (AFM) cantilever tip isplaced under the infrared microscope objective at the focus of the syn-chrotron light. Though the infrared light covers an area defined by thediffraction limit, the probe can potentially sense thermal absorptionsfrom much smaller regions. By modulating the light, the output fromthe sensor can be used to generate an absorbance spectrum. (b) Thescanning probe instrument is coupled to the infrared beamline at theSRS, Daresbury, using a side-port accessory.