J-aggregate to J-aggregate relaxations in langmuir films of amphiphilic merocyanine dye derivatives studied by optimum difference spectrum method

Colloids and Surfaces A: Physicochem. Eng. Aspects 284–285 (2006) 212–216

J-aggregate to J-aggregate relaxations in langmuir films of amphiphilic merocyanine dye derivatives studied by optimum difference spectrum method Keiichi Ikegami ∗ Nanotechnology Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), AIST Tsukuba Central 2, 1-1-1 Umezono, Tsukuba, Ibaraki 305-8568, Japan Received 20 June 2005; received in revised form 10 October 2005; accepted 28 October 2005 Available online 20 December 2005

Abstract Langmuir (L) films of amphiphilic merocyanine dye derivatives (DS and DSe) mixed with arachidic acid have been prepared on a Mg2+ containing aqueous subphase. Time evolution of the molar absorption coefficient (ε) spectra observed for them has shown that J-aggregates with relatively large red shifts (ca. 2800 cm−1 ) are the majorities at the initial stage, but J-aggregates with smaller shifts (ca. 2400 cm−1 ) become major within few minutes of compression. Isosbestic points clearly seen in the time evolution of the spectra imply those films include no other appreciable components. A similar phenomenon has been observed for DS films on Ca2+ -containing subphase. The pure spectra of the initial and final J-aggregates have been extracted from the experimentally obtained ε spectra by employing the optimum difference spectrum (ODS) method. Semi-quantitative discussion based on the extracted pure spectra has suggested that the observed phenomena are structural relaxations, i.e., changes in the molecular arrangements in the J-aggregates, and that the interactions between the static dipoles born by the dye molecules are not the main driving force of the J-aggregation of DS and DSe. © 2005 Elsevier B.V. All rights reserved. Keywords: J-aggregate; Relaxation; Langmuir film; Spectrum analysis

1. Introduction J-aggregates [1] have attracted much attention and been studied intensively by many researchers because of their fascinating optical properties: a red-shifted and narrowed electronic absorption band (J-band), enhanced photoluminescence, and a very small Stokes shift. J-aggregate to J-aggregate relaxations are interesting phenomena from both the technological and scientific viewpoints, since they may be used for information storage, and may reflect the mechanism of J-aggregation. An amphiphilic merocyanine dye derivative, 3-carboxymethyl-5-[2-(3-octadecyl-2(3H)-benzothiazolylidene)ethylidene]-2-thioxo-4-thiazolidinone (DS) [2] and its derivative of DSe [3–5] (Fig. 1) form J-aggregates in their pure and mixed Langmuir (L) films [6] prepared upon appropriate aqueous subphases. These L films are suited for studying

∗

Tel.: +81 298 61 5545; fax: +81 298 61 5400. E-mail address: [email protected].

0927-7757/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfa.2005.10.095

the J-aggregates under variation of physical and chemical conditions such as the surface pressure and salts dissolved in the subphase. In fact, the dependence of the spectral features of the J-band upon the cation species in the subphase was intensively studied for L films of DS by Yoneyama et al. [7] and Miyata et al. [8] and for those of DSe by Kawaguchi and Iwata [3–5]. These authors found that the J-band spectrum depends on time in some cases, but were not interested in the mechanism of the observed spectral changes and then did not report detailed data. Recently, Kato et al. [9,10] have reported a thermal J-aggregate-to-J-aggregate transition in DS films upon a subphase containing both Cd2+ and Mg2+ . They are interested in the transition mechanism and performing extensive studies, but in their case both of the physical and chemical conditions, i.e., the temperature and the cation species associating with the L films, are varied and the mechanism may be complicated. This work aims at providing a new insight into the mechanism of the spectral changes taking place in L films of DS and DSe under constant chemical conditions. With this purpose, the spectral changes have been observed in detail and the recorded

K. Ikegami / Colloids and Surfaces A: Physicochem. Eng. Aspects 284–285 (2006) 212–216

Fig. 1. Chemical structures of amphiphilic merocyanine dye derivatives used in this work.

spectra have been analyzed carefully. The followings are the keys in this work that enable significant extension of the past studies in Refs. [3–5,7,8]. (i) Drawing molar absorption coefficient (ε) spectra instead of usual absorbance spectra. This can be done by simultaneous recording of the surface area and the absorbance spectrum of an L film under a finite surface pressure. In addition, the effect of side reactions could be neglected with restricting the period of time for observation to several minutes. (Then, the word “final” in this work is used to indicate quasistable states realized in the short periods of time. Long-time effects, such as reported in Ref. [8], cannot be discussed in this way.) (ii) Analyzing the obtained ε spectra by the optimum difference spectrum (ODS) method, which is a recently proposed [11–14] powerful tool for analyzing two-component spectra. It is based on the least-squares regression and gives the maximum likelihood estimators of the difference between the pure spectra of the components as ODS, even when the components’ quantities in the samples are unknown. Pure spectra of J-aggregates have been extracted through the analyses. (iii) Semi-quantitative discussion upon the relationship between the shift due to Jaggregation (J-shift) and the number of dye molecules in the aggregation (aggregation number).

213

fibers as a probe. The probing light emitted from the bundle propagates into the water subphase at normal incidence and was reflected by a mirror put on the bottom of the trough and then was collected by the bundle. The molar absorption coefficient, ε, of a dye in its Langmuir film was obtained through ε(λ) = −1/2 × 10−17 NA A log10 [I(λ)/I0 (λ)], where I and I0 are the intensities of the detected light with and without the film, respectively, at the wavelength of λ. NA is the Avogadro number and A [nm2 ] is the area per dye molecule. Note that A is not the so-called “limited area,” but the actual area observed at the same time of the optical measurement. 3. Results and discussion 3.1. Changes in J-bands Figs. 2 and 3 show time evolution of the molar absorption coefficient (ε) spectra of DS–AA and DSe–AA mixed L films, respectively, prepared on the surface of an aqueous solution of MgCl2 and kept under a surface pressure of π = 25 mN/m. To neglect spectral changes due to side reactions, such as discoloration of the dyes, only the spectra observed within restricted periods of time are displayed in these figures. During these periods of time, the area per dye molecule, which is obtained by simply dividing the film area by the number of the dye molecules (and then not the area occupied by a dye molecule), gradually decreases from 0.73 to 0.70 nm2 and from 0.70 to 0.68 nm2 , respectively, in the DS–AA and DSe–AA films. (These decreases in the areas are not asymptotic and

2. Experimental The merocyanine dyes (DS and DSe, Fig. 1) and arachidic acid (AA) were purchased from the Japanese Research Institute for Photosensitizing Dyes Co. and Fluka Co., respectively, and were used without further purification. Pure or AA-mixed (1:1 molar ratio) solutions of the dyes were made using spectroscopic grade chloroform as the solvent. The concentration of the dyes was ca. 5 × 10−4 mol/l. These solutions were spread on a subphase contained in a Lauda Filmwaage trough under air at 20 ◦ C. The initial area of the subphase surface was 500 cm2 and the amount of the spread solution was 0.14 or 0.10 cm3 in the pure or mixed case, respectively. The L films were compressed to condensed states in ca. 180 s. Aqueous 4 × 10−4 mol/l solutions of MgCl2 and CaCl2 were used as subphase, after adding KHCO3 (3–5 × 10−6 mol/l). The pH value of these solutions kept in the trough was 6–7. Water with a resistivity greater than 1.8 × 107  cm was prepared by using a Millipore Milli-Q system. The electronic absorption spectra of the Langmuir films were recorded by a WRM-10TP polychrometer of Jasco Co. Ltd. with a halogen lamp as a light source and a bundle of optical

Fig. 2. Time evolution of the molar absorption coefficient (ε) spectrum observed for a DS–AA 1:1 (molar ratio) mixed Langmuir film prepared upon a Mg2+ containing subphase kept under a surface pressure of 25 mN/m. Inset: extraction of pure spectra of two types of J-aggregate based on the optimum difference spectrum (ODS) method. Average and difference spectra are also indicated.

214

K. Ikegami / Colloids and Surfaces A: Physicochem. Eng. Aspects 284–285 (2006) 212–216 Table 1 Spectral profiles of the J-band seen in L films of DS and DSe Cation Mg

Ca

Dye

Stage

λmax (nm)

ν (cm−1 )

FWHM (nm)

DS–AA

Initiala Final

617 605

2730 2410

15 19

DSe–AA

Initialb Final

620 609

2850 2550

18 21

DSc

Initial Final

617 602

2730 2330

13 19

DSe

Initial Final

620 602

2850 2360

– 20

610

2580

18

DSe–AA a b c

Fig. 3. Time evolution of the ε spectrum observed for a DSe–AA 1:1 mixed Langmuir film prepared upon a Mg2+ -containing subphase kept under a surface pressure of 25 mN/m. Inset: extraction of pure spectra of two types of J-aggregate.

discussions based on them are difficult.) The common remarkable feature of those spectral changes is the appearance of clear isosbestic points, showing that the L films consist of two components: J-aggregates with larger and smaller J-shifts. (Strictly speaking, unchanging components other than those J-aggregates may exist, but they have no significant influence on our discussion.) As discussed later, this fact provides an insight into the polymorphic behavior of those J-aggregates. No such spectral changes have been observed for pure DS and DSe films prepared on the MgCl2 solution. In these pure cases, the J-aggregates with the larger J-shifts are stable. More importantly, the J-aggregates with the larger J-shifts are much more stable under low surface pressures, e.g., 5 mN/m, both in the DS–AA and DSe–AA cases. The optimum difference spectrum method has been proved to be a powerful tool for extracting pure spectra from experimentally observed spectra of two component systems. This method calculates an “average spectrum” and an “optimum difference spectrum” based on the least-squares regression. In the virtue of the noise-reduction ability of the least-squares regression, the calculated spectra are usable in estimating the pure spectra through making linear combinations of them. Detailed explanation of ODS method is given in Ref. [14]. The extracted pure spectra of the two types of J-aggregate in the DS–AA and DSe–AA films on the MgCl2 solution are displayed in the insets of Figs. 2 and 3, respectively. These pure spectra are linear combinations of the averaged and difference spectra, which are drawn as the dotted and dashed curves, respectively, in the insets. From those pure spectra, the J-shift (ν) and full width at half maximum (FWHM) of the initial J-band of the DS–AA film are estimated at 2730 cm−1 and 15 nm, respectively

Pure DS films show a similar J-band. Pure DSe films show a similar J-band. DS–AA mixed films show similar J-bands.

(Table 1), and those of the final J-band of the film are estimated at 2410 cm−1 and 19 nm, respectively. Here, ν is defined as the shift from the absorption band of the dyes’ pure and dilute chloroform solution (seen at 528 and 527 nm for the DS and DSe cases, respectively). As for the DSe–AA film, ν of 2850 cm−1 and FWHM of 18 nm are obtained for the initial J-band, and ν of 2550 cm−1 and FWHM of 21 nm are obtained for the final J-band. The change in ν (FWHM) in the DS–AA case is approximately identical with that in the DSe–AA case, suggesting close similarity between the mechanisms of the spectral changes detected in these cases. Polymorphism of J-aggregates can be observed in pure L films of DS and DSe, too, when they are prepared on the surface of an aqueous solution of CaCl2 . In this case, the initial J-aggregates are not stable even under undetectably low surface pressures and no clear effect of the surface pressure has been detected in the range of 0–40 mN/m. Fig. 4 shows time evolution of the absorption spectrum of a DS (Ca) film. (In this experiment the surface pressure of the film is gradually increased from 0 to 5 mN/m and consequently the molecular area decreases from 0.69 to 0.59 nm2 , but effects of these changes upon the spectral change is unclear.) The ODS method has been applied to the data as shown in the inset of this figure and ν of 2730 (2330) cm−1 and FWHM of 13 (19) nm are obtained for the initial (final) J-band. The ν and FHWM values of the initial J-band in the DS (Ca) film are approximately identical with those of the initial J-band in the DS–AA (Mg) film, strongly suggesting that these J-aggregates are isostructural. In the DSe (Ca) film, unfortunately, the initial J-aggregate is too unstable to properly characterize, but its ν can be estimated at 2850 cm−1 . The more stable J-aggregate in the DSe (Ca) is characterized by ν of 2360 cm−1 and FWHM of 20 nm. The mixing effect of AA in the Ca-containing case should be mentioned. In DS films, mixing of AA (1:1 molar ratio) simply accelerates the spectral change. On the other hand, a J-band with ν of 2580 cm−1 and FWHM of 18 nm appears in DSe–AA (Ca) films. Reverse phenomena of the spectral changes described in this subsection have not been observed. Consequently, it has not been determined even for the DS–AA (Mg) and DSe–AA (Mg) cases

K. Ikegami / Colloids and Surfaces A: Physicochem. Eng. Aspects 284–285 (2006) 212–216

Fig. 4. Time evolution of the ε spectrum observed for a pure DS Langmuir film prepared upon a Ca2+ -containing subphase under low surface pressures (gradually increased from 0 to 5 mN/m). Inset: extraction of pure spectra of two types of J-aggregate.

whether the initial J-aggregates are the stable states under low surface pressures. 3.2. Size or geometry? It is well known that the J-shift is determined by the size and geometry of the J-aggregate [8,15–17]. Since the chemical conditions are maintained, the changes in ν observed in the DS and DSe films must be driven by changes in the aggregation numbers or changes in the molecular arrangements. The appearance of the isosbestic points (Figs. 2–4) implies that the spectral change in each case corresponds to a relaxation from a well-defined state to another well-defined state. As for the aggregation number (N), “well-defined” means (N/N)1/2  1, where N and N denote the average and deviation of N, respectively. The observed facts must give a clue to answer the question: which is responsible for the spectral changes, the size or geometry. If the geometry of the aggregate does not change, ν is a function of N: ν = f(N). In this case, writing the N values of the final and initial J-aggregates as n and n + m (m > 0), respectively, the ratio between the ν values of the aggregates (r < 1) is given by r=

f (n) . f (n + m)

(1)

Since it is obvious that f(n) monotonically increases with n, an assumption of m ≤ m gives an upper limit of n through the above equation as f (n) = rf (n + m) ≤ rf (n + m ).

(2)

215

As for the real form of f, for example, f(n) ∝ cos[π/(n + 1)] was given for non-periodic linear aggregates with nearest neighbor interactions (type I) [15,16]. Another simple example of f was given for cyclic linear aggregates (type II) as f(n) ∝ (n−1)/n [8,18]. The most rigorous estimation of m , i.e., m = ∞, gives 5.8, 5.4, 4.7 and 4.3 to the upper limits of n in the DSe–AA (Mg), DS–AA (Mg), DS (Ca) and DSe (Ca) cases, respectively, for type I, and gives 9.5, 8.6, 6.8 and 5.8 for type II, although the cyclic model is inconsistent with these small n values [19]. The fact that the present systems are condensed monolayers bound to water surfaces can restrict m . Diffusion of the molecules in such films is much slower than that in solutions. If the initial J-aggregates ((n + m)–mer) dissociates into two smaller aggregates (n–mer and m–mer), reorganization of m–mer to n–mer must take some time. If the (n + m)–mer dissociates into n–mer and m (>n) monomers, reassociation of monomers to n–mer must take some time, too. However, no clear indication of such reorganization or reassociation is seen in the spectra (Figs. 2–4). A special case of m = n is free from this difficulty. On putting m = n, the upper limits of n in the DSe–AA (Mg), DS–AA (Mg), DS (Ca) and DSe (Ca) cases are estimated at 4.8, 4.5, 3.9, and 3.5, respectively, for type I, and at 5.3, 4.8, 3.9 and 3.4 for type II [20]. On one hand, the above discussion shows that N should be small (≤5) if the observed spectral changes are driven by changes in N. On the other hand, however, the ν values of the present J-aggregates, which are very large in comparison with the transition dipole moments of the dyes (ca. 10 Debye), imply that N is large (≥several tens) [13,17,21]. In addition, the nucleation behavior found in metal-free DSe–AA mixed system showed that N is large (≥several tens) [11]. Thus, the assumption that the geometry of the aggregate does not change seems unlikely. In other words, although further discussion is needed for drawing a concrete conclusion, it is suggested that the main cause of the observed spectral changes is changes in the molecular arrangements of the aggregates, i.e., structural relaxations of the J-aggregates. 3.3. Driving force of J-aggregation Fig. 5 shows the correlation between ν and FWHM. Data observed for L films of DS(–AA) and DSe(–AA) prepared on an aqueous solution of CdCl2 are also displayed in this figure. It is clearly noted that FWHM monotonically decreases as ν increases. This monotonic behavior shows that the widths of the present J-bands are narrowed by exciton motion through the offdiagonals that cause the J-shifts [22]. The off-diagonals are given by the interactions between the transition dipoles and depend on the geometry of the aggregates. Therefore, both of the increase in FWHM and the decrease in ν seen in each DS or DSe film (see arrows in Fig. 5) indicate that the structural relaxation give rise to decreases in the interactions between the transition dipoles, and indicate that the aggregate with the smaller transition-dipole interactions is more stable than that with the larger transition-dipole interactions. Here, it should be remembered that the dyes bear large static dipoles along the long molecular axis (nearly parallel to

216

K. Ikegami / Colloids and Surfaces A: Physicochem. Eng. Aspects 284–285 (2006) 212–216

the aggregate. In other words, J-aggregate-to-J-aggregate structural relaxations take place in those L films. In addition, the fact that the ν (FWHM) decreases (increases) upon the relaxations implied that the interaction between the static dipoles born by the dye molecules cannot be the main driving force of the Jaggregation of DS and DSe. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] Fig. 5. Correlation between the J-shift (ν) and the full width at half maximum (FWHM) of the J-bands observed for DS(–AA) and DSe(–AA) films prepared on aqueous solutions of different cations. It should be noted that the initial Jaggregate in the DS (Ca) film exhibits ν of 2850 cm−1 , but its FWHM is unknown.

the transition dipoles), as shown by density-functional calculations [10,23,24], and the interaction between static dipoles must be approximately proportional to that between the transition dipoles. Then, the observed fact implies that the aggregate with smaller static-dipole interactions is more stable than that with larger static-dipole interactions, and that the main driving force of the J-aggregation cannot be the interactions between the static dipoles. Rather, it should be a local interaction that is more sensitive to the molecular shapes and arrangements. A recent study [13,25] upon the J-aggregates of dyes analogous to DS suggested that the main driving force of the J-aggregation is the generation of intermolecular hydrogen bonding or metal chelation. The present observations are consistent with that conclusion.

[11] [12] [13] [14] [15] [16] [17] [18]

[19]

[20]

4. Concluding remarks The time evolution of the ε spectrum was experimentally obtained for L films of DS–AA (Mg), DSe–AA (Mg) and DS (Ca). The appearance of the isosbestic point in each case indicated that the film consists of two types of J-aggregates: the one with the larger J-shift (initial state) and the one with the smaller J-shift (final state). By employing the ODS method, the pure spectra of these types of J-aggregate were extracted from the experimental data and then the ν and FWHM values were evaluated for each aggregate. Semi-quantitative consideration about ν suggested that the time evolution of the spectrum is a consequence of a change in the molecular arrangement in

[21]

[22]

[23] [24] [25]

T. Kobayashi (Ed.), J-Aggregates, Word Scientific, Singapore, 1996. M. Sugi, S. Iizima, Thin Solid Films 68 (1980) 199. T. Kawaguchi, K. Iwata, Thin Solid Films 165 (1988) 323. T. Kawaguchi, K. Iwata, Thin Solid Films 180 (1989) 235. T. Kawaguchi, K. Iwata, Thin Solid Films 191 (1990) 173. A. Ulman, An Introduction to Ultrathin Organic Films, Academic Press, Boston, 1991. M. Yoneyama, T. Nagao, T. Murayama, Chem. Lett. (1989) 397. A. Miyata, D. Heard, Y. Unuma, Y. Higashigaki, Bull. Chem. Soc. Jpn. 66 (1993) 999. N. Kato, K. Saito, T. Serata, H. Aida, Y. Uesu, J. Chem. Phys. 115 (2001) 1473. N. Kato, M. Yamamoto, K. Itoh, Y. Uesu, J. Phys. Chem. B 107 (2003) 11917. K. Ikegami, Jpn. J. Appl. Phys. 41 (2002) 5444. K. Ikegami, Trans. Mater. Res. Soc. Jpn. 28 (2003) 31. K. Ikegami, J. Chem. Phys. 121 (2004) 2337. K. Ikegami, Trans. Mater. Res. Soc. Jpn. 30 (2005) 147. P.O.J. Scherer, Chapter 4 in Ref. [1], pp. 99–100. It is obvious that “4V cos(π/N + 1)” in Ref. [15] should be read as “4V cos[π/(N + 1)]”. Y. Hamanaka, H. Kurosawa, A. Nakamura, Y. Uchiyama, K. Marumoto, S. Kuroda, Chem. Phys. Lett. 363 (2002) 233. Note that if only the nearest neighbor interactions are taken into account, the J-shift of a periodic linear aggregate is independent of N, as is shown by Eq. (10) in Ref. [15] with k = 0 and V1,j = 0 for j ≥ 3. The small aggregation numbers (≤6) proposed in Ref. [8] are inconsistent with the cyclic model (ansatz of the proposal), too, and then they are neglected in this work. In case of a simple two-dimensional (2D) aggregate, v may be expressed as v = f(nx , ny ) = fx (nx ) + fy (ny ) with the nearest neighbor approximation, where fk (n) = wk cos[π/(nk + 1)] (k = x, y). Here, the case in which a nx × 2ny aggregate dissociates into two nx × ny aggregates is considered. Under nearly isotropic conditions, i.e., wx ≈ wy and nx ≈ ny , Eq. (2) is modified as fx (nx ) + fy (ny ) ≈ 2fy (ny ) ≤ r[fx (nx ) + r fy (2ny ). fy (2ny )] ≈ r[fy (ny ) + fy (2ny )], and then as fy (ny ) ≤ 2−r Since r/(2 − r) is smaller than r, this condition gives a tighter restriction than Eq. (2). Thus, nx × ny may be estimated at 10 or smaller for nearly isotropic 2D aggregates. This discussion does not hold when wx  wy and nx ny (the intermolecular interaction is stronger in the y direction and the aggregate grows in the x direction), but such cases are not very likely. Strictly speaking, Eqs. (1) and (3) in Ref. [17] are not rigorous (they are inconsistent with each other). Nevertheless, the discussion in Ref. [17] estimates the lower limit of the aggregation number of the J-aggregates of a merocyanine dye analogous to DS. The widths of the J-bands include a contribution of the inhomogeneous broadening effect, but such a contribution is not expected to have a strong correlation with ν. Thus, the change in FWHM correlating with ν can be considered a contribution of the motional narrowing effect. K. Ikegami, S. Kuroda, Chem. Phys. 295 (2003) 205. K. Ikegami, M. Lan, Colloids Surf. A 257–258 (2005) 143. K. Ikegami, Curr. Appl. Phys., in press, doi:10.1016/j.cap.2005.04.046.