Tsd

Ha

b

a

ARRAA

KTsIE(OS

1

iuOcnddkqifpredpt

0d

Vibrational Spectroscopy 60 (2012) 180–184

Contents lists available at SciVerse ScienceDirect

Vibrational Spectroscopy

journa l homepage: www.e lsev ier .com/ locate /v ibspec

wo-dimensional infrared (2D IR) correlation spectroscopy study ofelf-assembly of oleic acid (OA) in conjunction with partial attenuation ofominant factor by eigenvalue manipulating transformation (EMT)

ideyuki Shinzawaa,∗, Isao Nodab

National Institute of Advanced Industrial Science and Technology (AIST), Nagoya 463-8560, JapanThe Procter & Gamble Company, 8611 Beckett Road, West Chester, OH 45069, USA

r t i c l e i n f o

rticle history:eceived 13 September 2011eceived in revised form 1 December 2011ccepted 12 January 2012vailable online 21 January 2012

eywords:

a b s t r a c t

Highly selective two-dimensional (2D) correlation analysis, achieved by attenuating a major principalcomponent in spectral data with eigenvector manipulating transformation (EMT) technique, was demon-strated. A spontaneous evaporation process of a binary mixture solution of oleic acid (OA) and carbontetrachloride (CCl4) was monitored by attenuated total reflection infrared (ATR-IR) spectroscopy. Finedetails of the dynamic behavior of OA molecules undergoing self-assembly into smectic liquid crystalswere analyzed by 2D correlation spectroscopy scheme in conjunction with EMT technique. Correlationfeature derived from the original spectral data is dominated with the contribution from the major dimer

wo-dimensional (2D) correlationpectroscopynfraredigenvector manipulating transformationEMT)leic acid

components of OA. On the other hand, 2D correlation spectra generated form the reconstructed databy attenuating the first principal component (PC1) showed selective enhancement of the correlationalfeature associated with the minor monomer component, which makes it possible to identify distinctpopulations, each having different dynamic behavior during the self-assembly.

© 2012 Elsevier B.V. All rights reserved.

elf-assembly

. Introduction

Main feature of two-dimensional (2D) correlation spectrums often dominated with the intensity variation arising fromnwanted signals arising from interfering constituents in system.ne of the important benefits derived from chemometric analysisombined with 2D correlation analysis is the ability to ratio-ally reject such irrelevant information presented in the spectralata [1]. For example, 2D correlation analysis in conjunction withata reconstruction based on principal component analysis (PCA),nown as PCA-2D, was proposed by Jung et al. to improve theuality of the analysis [2]. The basic hypothesis of PCA is that the

mproved proxy of the original data matrix can be reconstructedrom only a limited number of significant factors, namely princi-al components (PCs). 2D correlation analysis based on the dataeconstructed by rejecting unnecessary PCs makes it possible toffectively elucidate the most important features presented in theata without being hampered by noise or insignificant minor com-

onents. Thus, PCA-2D may be viewed as a form of denoisingechnique for the pretreatment of data.

∗ Corresponding author.E-mail address: h-shinzawa@aist.go.jp (H. Shinzawa).

924-2031/$ – see front matter © 2012 Elsevier B.V. All rights reserved.oi:10.1016/j.vibspec.2012.01.007

Recently, another type of PCA-based data transformation for 2Dcorrelation analysis, called eigenvector manipulating transforma-tion (EMT), was proposed by the same group [3–5]. EMT producesa new reconstructed data by systematic substitution of individ-ual factors. By partially attenuating specific PCs with EMT scheme,it becomes possible to suppress the unwanted contribution fromselected components in the system. A characteristic advantage ofEMT lies in the fact that rejection of the contribution from noise orinsignificant minor components as well as selective enhancementof certain correlational features becomes possible by systematicallymanipulating the weight on PCs.

In this study, the effect of the attenuating major PC was demon-strated with time-dependent attenuated total reflection infrared(ATR-IR) spectra of a binary mixture solution of oleic acid (OA) andcarbon tetrachloride (CCl4). The understanding of the nature con-cerning self-assembling of OA is of great scientific interest, fromfundamental research and practical technological points of view[6–8]. Spectroscopic study of self-assembling of OA provides inter-esting opportunity to derive adequate molecular level insight intothe underlying mechanisms of the system.

Transient evaporation of binary mixture solution of OA and

CCl4 was monitored by ATR-IR spectroscopy. During the evap-oration of CCl4 from the system, OA molecules substantiallygo through several transitions. For example, it is believed thatthe isolated monomers of OA show spontaneous organization of

nal Spectroscopy 60 (2012) 180–184 181

mpbiicatstvao

2

mpd

A

stfidvo

mdg

A

atvieTrt

3

OmdmbeoimCr

H. Shinzawa, I. Noda / Vibratio

olecular units into a certain kind of ordered structure. Thisrocess includes the formation of dimer based on hydrogenonding of carboxyl group and aggregation of dimer units

nto tightly packed smectic liquid crystals. Such phenomenons closely related to the degree of hydrogen bonding of thearboxyl group and segmental movements of carbon atomslong the alkyl chain [9–13]. Consequently, the variation of theime-dependent ATR-IR spectra of the binary mixture solution sub-tantially reflects such transition of OA molecules. Thus, in turn,he detailed analysis of the dynamic behavior of the spectra pro-ides useful background information on how the OA moleculesggregate into the smectic liquid crystals during the evaporationf the CCl4.

. Background of EMT

Assume a spectral data matrix A of m by n dimension, whereis the number of spectral traces and n is the number of data

oints per spectrum. Reconstructed matrix A* by singular valueecomposition (SVD) is given as follows

∗ = USVt (1)

The superscript t indicates the transpose of matrix. The m by ro-called left singular matrix U contains the first r eigenvectors ofhe matrix AAt, and the n by r right singular matrix V contains therst r eigenvectors of AtA. The r by r matrix S is a relatively smalliagonal matrix. The diagonal elements of S are the first r singularalues of A, which are the positive square roots of the eigenvaluesf either AAt or AtA, arranged in the decreasing order.

The new EMT-reconstructed data matrix A** is obtained byanipulating and replacing eigenvalues of A*. The reconstructed

ata matrix A** based on general form of PC attenuating EMT isiven by

∗∗ = A∗ −∑

i

kisiUiVti (2)

This form indicates that the data matrix A** is reconstructed byttenuating the contribution from ith singular values. The attenua-ion parameter ki can be chosen individually for different PCs. Thealue of individual ki can be set to 0 (no attenuation), 1 (full elim-nation) and so on, depending on the specific strategy of EMT. Forxample, it is possible to suppress the effect of PC1 by setting i = 1.he reconstructed data are then analyzed with conventional 2D cor-elation spectroscopy scheme to elucidate the dynamic behavior ofhe components presented in the data [1,14,15].

. Experimental

Time-dependent ATR-IR spectra of a binary mixture solution ofA and CCl4, undergoing a spontaneous evaporation process wereeasured by a NEXS 870 FT-IR spectrometer equipped with a MCT

etector (Thermo Nicolet). The initial mole fraction of the OA in theixture solution was 0.02. The 20 �L sample solution was analyzed

y depositing it on a horizontal ZnSe ATR plate. The sample wasxposed to open atmosphere at room temperature (24 ◦C), and setsf IR spectra were collected at intervals of 4 s, with each set consist-

ng of eight coadded scans at a 4 cm−1 resolution. Once the solution

ixture was exposed to air, CCl4 started evaporating. Eventually,Cl4 was completely removed from the system and only oleic acidemained behind.

Fig. 1. Time-dependent ATR-IR spectra of binary mixture solution of OA and CCl4.

4. Results and discussion

4.1. 2D correlation analysis of original data matrix

Fig. 1 represents the time-dependent ATR-IR spectra of thebinary mixture solution. Peaks observed in this region are spe-cific to vibrational modes of carboxyl group of OA. For example,a minor peak observed at 1740 cm−1 is assignable to the monomerof OA [16]. A major peak observed at around 1710 cm−1 exhibitsgradual increase in the spectral intensity and shift in position from1714 to 1708 cm−1 [16]. Such variation of the spectral feature inthis region may be explained as the co-existence of the contribu-tions from the dimers with disordered orientation and the dimersforming quasismectic liquid crystals in which the dimers are tightlypacked together and have only short-range positional order. Iwa-hashi et al. reported that OA tends to form a specific self-assembledmodel, which provides most condensed packing form of the dimersof OA due to the segmental movements of carbon atoms alongthe alkyl chain [10]. The variation of the spectral intensity heresubstantially reflects the structural alternation of OA induced bythe change in the concentration. Thus, the detailed analysis of thechange in the spectral feature may, in turn, provide useful intima-tion on how OA molecules undergo the self-assembly. However, itis not straightforward to investigate these spectral changes fromthe conventional one-dimensional stack of the spectra. The appli-cation of 2D correlation analysis becomes useful to elucidate suchsubtle but pertinent information.

Fig. 2(A) represents synchronous correlation spectrum cal-culated from the set of IR spectra shown in Fig. 1. Thesynchronous correlation spectrum provides only one specific auto-peak at around 1710 cm−1, indicating the gradual increase in theconcentration of the OA during the evaporation. It is importantto point out that some minor synchronous correlation peaks, forexample an auto-peak and cross peaks arising from the monomer,do not show up in this synchronous correlation spectrum. Havinga band with a large magnitude of intensity variations creates theproblem in simultaneously displaying the fine features of bandsexhibiting only small amount of intensity variations. Consequently,the generation of such seemingly simple correlation pattern sug-gests that large portion of the variation of the spectral feature isdominated with the intensity change of the dimer peaks.

The corresponding asynchronous correlation spectrum is shownin Fig. 2(B). The generation of a negative cross peak between1708 and 1714 cm−1 reveals that the change in the dimer occurs

before the change in the dimer cluster. Thus, it is likely that for-mation of the dimer is followed by the aggregation of the dimerunits into a certain cluster structure. Such sequential order ofthe events agrees well with the development of smectic liquid

182 H. Shinzawa, I. Noda / Vibrational Spectroscopy 60 (2012) 180–184

Ft

c[tctbaoddtaia

4

cfmdoasaaosT

ig. 2. (A) Synchronous and (B) asynchronous correlation spectra calculated fromime-dependent ATR-IR spectra of binary mixture solution of OA and CCl4.

rystals via the packing of dimer units reported by Iwahashi et al.10]. On the other hand, one can see the development of correla-ion feature along the 1740 cm−1 coordinate of the asynchronousorrelation spectrum, suggesting the dissimilar behavior betweenhe monomer and dimer components. However, the demarcationetween the monomer and dimer bands is somewhat unclear in thesynchronous correlation spectrum. With the specific knowledgef the spectral coordinate of the monomer peak, one may expect toeduce the sequential order between the monomer and dimer (orimer cluster) from the positive asynchronous correlation intensi-ies at 1708 and 1714 cm−1 along the 1740 cm−1 coordinate of thesynchronous correlation spectrum. Yet the absence of the signif-cant peak center makes the interpretation somewhat ambiguous,nd it may run a risk of overinterpretation.

.2. 2D correlation analysis of reconstructed data matrix

So far, the main feature of the 2D synchronous and asynchronousorrelation spectrum is dominated with the major factor arisingrom the dimer components, making the identification of other

inor component difficult. Thus, it becomes useful to selectivelyiminish the contribution from the dominant peaks which obscurether small but significant spectral features. Fig. 3 shows (A) scorend (B) loading vectors of the first PC (PC1) derived from PCA of theet of IR spectra. The score value gradually decreases with the timend the entire feature of the corresponding loading vector results invery similar manner with the IR spectra of OA. Linear combination

f the score and loading vectors describes the major variation of thepectral intensity induced by the change in the concentration of OA.hus, the attenuation of the PC1 may in turn provide an opportunity

Fig. 3. (A) Score and (B) loading vectors of PC1.

to elucidate the subtle but important variation associated withminor components in the system.

Fig. 4 represents (A) synchronous and (B) asynchronous correla-tion spectra obtained from the reconstructed data, which is devoidof the dominant contribution by the PC1 by setting the tunableparameters in Eq. (2) as k1 = 1 and ki = 0 for i > 1. It is noted thatthe reconstructed data matrix resulted in the marked enhance-ment of selectivity in the 2D correlation spectrum. For example,one can find that the synchronous correlation pattern in Fig. 4(A) isnow quite different from that of the conventional synchronous cor-relation spectrum. Many correlation peaks, which are not readilynoticeable the conventional 2D correlation spectrum, become visi-ble in Fig. 4(A). Autocorrelation peaks are observable at 1708, 1714and 1740 cm−1, respectively. Cross correlation peaks are also obvi-ously identifiable in the synchronous correlation spectrum. Suchenhancement of the correlation feature is mostly due to the atten-uation of the PC1. Thus, it is demonstrated that the attenuation ofmajor PC is useful in simultaneously displaying the fine features ofbands exhibiting large and small amount of intensity variations.

It is noted that much more detailed features are also visiblein the asynchronous correlation spectrum shown in Fig. 4(B). Thedemarcation between overlapped asynchronous correlation peakbecame much sharper compared to the original asynchronous cor-relation spectrum. Thus, cluster of peaks connected by ridges arefragmented into distinct peaks separated by clear boundaries. Forexample, one can find the development of cross peaks at 1708 and1714 cm−1 along the 1740 cm−1 coordinate of the asynchronouscorrelation spectrum. Highly overlapped correlation features arenow well resolved into the monomer, dimer and cluster compo-nents, respectively, suggesting dissimilar behavior among these

components. Consequently, it reveals that there are three distinctpopulations, each having different absorption band and uniquevariation rate during the evaporation experiment.

H. Shinzawa, I. Noda / Vibrational Spectroscopy 60 (2012) 180–184 183

FP

pctaowismAtkcdoccsifTdlif

4

e

Fig. 5. (A) Synchronous and (B) asynchronous correlation spectra calculated from

ig. 4. (A) Synchronous and (B) asynchronous correlation spectra calculated fromC1-attenuated data.

It is important to point out here that the simple correlationeak-sign rules used for identifying the sequence of events in theonventional 2D spectrum are no longer reliable, since the quanti-ative features, especially the sequential order information in thesynchronous spectrum, will be inevitably corrupted by the EMTperation. The practical utility of the EMT-enhanced 2D spectrumill be found in the area of identifying overlapped bands exhibit-

ng similar or dissimilar patterns of intensity variations. Even a verymall amount of variation will be readily detected since such infor-ation is often carried more heavily in the secondary PC factors.ccordingly, it becomes possible to go back to the interpreta-

ion of the conventional 2D correlation spectrum with the specificnowledge of spectral coordinates for correlation peaks in the 2Dorrelation spectrum generated from the PC1-free reconstructedata. For example, now it seems safe to determine the sequentialrder between the monomer and dimer from the positive asyn-hronous correlations at 1708 and 1714 cm−1 along the 1740 cm−1

oordinate of the asynchronous correlation. All put together, theimple linear sequence relationship can be derived. The first eventn the intensity variations occurs for the band due to the monomer,ollowed by the intensity changes in the band assigned to the dimer.he intensity variation arising from cluster formation based on theimers takes place next. Consequently, it indicates that the iso-

ated monomer substantially undergoes the self-assembly processnto the development of smectic liquid crystalline structure via theormation of the dimer.

.3. Comparison with other techniques

It is useful to compare the effect on the correlation featurenhancement by EMT with other techniques, e.g. normalization

normalized spectra.

[17,18] and Pareto scaling [19]. The use of normalization of rawspectral data is a popular method to reduce various interferingeffects that can later obscure the information extracted by 2D cor-relation analysis. For example, it can be effectively used to suppressthe overwhelmingly strong nonselective effect of the linear spec-tral response proportional to the concentration (i.e., Beer–LambertLaw), which may obscure much more subtle but interesting spectralresponses associated with specific molecular interactions [17,18].Fig. 5 shows (A) synchronous and (B) asynchronous correlationspectra calculated from the normalized spectra. One can find thatcorrelation features appearing in the synchronous correlation spec-trum becomes similar to that derived from the EMT reconstructedspectra. Many correlation peaks, which are not readily noticeable inthe conventional 2D correlation spectra, become visible in Fig. 5(A).The negative correlation between the monomer and dimer (ordimer cluster) reveals that changes in the monomer and dimerconcentrations occur in the opposite directions. Thus, the over-whelmingly strong spectral intensity change proportional to theconcentration can effectively be suppressed by the normalization.On the other hand, one can find that the asynchronous correlationspectrum is dominated by numerical artifacts mostly due to theunwanted effect of the noise amplification by the normalization.The development of such numerical artifacts obviously makes theidentification of meaningful correlation peak difficult.

Normalized spectra are shown in Fig. 6. It is clear that the effectof the noise amplification by normalization is especially acute forthe IR spectra measured at the early stage of the evaporation mea-

surement due to weak IR signals. In Fig. 6, one can find that theexaggeration of the noise becomes obvious when the concentra-tion of the OA is low. Consequently, the normalization of the spectra

184 H. Shinzawa, I. Noda / Vibrational Sp

we

ysscwg

sssta

Fs

Fig. 6. Normalized IR spectra.

ith little signals exaggerates the contribution of noise rather thannhancing minor but meaningful variation of the spectra.

Proper scaling of spectral datasets prior to 2D correlation anal-sis may accentuate certain useful features of the resulting 2Dpectra. For example, Pareto scaling, i.e., dividing of dataset by thequare root of its standard deviation, often tends to sharpen demar-ation between overlapped peaks. The concept of Pareto scalingas first introduced by Wold et al. [20] and latter expended to the

eneralized scaling form by Noda [19].Fig. 7 shows (A) synchronous and (B) asynchronous correlation

pectra calculated from the spectra subjected to the generalized

caling. The scaled IR spectra were obtained by two scaling con-tants set to ˛ = 0.5 (i.e. Pareto scaling) and ˇ = 1 [19]. The main fea-ure of the synchronous correlation is dominated by the autopeakt around 1710 cm−1, indicating the gradual increase in the

ig. 7. (A) Synchronous and (B) asynchronous correlation spectra calculated fromcaled spectra.

[

[

[[[[[

[[[[

ectroscopy 60 (2012) 180–184

concentration of OA during the evaporation. Development of crosspeaks is not identified in the synchronous correlation spectra. Onthe other hand, it is interesting to point out that, in Fig. 7(B),the asynchronous correlation intensity arising from the monomerbecomes greater by the scaling, reveling the correlation enhance-ment of the monomer component. However, demarcation betweenthe monomer and dimer bands is still unclear in the asynchronouscorrelation spectrum based on the scaling. In practice, an advan-tage of the generalized Pareto scaling lies in the fact that the Noda’srules to interpret sign relations remain applicable to determine thesequential order of events encoded within the set of the data, sincethe signs of the cross peaks do not change by the scaling. Thus,practice of the scaling technique will often be a reasonable startingpoint to explore the even more detailed correlation relationship insynchronous and asynchronous spectra.

Consequently, it is likely that the attenuation of the spec-tral intensity variation associated with the concentration becomesimportant to enhance the minor correlation features encoded in theIR spectral data. The above results reveal that the selective removalof the PC1 (i.e. concentration) by the EMT works well to enhancethe minor correlation feature in this case.

5. Conclusion

Dynamic behavior of self-assembly system of OA was studied by2D IR correlation spectroscopy in conjunction with EMT technique.A spontaneous evaporation process of a binary mixture solutionof OA and CCl4 was monitored by ATR-IR spectroscopy. The cor-relation features of the conventional 2D correlation spectra aredominated by the contribution from the major monomer compo-nents of OA. Consequently, sufficient information is not providedon the dynamic behavior of the minor monomer component. Incontrast, a newly reconstructed PC1-free data generated distinctcorrelation features of the minor monomer component by attenu-ating the contribution of the major dimer component contributions.This study effectively demonstrated that 2D IR correlation anal-ysis combined with EMT technique is a powerful tool to unraveldynamic behavior of minor component in a complex system. Suchpartial attenuation of PC may potentially become a very useful toolin the analysis of highly congested 2D correlation spectrum oftenencountered in practice.

References

[1] I. Noda, Y. Ozaki, Two-Dimensional Correlation Spectroscopy, John Willey &Sons, Chichester, West Sussex, 2004.

[2] Y.M. Jung, S.H. Shin, S.B. Kim, I. Noda, Appl. Spectrosc. 56 (2002) 1562.[3] Y.M. Jung, S.B. Kim, I. Noda, Appl. Spectrosc. 57 (2003) 564.[4] Y.M. Jung, S.B. Kim, I. Noda, Appl. Spectrosc. 57 (2003) 557.[5] Y.M. Jung, S.H. Shin, S.B. Kim, I. Noda, Appl. Spectrosc. 57 (2003) 850.[6] D.M. Small, The Physical Chemistry of Lipids: From Alkanes to Phospholipids,

Plenum Press, New York, 1986.[7] R.J. Hamilton, A. Bhati, Recent Advances in Chemistry and Technology of Fats

and Oils, Elsevier Applied Science Publishers, London, 1987.[8] R.B. Gennis, Biomenbranes: Molecular Structure and Function, Springer-Verlag,

New York, 1989.[9] M. Iwahashi, Y. Kasahara, H. Minami, H. Matsuzawa, M. Suzuki, Y. Ozaki, J. Oleo

Sci. 51 (2002) 157.10] M. Iwahashi, Y. Kasahara, H. Matsuzawa, K. Yagi, K. Nomura, H. Terauchi, Y.

Ozaki, M. Suzuki, J. Phys. Chem. B 104 (2000) 6186.11] F. Pi, H. Shinzawa, M.A. Czarnecki, M. Iwahashi, M. Suzuki, Y. Ozaki, J. Mol.

Struct. 974 (2010) 40.12] H. Shinzawa, M. Iwahashi, I. Noda, Y. Ozaki, J. Mol. Struct. 883–884 (2008) 27.13] H. Shinzawa, M. Iwahashi, I. Noda, Y. Ozaki, J. Mol. Struct. 883–884 (2008) 73.14] I. Noda, Appl. Spectrosc. 47 (1993) 1329.15] I. Noda, Appl. Spectrosc. 54 (2000) 994.16] M. Iwahashi, A. Umehara, K. Wakisaka, Y. Kasahara, H. Minami, H. Matsuzawa,

H. Shinzawa, Y. Ozaki, M. Suzuki, J. Phys. Chem. B 111 (2007) 740.

17] M.A. Czarnecki, Appl. Spectrosc. 57 (2003) 107.18] I. Noda, Y. Ozaki, Appl. Spectrosc. 57 (2003) 110.19] I. Noda, J. Mol. Struct. 883–884 (2008) 216.20] S. Wold, E. Johansson, M. Cocchi, in: H. Kubinyi (Ed.), 3D QSAR in Drug Design:

Theory, Methods and Applications, ESCOM Science Publishers, Leiden, 1993.