Pre-Event Dam Failure Analyses for Emergency Management

Colloids and Surfaces

A: Physicochem. Eng. Aspects 171 (2000) 225–239

Application of 2D IR correlation analysis to phasetransitions in Langmuir monolayer films

Douglas L. Elmore, Richard A. Dluhy *Department of Chemistry, Uni6ersity of Georgia, Athens, GA 30602-2556, USA

Abstract

The surface pressure-dependent conformational state of a monolayer film of 1,2-dipalmitoyl-sn-glycero-3-phospho-choline (DPPC) at the air–water (A/W) interface was studied using infrared external-reflection spectroscopy andtwo-dimensional infrared (2D IR) correlation analysis. When the IR spectra of the DPPC monolayer was collectedusing polarized IR radiation, a band splitting was observed in both the antisymmetric (na) and symmetric (ns)methylene CH2 stretching modes that was not observed with unpolarized radiation. This band splitting wasinterpreted as being due to the presence of co-existing ordered and disordered conformational states, however,definitive identification of conformational sub-bands is problematic due to the low signal-to-noise inherent in thepolarized IR spectra. To further investigate the spectral changes observed in the C�H region, 2D IR correlationanalysis was applied to a set of pressure-dependent unpolarized IR spectra of the DPPC monolayer. When theseunpolarized spectra were analyzed using 2D IR methods, the 2D asynchronous correlation spectrum of the DPPCmonolayer clearly showed that cross peaks attributable to the na and ns CH2 bands both split into two components,in agreement with the polarized IR monolayer spectra. Since band splitting in 2D IR spectra may be due to severalcauses, computer simulations were undertaken to help elucidate the exact cause of the observed splitting in the DPPC2D asynchronous spectrum. Synthetic monolayer IR spectra were calculated for two limiting cases. The first was a‘frequency shifting’ model in which a single band underwent a simple frequency shift. The second limiting case wasan ‘overlapped peaks’ model in which an overall vibrational band was calculated as the sum of two individualsub-bands whose frequencies remained constant, but whose relative intensities changed through the simulatedmonolayer transition. The results of the computer simulations indicated that a simple frequency shift could bedistinguished in the 2D asynchronous spectrum by the presence of a quartet of cross peaks, two with positivecorrelation intensities, and two with negative. In addition, a curved elongation of these cross peaks along the diagonalwas associated with this frequency shift. In contrast, the 2D asynchronous spectrum for two overlapped peaksresulted in a correlation intensity cross peak doublet, one positive and one negative with no elongation along thediagonal. The experimentally measured 2D IR asynchronous correlation spectrum for the DPPC monolayer closelyresembled the computer-simulated spectra for the ‘overlapped peaks’ model. Therefore, the origin of the bandsplitting in the na and ns CH2 bands in the 2D asynchronous spectrum is due to overlapping sub-bands that representthe ordered and disordered conformational states of the monolayer. Furthermore, these results also support the

www.elsevier.nl/locate/colsurfa

* Corresponding author. Tel.: +1-706-5421950; fax: +1-706-5429454.E-mail address: [email protected] (R.A. Dluhy)

0927-7757/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.

PII: S0927 -7757 (99 )00542 -7

D.L. Elmore, R.A. Dluhy / Colloids and Surfaces A: Physicochem. Eng. Aspects 171 (2000) 225–239226

interpretation that the sub-bands observed in the polarized monolayer IR spectra are correlated with ordered anddisordered monolayer states. © 2000 Elsevier Science B.V. All rights reserved.

Keywords: 1,2-Dipalmitoyl-sn-glycero-3-phosphocholine; Air–water interface; Two-dimensional infrared correlation analysis

1. Introduction

1.1. IR spectroscopy at the air–water interface

In 1985 Dluhy and Cornell published the firstpaper which reported that IR reflection spec-troscopy could be used to acquire spectra ofmonomolecular films on aqueous substrates [1].Since this initial proof of applicability tomonomolecular films at the air–water (A/W) in-terface, infrared external reflection–absorptionspectroscopy has been utilized to study manydifferent types of insoluble monolayers, and thereis a growing body of literature in this field (see,[2,3]). Although the major application of thismethod has been in the conformational analysisand phase transitions of model monolayers, therealso have been several groups that have appliedIR spectroscopy to study monolayer headgroupinteractions as well as polymer and peptide mono-layers [4–8].

Previous experimental work has used both un-polarized and plane polarized (parallel, Rp, andperpendicular, Rs) IR radiation to studymonomolecular films at the A/W interface. Al-though polarized monolayer IR spectra are valu-able in as much as the angle-dependent polarizedreflectivities can be used in the calculation ofmonolayer molecular orientation [9,10], both the-oretical and experimental studies have shown thatthe use of polarized IR radiation at the A/Winterface significantly degrades the final signal-to-noise ratio and is highly dependent on the angleof incidence employed [11–13].

Recent experiments in our laboratory have fo-cused on the use of polarized IR external reflec-tion–absorption spectroscopy to studybiophysical monolayers [14]. In experiments withwell-defined single component monolayers, we ob-served a band splitting in the methylene stretchingvibrations in the C�H stretching region of the

spectrum that has not been previously describedfor IR monolayer spectra. The polarized mono-layer spectra were able to distinguish individual,overlapping sub-bands within the methylene C�Hvibrations that were correlated to ordered anddisordered conformational states. Using the inte-grated intensities of these sub-bands, we havebeen able to semi-quantitatively track the forma-tion of ordered and disordered conformationalstates of a monolayer film throughout its phasetransition [14] (Faucher and Dluhy, manuscriptsubmitted).

Although these experiments demonstrate thepower of polarized IR reflection spectroscopy forstudying Langmuir monolayers, we are ultimatelylimited in our ability to make quantitative analy-ses of monolayer properties by the relatively poorsignal-to-noise ratio in the polarized spectra. Inorder to enhance our ability to interpret IRmonolayer spectra, we recently applied two-di-mensional infrared correlation spectroscopy (2DIR) correlation methods to study surface-pressureinduced dynamical changes in monolayers at theA/W interface. In particular, we are interested inwhether 2D IR methods can be used to confirmand further interpret the band splitting and spec-tral changes that we observe in the C�H region ofpolarized IR monolayer spectra. The results ofthese experiments are the subject of this paper.

1.2. 2D IR correlation analysis: background

2D IR is a recently-developed analyticalmethod that aids in the interpretation of complexspectra, especially spectra with the kind of broad,multiply overlapped peaks commonly encoun-tered in condensed phase vibrational spec-troscopy. As such, 2D IR may be considered asanother tool for spectral resolution enhancement,much as curve fitting, spectral derivatives or de-convolution are mathematical techniques used to

D.L. Elmore, R.A. Dluhy / Colloids and Surfaces A: Physicochem. Eng. Aspects 171 (2000) 225–239 227

identify the number and position of underlyingbands in a complex bandshape [15,16]. In the caseof 2D IR, the resolution of overlapping spectralfeatures is enhanced by subjecting the sample toan environmental perturbation. The nature of thisperturbation is arbitrary and can be any physicalor chemical modification to the sample, the onlyrequirement being that this modification must im-part some kind of measurable change in the re-sulting spectrum. In 2D IR, the variations in theIR spectra obtained as a result of the sampleperturbation are then mathematically cross-corre-lated to produce a 2D correlation map. The re-sulting 2D IR spectra are able to identify thosevibrational modes which selectively respond (ei-ther in-phase or out-of-phase) to the externalperturbation. Spectral resolution enhancement re-sults when the sample perturbation affects thevibrational modes occurring at different wave-lengths in a different manner.

The mathematical formalism for 2D IR wasfirst introduced by Noda in 1986 [17]. In thisoriginal formalism, the correlation analysis waslimited to macromolecular systems in which theexternal perturbation to the sample was restrictedto having a simple time-dependent sinusoidalwaveform [18]. In 1993, however, Noda intro-duced a generalized method for obtaining 2Dcorrelation spectra using external perturbationshaving any arbitrary waveform that may be afunction of time or other physical variable [19].The mathematical formalism for this generalizedmethod was somewhat more complicated than theoriginal method, in that it required the complexFourier transformation of dynamic spectra. Re-cently, a modification of the generalized methodwas introduced which uses the discrete Hilberttransform in place of the complex Fourier trans-form [20]. The new Hilbert transform approachsubstantially simplifies the computationaldifficulties associated with the generalizedmethod.

The 2D IR correlation spectra are characterizedby two independent wavenumber axes (n1, n2) anda correlation intensity axis. In general, two typesof spectra are obtained, commonly referred to asthe 2D synchronous spectrum and the 2D asyn-chronous spectrum. Vibrational modes that are

significantly coupled, or whose transition dipolemoments change in-phase at similar rates in re-sponse to the external sample perturbation (i.e.modes that are synchronized) appear in the 2Dsynchronous spectrum. Conversely, bands that aresignificantly decoupled, or whose transition dipolemoments respond out-of-phase at different ratesto the external sample perturbation (i.e. modesthat are asynchronized) appear in the 2D asyn-chronous spectrum. The correlation intensity inthe 2D synchronous and asynchronous mapsreflects the relative degree of in-phase or out-of-phase response, respectively.

The 2D synchronous spectra are symmetricwith respect to the diagonal line in the correlationmap. Intensity maxima appearing along the diag-onal are called autopeaks (corresponding to theautocorrelation of perturbation-induced molecu-lar vibrations), and are always positive. Intensitymaxima located at off-diagonal positions arecalled cross peaks (corresponding to the cross-cor-relation of perturbation-induced molecular vibra-tions at two different wavenumbers). A pair ofcross peaks may be positive or negative.

The 2D asynchronous spectra are antisymmet-ric with respect to the diagonal line in the correla-tion map. Only cross peaks located atoff-diagonal positions appear in asynchronousspectra; pair of cross peaks consist of two inten-sity maxima/minima, one of which is necessarilypositive and the other necessarily negative. Asyn-chronous cross peaks represent a lack of strongchemical interactivity, since their presence reflectsthe mutually independent nature of the reorienta-tion of the dipole moments of the molecule’sfunctional groups in response to an externalperturbation.

A major advantage of 2D IR correlation spec-troscopy is the possibility for enhanced resolutionobserved in the asynchronous spectra. If two IRdipole transition-moments change orientations in-dependently of each other at different rates, over-lapped bands appear as two cross peaks in theasynchronous spectrum. This is true even forheavily overlapped bands for which the corre-sponding cross peaks may appear close to thediagonal [21].


The temporal, or phase, relationship, betweentwo cross-correlated IR bands changing intensityunder a time-dependent perturbation determinesthe sign of the crosspeaks. Two positivecrosspeaks (intensity maxima) are observed in the2D synchronous spectrum when two transition-moments change orientations identically, and in-phase. Two negative crosspeaks (intensityminima) are observed in the 2D synchronousspectrum when two transition moments changeorientations identically, but anti-phase. A positiveand a negative cross peak are observed in the 2Dasynchronous spectrum when two transitiondipole moments change orientations out-of-phase;specifically, a maximum and a minimum occurwhen the orientations are 90° out-of-phase.

The interpretation of 2D IR correlation mapscan be difficult. The very reason that IR spec-troscopy is valuable in molecular analysis, i.e. thehighly sensitive nature of vibrational spectra tolocal environment, means that the 2D syn-chronous and asynchronous spectra can be com-plex, even for simple systems. Several articles haveappeared that explore the current state-of-the-artfor 2D IR spectral interpretation. The effect ofcommonly encountered changes in IR bandparameters, such as frequency, bandwidth, inten-sity changes and errors in band position havebeen described by Gericke et al. [22]. Other com-mon complications encountered in spectral analy-sis, including the effect of noise and baselinefluctuations on 2D spectra, were investigated byCzarnecki [23]. However, neither of these studiesspecifically addresses the assumptions and compli-cations encountered in the use of 2D IR correla-tion analysis for external reflection–absorptionspectroscopy of Langmuir monolayers. Usingcomputer simulations, we address these questionsin this article, and show how this analysis can beused to confirm our polarized monolayer data.

2. Materials and methods

2.1. Surface chemistry

The phospholipid used in these experiments was1,2 - dipalmitoyl - sn - glycero - 3 - phosphocholine

(DPPC); this material was obtained from AvantiPolar Lipids (Alabaster, AL) at 99+% statedpurity and used without further purification. Sam-ple solutions of DPPC were prepared by dissolv-ing the lipid in a chloroform solution (Baker,HPLC grade) at a concentration of approximately1.0 mg ml−1. The lipid sample was spread onto aNima 601M Series Langmuir film balance (Nima,Coventry, UK) and allowed to settle for at least10–15 min before compression began. Subphasewater was obtained from a Barnstead (Dubuque,IA) ROpure/Nanopure reverse osmosis/deioniza-tion system having a nominal resistivity of 18.3MV cm and a pH of approximately 5.6.

2.2. Fourier transform infrared externalreflectance measurements

Infrared external reflection–absorbance spectraof monolayers at the A/W interface were collectedusing a Perkin-Elmer Spectrum 2000 FT IR spec-trometer (Perkin-Elmer, Norwalk, CT). The opti-cal interface of the IR spectrometer to theLangmuir film balance was designed in our labo-ratory. The IR beam coming from the spectrome-ter’s external beam port reflected off a 60°,gold-coated, off-axis parabolic mirror (JanosTechnology Inc., Townshend, VT), and throughan infrared bandpass filter (OCLI, Santa Rosa,CA) onto the surface of a Nima 601M film bal-ance (Coventry, UK) at an incidence angle of 30°to the surface normal. The IR beam reflected offof the subphase, sampling the film, and was recol-limated by a second parabolic mirror that directedit onto the focusing mirror of a liquid N2-coolednarrow band HgCdTe detector. The IR bandpassfilter (3300–2000 cm−1) was placed just above thewater to reduce any localized heating effect [24].For polarized spectra, an Al wire-grid polarizeron KRS-5 (Model IGP225, Molectron DetectorInc., Portland, OR) was used. The film balance,optical components, and detector are housed in asealed, Plexiglas chamber that allows humiditycontrol of the local trough environment thus im-proving water vapor background subtraction.

A single beam spectrum of a pure water sub-phase with the correct polarization characteristicswas used as the background spectrum. The sub-


Fig. 1. Infrared external reflection–absorption spectra of DPPC monolayer films at the A/W interface. Reflection spectra werecollected as a function of monolayer surface pressure and were acquired with (A) unpolarized; and (B) parallel (Rp) polarizedradiation.

phase temperature was held constant at 2091°Cby flowing thermostatted water through the hol-low body of the trough. The temperature in theenclosed chamber was 25°C with a relative humid-ity of 70%. The lipid sample in CHCl3 solutionswas spread via syringe onto the trough surfacewhere it was allowed to equilibrate for a period of15 min before data collection began. The mono-layer film was compressed intermittently and spec-tra collected over a range of surface pressuresfrom 0.7 to a maximum of 50.0 mN m−1.

External reflection–absorption spectra werecollected with 1024 scans at 4 cm−1 resolution,apodized with a Norton-Beer (medium) function,and was Fourier transformed with one level ofzero filling. All monolayer IR spectra are pre-sented as reflection–absorption spectra, i.e. A−log(R/R0) where R is the IR reflectivity of themonolayer-covered surface, and R0 is the IRreflectivity of the bare water subphase back-ground. The reflectance IR spectra used in theanalyses presented here were baseline corrected


using the GRAMS/32 (Galactic Industries, Salem,NH) program, but have not otherwise beensmoothed or further processed.

Synthetic spectra were calculated using an Ar-ray Basic program written in our laboratory forthe GRAMS/32 environment (Dluhy, unpub-lished). All synthetic spectra were calculated as50% Lorentzian and 50% Gaussian peak shapeswith full widths at half maximum equal to 20

cm−1. No noise was added to the syntheticspectra.

2.3. Calculation of 2D IR correlation spectra

The 2D IR synchronous and asynchronousspectra were calculated using the Array Basicprogram ‘KG2D’ written for Grams/32 and gen-erously provided by Professor Yukihiro Ozaki ofKwansei-Gakuin University, Japan. This programuses Noda’s most recent mathematical formalismthat replaces Fourier transforms with Hilberttransforms [20]. The 2D IR correlation maps werecalculated from the unpolarized surface pressure-resolved IR monolayer spectra shown in Fig. 1A.The average spectrum was subtracted from eachspectrum in the set prior to the cross correlationanalysis. The synchronous spectrum, F(n1, n2),and the asynchronous spectrum, C(n1, n2), werethen calculated using Eqs. (1) and (2).

F(61, 62)=1

n−1%n

j=1

y(61, Pj)y(62, Pj) (1)

C(61, 62)=1

n−1%n

j=1

y(61, Pj) %n

k=1

Mjky(62, Pk)

(2)

In Eqs. (1) and (2), n1 and n2 represent twoindependent frequencies, n represents the numberof spectra used in the calculation, and Mjk is theHilbert transform matrix, which is defined in Eq.(3).

Mjk=!0 if j=k

1/p(k− j) otherwise(3)

To reduce the effect of noise in the experimen-tal 2D IR spectra, correlation intensities less than1% of the full scale were cut, as previously de-scribed [25].

3. Results and discussion

3.1. 2D IR spectroscopy at the air–waterinterface: experiments

Fig. 1A presents the unpolarized external reflec-tance IR spectra of a monomolecular film of

Fig. 2. Experimentally measured 2D IR correlation spectra forthe DPPC monolayer film at the A/W interface in the region3000–2800 cm−1. Eleven individual monolayer IR spectracollected between surface pressures of 0.7 and 50.0 mN m−1

were used to calculate the 2D spectra in this figure. Solid linesindicate regions of positive correlation intensity, while dashedlines indicate regions of negative correlation intensity. (A)Synchronous 2D correlation spectrum; (B) asynchronous 2Dcorrelation spectrum.


Fig. 3. A 3D wire mesh surface plot representation of the 2D asynchronous correlation spectrum for the DPPC monolayer film.Data used for this plot was the same as used to calculate the 2D asynchronous correlation map in Fig. 2B. (A) Surface plotrepresentation of entire C�H region between 3000 and 2800 cm−1; (B) surface plot representation of the spectral region between2980 and 2880 cm−1 surrounding the antisymmetric na CH2 band at �2920 cm−1.

DPPC in the C�H stretching region between3000–2800 cm−1. To acquire these spectra, themonolayer film was applied to the surface of thefilm balance, allowed to equilibrate, and thencompressed step-wise over a range of surfacepressures from 0.7 to 50.0 mN m−1 with spectracollected during the interval between eachsuccessive compression. Fig. 1A presentsmonolayer IR spectra obtained with unpolarizedincident radiation. Clearly evident in the spectraare the antisymmetric (na) CH2 stretching band(�2920 cm−1), the symmetric (ns) CH2 stretchingband (�2850 cm−1) and the antisymmetric CH3

stretching band (�2960 cm−1). The symmetricmethyl vibration (�2860 cm−1) is barelyapparent as a slight shoulder on the symmetricCH2 band. The C�H vibrations grow in intensityas the lipid surface density increases withincreasing surface pressure and the lipid molecules

become more ordered. Also observed in thesespectra as the monolayer surface pressureincreases is the shift of the IR peak maximafrequencies to lower wavenumbers. Specifically,the na CH2 band maximum shifts from �2925 to2919 cm−1 while the ns CH2 band maximum shiftsfrom �2855 to 2851 cm−1. The shift of the CH2

peak frequency to lower wavenumbers has longbeen used to distinguish hydrocarbon order in avariety of alkane systems [26]. The magnitude ofthe wavenumber shifts for the CH2 modes seen inFig. 1 are consistent with previously reportedvalues for monolayer systems, and have been usedto provide a direct, although qualitative, measureof acyl chain order in monolayers at the A/Winterface [2,3].

Fig. 1B also presents the C�H region of externalreflection IR spectra of DPPC monolayers at theA/W interface, but unlike the spectra shown


Fig. 4. Computer-simulated synthetic infrared external reflection–absorption spectra of a monomolecular film at the A/W interfacein the region 3000–2800 cm−1. Synthetic spectra were generated using the ‘frequency shifting’ model, in which a single IRbandshape increases intensity and shifts frequency as a function of surface pressure. In these spectra, the frequency shift of thesynthetic na CH2 band at �2920 cm−1 was calculated from 2926 to 2918 cm−1 in 1.0 cm−1 increments, while the shift of thesynthetic ns CH2 band at �2950 cm−1 was calculated from 2856 to 2850 cm−1 in 0.75 cm−1 increments. Further details of theconditions used in generating these synthetic spectra are provided in the body of the manuscript.

in Fig. 1A, these spectra were acquired with paral-lel polarized (Rp) incoming radiation. Even takinginto account the lower signal-to-noise ratio of thepolarized spectra, a comparison of Fig. 1A and Bclearly show a subtle but reproducible splitting ofthe na CH2 vibration at 2920 cm−1. Two sub-bands are observed at low surface pressures thatare directly related to the populations of orderedand disordered conformational states [14](Faucher and Dluhy, manuscript submitted). Asseen in Fig. 1B, these sub-bands appear to mergeat intermediate pressures. At higher surface pres-sures, the overall band shape appears to shift tolower wavenumber, indicating a mostly orderedconformational state in the monolayer. Usingcurve-fitting procedures based on these polarizedspectra, the fractional conformational state of themonolayer as a function of surface pressure canbe determined [14].

While these experiments show that polarized IRspectra are capable of distinguishing monolayer

conformational states, the relatively poor signal-to-noise ratio in these polarized spectra limits ourability to use IR spectroscopy to more fully quan-tify monolayer properties. In order to increase ourability to interpret IR monolayer spectra, we haveapplied 2D IR correlation methods to study thesurface-pressure induced dynamical changes ap-parent in the IR spectra of DPPC monolayers atthe A/W interface. In particular, we studiedwhether 2D IR methods are able to confirm theband splitting observed in the C�H region ofpolarized IR monolayer spectra.

For all the 2D IR analyses described here, weused the unpolarized IR monolayer spectra fromFig. 1A. The results of these 2D calculations forthe DPPC monolayer IR spectra are shown inFigs. 2 and 3. Fig. 2A represents the synchronouscorrelation map obtained from the cross-correla-tion of the C�H stretching region (3000–2800cm−1) of the DPPC monolayer at differing sur-face pressures. Two auto peaks, six positive cross


peaks and two negative cross peaks are observedin the synchronous map shown in Fig. 2A.1 Of allthe peaks observed in this 2D spectrum, the threemain peaks at 2959, 2919 and 2851 cm−1 are

readily assigned to the na CH3, na CH2, and the ns

CH2 stretching modes, respectively. The CH2 andCH3 cross peaks all have positive correlation in-tensities, indicating an in-phase reorientation oftransition moments as the surface pressure in-creases for these modes. This agrees with theone-dimensional (1D) spectra in Fig. 1 where it isseen that these peaks increase in intensity as thesurface pressure increases. The largest autopeaksand cross peaks appear at 2919 and 2851 cm−1

since the na and ns CH2 stretching bands experi-ence the largest relative intensity increase in themonolayer spectra [27].

The 2D IR asynchronous correlation spectrumfor the C�H stretching region of the DPPC mono-layer is shown in Fig. 2B. The most dominantcross peaks appear at 2923, 2916, 2854, and 2849cm−1. As in the synchronous spectrum, thesepeaks can be assigned to the na and ns CH2

stretching modes. Similarly, two cross peaks at2960 and 2956 cm−1 are assigned to the na CH3

stretch. The peak at 2865 cm−1 may be attributedto the ns CH3 mode but with a large degree ofuncertainty because of the low correlation inten-sity. As well, the peaks at 2899 and 2890 cm−1

have very low correlation intensities and are ofuncertain origin. Since the cross peaks at 2923,2916, 2854, and 2849 cm−1, attributable to the na

and ns CH2 stretching modes, contain the highestcorrelation intensities, we will restrict our analysisto these peaks.

It is clear from the 2D IR asynchronous spec-trum presented in Fig. 2B that both the na and ns

CH2 bands split into two components, in agree-ment with the polarized IR monolayer spectra ofDPPC (Fig. 1B). These components are located at2923 and 2916 cm−1 for the na CH2 band and at2854 and 2849 cm−1 for the ns CH2 band. Theasynchronous map further shows that the bandsplitting results in one positive and one negativeband, as predicted by 2D IR theory [21]. Thisband splitting, as well as the positive and negativebands, is further illustrated in Fig. 3A, which is a3D wire frame representation of the asynchronousspectrum of the entire CH2 region shown in Fig.2B. A similar 3D wire frame plot of only the na

CH2 peak from the 2D IR asynchronous map ispresented in Fig. 3B, providing a close up view of

Fig. 5. Computer simulated 2D IR correlation spectra calcu-lated for the synthetic IR spectra shown in Fig. 4. Individualmonolayer IR spectra calculated according to the ‘frequencyshifting’ model were used to produce the 2D spectra in thisfigure. Solid lines indicate regions of positive correlation inten-sity, while dashed lines indicate regions of negative correlationintensity. (A) Synchronous 2D correlation spectrum; (B) asyn-chronous 2D correlation spectrum.

1 In all the 2D correlation maps presented in this article, asolid line indicates that the peak has a positive correlationintensity, while a dashed line indicates that the peak has anegative correlation intensity.


Fig. 6. A 3D wire mesh surface plot representation of the 2D asynchronous correlation map for the ‘frequency shifting’ computersimulated monolayer film spectra shown in Fig. 4. Data used for this plot was the same as used to calculate the 2D asynchronouscorrelation map in Fig. 5B. (A) Surface plot representation of entire C�H region between 3000 and 2800 cm−1; (B) surface plotrepresentation of the spectral region between 2980 and 2880 cm−1 surrounding the antisymmetric na CH2 band at �2920 cm−1.

the band splitting and the positive–negative dou-blet for this peak.

3.2. 2D IR spectroscopy at the air–waterinterface: computer simulations

It is known that the coexistence of two over-lapped sub-bands can result in the formation ofcross peaks near the diagonal in the asynchronousspectrum [21]. However, just a simple frequencyshift of one IR band can also result in the formationof cross peaks near the diagonal. For example,Gericke et al. have published synthetic 2D IRspectra for a peak undergoing a simple frequencyshift [22]. Czarnecki has also published synthetic2D spectra for a peak undergoing a frequency shiftas well as for a peak containing two sub-bands thatchange intensity in the same direction with slightlydifferent rates [23]. Both these studies show that a

band splitting observed in the asynchronous 2Dspectrum can be due solely to a frequency shiftwithout the presence of underlying sub-bands.

While these previous studies provide valuablehints in how to interpret 2D IR spectra, they donot directly address the situation that pertains inthe monolayer IR experiment. In particular, theeffect of the simultaneous changes in the charac-teristic frequencies, intensities, bandwidths andnumber of sub-bands found in monolayer IRspectra has not been adequately addressed. Tohelp determine the physical cause of the splittingin the 2D IR asynchronous spectrum of DPPCmonolayers, we have simulated 2D IR syn-chronous and asynchronous spectra for two pos-sible physical causes under the specificconditions pertaining to the monolayer experi-ment and compared the simulated 2D spectrawith the experimental 2D spectra.


Fig. 7. Computer-simulated synthetic infrared external reflection–absorption spectra of a monomolecular film at the A/W interfacein the region 3000–2800 cm−1. Synthetic spectra were generated using the ‘overlapped peaks’ model, in which two sub-bandscontribute to the overall bandshape for the na CH2 and ns CH2 bands at �2920 and �2850 cm−1, respectively. In these spectrathe overall na CH2 band was calculated as the sum of two individual sub-bands located at 2926 and 2918 cm−1. The overall ns CH2

band was calculated as the sum of two individual sub-bands located at 2856 and 2850 cm−1. Further details of the conditions usedin generating these synthetic spectra are provided in the body of the manuscript.

Our computer simulated monolayer spectra fallinto two limiting cases, one in which a single bandshifts frequency only, and one in which an IRband shape is composed of two sub-bands that donot change frequency, but only relative intensity.We refer to the simulated spectra for the case inwhich a single IR peak undergoes a simple fre-quency shift as the ‘frequency shifting’ model. Werefer to the simulated spectra for the case in whichone overall IR bandshape is composed two over-lapping sub-bands as the ‘overlapped peaks’model. In the ‘overlapped peaks’ model one sub-band decreases in intensity while the other in-creases in intensity as the simulated monolayersurface pressure increases. In our calculations wehave simulated the C�H stretching region from3000–2800 cm−1 as being composed of two ma-jor bands, corresponding to the na CH2 band at�2920 cm−1 and the ns CH2 band at �2850cm−1. To accurately simulate the experimentaldata, the following conditions were applied to the

synthetic monolayer spectra. (1) IR band intensi-ties increased as the surface area available to themonolayer decreases; (2) the intensity ratio of thena CH2 band to the ns CH2 band was assumed tobe 100:75; (3) spectral lineshapes used were 50%Gaussian and 50% Lorentzian in character; (4) 20cm−1 linewidths (full width at half height) wereused to simulate broadened condensed phasepeaks; (5) the initial band position and the finalband position in the ‘frequency shifting’ modelcorresponds to the positions of the two over-lapped sub-bands in the ‘overlapped peaks’model; and (6) spectral intensities increased pro-portionately to the decrease in surface area avail-able to the monolayer as the simulated experimentproceeded (i.e. a 5% decrease in surface arearesulted in a 5% increase in spectral intensity).

The simulated monolayer spectra used in thecalculations for the ‘frequency shifting’ model areshown in Fig. 4. In these spectra, the frequencyshift of the na CH2 band was calculated from 2926


to 2918 cm−1 in 1.0 cm−1 increments. Con-versely, the frequency shift of the ns CH2 bandwas calculated from 2856 to 2850 cm−1 in 0.75cm−1 increments.

The 2D IR synchronous spectrum calculatedfor the computer simulated spectra correspondingto the ‘frequency shifting’ model is shown in Fig.5A. All together, there are four autopeaks, fourpositive cross peaks, and eight negative cross

peaks2. The negative cross peaks are present be-cause certain frequencies change intensity in op-posite directions. Specifically, the correlationintensities decrease for the highest wavenumbersand increase for the lowest wavenumbers of bothpeaks as the bands shift to lower wavenumbers.The appearance of negative cross peaks is incontrast to the measured synchronous spectrumfor DPPC. This is due to the fact that the experi-mental IR monolayer intensities do not decreasein any portion of the methylene band region inthe measured IR spectra.

The 2D IR asynchronous map calculated forthe computer simulated spectra corresponding tothe ‘frequency shifting’ model is shown in Fig. 5B.The most obvious feature of this correlation mapis the presence of a quartet of cross peaks, twowith positive intensities and two with negativeintensities, in each quadrant of the spectrum. Inaddition, a curved elongation along the diagonalis associated with these asynchronous cross peaks.Similar features in simulated 2D asynchronousspectra have been previously described [22,23].These features are diagnostic of a single bandundergoing a frequency shift.

The 3D wire frame surface plots of the correla-tion intensities calculated from the 2D asyn-chronous spectrum of the ‘frequency shifting’simulated spectra are shown in Fig. 6. Fig. 6Aillustrates the entire simulated C�H region, whileFig. 6B presents a closer view of only the calcu-lated, frequency-shifted, na CH2 band. The char-acteristic elongated quartet pattern is evidentfrom this view, although the fourth small negativecorrelation peak is hidden behind the large posi-tive peak.

The simulated monolayer spectra used in thecalculations for the ‘overlapped peaks’ model areshown in Fig. 7. In these spectra the overallantisymmetric na CH2 bandshape was calculatedas the sum of two individual sub-bands located at2926 and 2918 cm−1. The peak at 2926 cm−1

Fig. 8. Computer simulated 2D IR correlation spectra calcu-lated for the synthetic IR spectra shown in Fig. 7. Individualmonolayer IR spectra calculated according to the ‘overlappedpeaks’ model were used to produce the 2D spectra in thisfigure. Solid lines indicate regions of positive correlation inten-sity, while dashed lines indicate regions of negative correlationintensity. (A) Synchronous 2D correlation spectrum; (B) asyn-chronous 2D correlation spectrum.

2 Two of the peaks described do not appear in the syn-chronous spectrum presented in Fig. 5A because of the num-ber of contour levels used. If a larger number of contour levelsare used the peaks are observed; however, more contour levelsfurther emphasise noise and baseline fluctuations, and thespectrum becomes more difficult to interpret.


Fig. 9. A 3D wire mesh surface plot representation of the 2D asynchronous correlation map for the ‘overlapped peaks’ computersimulated monolayer film spectra shown in Fig. 7. Data used for this plot was the same as used to calculate the 2D asynchronouscorrelation map in Fig. 8B. (A) Surface plot representation of entire C�H region between 3000 and 2800 cm−1; (B) surface plotrepresentation of the spectral region between 2980 and 2880 cm−1 surrounding the antisymmetric na CH2 band at �2920 cm−1.

(corresponding to the disordered sub-band) beganthe calculations at maximum intensity and de-creased in 10% increments to a minimum as thesimulated surface pressure increased. In contrast,the peak at 2918 cm−1 (corresponding to theordered sub-band) began the calculations at mini-mum intensity and increased in 10% increments toa maximum as the simulated surface pressureincreased. A similar situation existed in the calcu-lations for the symmetric ns CH2 bandshape. Thisband was also calculated as the sum of two indi-vidual sub-bands, now located at 2856 and 2850cm−1. As the simulated surface pressure in-creased, the peak at 2856 cm−1 (corresponding tothe disordered sub-band) began the calculations atmaximum intensity and decreased in 10% incre-ments to a minimum, while the peak at 2850cm−1 (corresponding to the ordered sub-band)began the calculations at minimum intensity and

increased in 10% increments to a maximum. Inthese calculations only the individual sub-bandintensities were varied; the frequencies of the sub-bands stayed constant.

The 2D synchronous and asynchronous corre-lation maps calculated for the computer-simulatedspectra corresponding to the ‘overlapped peaks’model is shown in Fig. 8. Fig. 8A presents the 2Dsynchronous spectrum for the ‘overlapped peaks’model. It is clear from a comparison of Fig. 8Awith Fig. 5A that the synchronous correlationmaps for the ‘frequency shifted’ and the ‘over-lapped peaks’ models are very similar. The resem-blance of the 2D synchronous spectra for the twomodels is related to the inherent assumptions ofthe synchronous correlation analysis. That is, theintensity of the 2D synchronous spectrum be-comes significant if the perturbation-induced spec-tral changes of the two IR wavenumber axes are


similar to each other. For the case of the twomodels under consideration here, the overall spec-tral bandshapes are very similar, regardless of thenumber of underlying sub-bands. Therefore, sincethe input spectra for the two models closely re-semble each other, the synchronous correlationintensities will also be comparable, as seen inFigs. 5A and 8A.

The 2D asynchronous spectrum for the ‘over-lapped peaks’ simulated model is shown in Fig.8B. In contrast to the similarity observed in thecase of the 2D synchronous spectra, it is clear thatthe ‘frequency shifted’ and ‘overlapped peaks’models give very different 2D asynchronous spec-tra (i.e. compare Fig. 8B and Fig. 5B). For the‘overlapped peaks’ model, a correlation intensitypeak doublet, consisting of one positive intensitypeak and one negative intensity peak, is observedin each quadrant of the 2D asynchronous spec-trum (Fig. 8B). This contrasts with the case of the‘frequency shifting’ simulation model, in which acorrelation intensity peak quartet is observed(Fig. 5B). In further contrast with the ‘frequencyshifting’ model, no curved elongation of the crosspeaks along the diagonal is observed.

The 3D wire frame surface plots of the correla-tion intensities calculated from the 2D asyn-chronous spectrum of the ‘overlapped peaks’simulated spectra are shown in Fig. 9. Fig. 9Aillustrates the entire simulated C�H region, whileFig. 9B presents a closer view of only the calcu-lated, overlapped, na CH2 bands. The characteris-tic asynchronous doublet pattern associated withtwo overlapping sub-bands is evident from thisview.

The computer simulations of 2D spectra for thetwo competing models (‘frequency shifting’, Figs.5 and 6; and ‘overlapped peaks’, Figs. 8 and 9)can now be used to interpret the measured, exper-imental 2D spectra for the DPPC monolayer(Figs. 2 and 3). In Fig. 2B, the 2D asynchronousspectrum for DPPC, a clear doublet is observed ineach quadrant of the correlation map, and nocurved elongation of the cross peaks along thediagonal is seen. This asynchronous correlationdoublet for DPPC is further viewed in the surfaceplots of Fig. 3. In comparison with the two modelsystems, it is apparent that the experimentally

measured asynchronous 2D IR spectrum forDPPC closely resembles the computer-simulatedspectra for the ‘overlapped peaks’ model. There-fore, the origin of the splitting in the na CH2 andns CH2 bands in the 2D asynchronous spectrum(Fig. 2B) is due to overlapping sub-bands thatrepresent the ordered and disordered conforma-tional states of the monolayer.

Although 2D IR correlation spectroscopy is arecent development, other groups have previouslyused this technique to investigate the C�H regionof macromolecules. Two studies, in particular, arerelevant to the current situation of co-existingordered and disordered conformational states in amonomolecular film. First, a 2D IR study oftemperature-induced changes in nylon 12 discov-ered a splitting in the C�H stretching modes in the2D asynchronous spectrum [28]. In this case, theauthors attributed the observed band splitting tothe presence of crystalline and amorphous formsof the nylon polymer, an analogous situation tothat encountered with co-existing ordered anddisordered regions in monomolecular films. In thesecond instance, a splitting of the antisymmetricmethylene and methyl stretching modes was re-ported in a 2D IR study of poly(b-hydroxybu-tyrate) using a sinusoidal strain perturbation anda dynamic IR linear dichroism method [27]. Inthis case, the splitting of both the na and ns CH2

vibrational modes was again seen in the 2D asyn-chronous correlation spectrum. The pattern of theCH2 splitting observed in the 2D asynchronousmap in this study matches exactly the patternobserved by us for the DPPC monolayer and the‘overlapped peaks’ model. As in the previousstudy, the splitting seen in the 2D spectrum wasattributed to the presence of crystalline solid andnon-crystalline disordered regions in poly(b-hydroxybutyrate).

The presence of cross peaks in a 2D asyn-chronous spectrum requires that the correlatedwavenumbers are responding out-of-phase witheach other. The existence of a splitting in theDPPC monolayer 2D spectrum strongly suggeststhat it is the result of two populations of lipid inthe monolayer that exist in slightly different envi-ronments, and which are affected differently bythe external sample perturbation, i.e. an increase


in surface pressure. This observation and interpre-tation is in agreement with the results and conclu-sions of previous polarized IR experimentsreported by our laboratory [14] (Faucher andDluhy, manuscript submitted).

Acknowledgements

We would like to thank Professor YukihiroOzaki and Dr Yan Wang of Kwansei-GakuinUniversity, Nishinomiya, Japan for providing uswith their Array Basic program used to calculatethe 2D IR spectra. This work was supported bythe US Public Health Service through NationalInstitutes of Health grant GM40117 (RAD).

References

[1] R.A. Dluhy, D.G. Cornell, J. Phys. Chem. 89 (1985)3195–3197.

[2] R.A. Dluhy, S.M. Stephens, S. Widayati, A.D. Williams,Spectrochim. Acta 51A (1995) 1413–1447.

[3] R. Mendelsohn, J.W. Brauner, A. Gericke, Annu. Rev.Phys. Chem. 46 (1995) 305–334.

[4] J.T. Buontempo, S.A. Rice, J. Chem. Phys. 99 (1993)7030–7037.

[5] D. Blaudez, T. Buffeteau, J.C. Cornut, B. Desbat, N.Escafre, M. Pezolet, J.M. Turlet, Appl. Spectrosc. 47(1993) 869–874.

[6] C.R. Flach, J.W. Brauner, R. Mendelsohn, Biophys. J. 65(1993) 1994–2001.

[7] Y. Ren, M.S. Shoichet, T.J. McCarthy, H.D. Stidham,S.L. Hsu, Macromolecules 28 (1995) 358–364.

[8] A. Gericke, C.R. Flach, R. Mendelsohn, Biophys. J. 73(1997) 492–499.

[9] T. Hasegawa, S. Takeda, A. Kawaguchi, J. Umemura,Langmuir 11 (1995) 1236–1243.

[10] C.R. Flach, A. Gericke, R. Mendelsohn, J. Phys. Chem.B 101 (1997) 58–65.

[11] R.A. Dluhy, J. Phys. Chem. 90 (1986) 1373–1379.[12] J.A. Mielczarski, J. Phys. Chem. 97 (1993) 2649–2663.[13] D. Blaudez, T. Buffeteau, B. Desbat, P. Fournier, A.M.

Ritcey, M. Pezolet, J. Phys. Chem. B 102 (1998) 99–105.[14] R.A. Dluhy, Z. Ping, K. Faucher, J.M. Brockman, Thin

Solid Films 329 (1998) 308–314.[15] J.K. Kauppinen, D.J. Moffatt, H.H. Mantsch, D.G.

Cameron, Appl. Spectrosc. 35 (1981) 271–276.[16] D.G. Cameron, D.J. Moffatt, J. Test. Eval. 12 (1984)

78–85.[17] I. Noda, Bull. Am. Phys. Soc. 31 (1986) 520–524.[18] I. Noda, J. Am. Chem. Soc. 111 (1989) 8116–8118.[19] I. Noda, Appl. Spectrosc. 47 (1993) 1329–1336.[20] I. Noda, Abstract of Papers in the Second International

Syposium on Advanced Infared Spectroscopy, Durham,NC, 1996, paper A-16.

[21] I. Noda, Appl. Spectrosc. 44 (1990) 550–554.[22] A. Gericke, J.G. Sergio, J.W. Brauner, R. Mendelsohn,

Biospectroscopy 2 (1996) 341–351.[23] M.A. Czarnecki, Appl. Spectrosc. 52 (1998) 1583–1590.[24] H. Sakai, J. Umemura, Langmuir 13 (1997) 502–505.[25] M.A. Czarnecki, H. Maeda, Y. Ozaki, M. Suzuki, M.

Iwahashi, Appl. Spectrosc. 52 (1998) 994–1000.[26] R.G. Snyder, S.L. Hsu, S. Krimm, Spectrochim. Acta

34A (1978) 395–406.[27] C. Marcott, I. Noda, A.E. Dowrey, Anal. Chim. Acta 250

(1991) 131–143.[28] M.A. Czarnecki, P. Wu, H.W.S., Chem. Phys. Lett. 283

(1998) 326–332.

.

Pre-Event Dam Failure Analyses for Emergency Management

Documents