2310
J. Phys. Chem. 1996, 100, 2310-2321
Liquid-Crystalline J-Aggregates Formed by Aqueous Ionic Cyanine Dyes William J. Harrison Research DiVision W93, Kodak Ltd., Headstone DriVe, Harrow, Middlesex, HA1 4TY, England
Donna L. Mateer and Gordon J. T. Tiddy* DiVision of Chemical Sciences, Science Research Institute, UniVersity of Salford, Salford, M5 4WT, England ReceiVed: August 29, 1995; In Final Form: October 25, 1995X
NMR spectroscopy (2H and 1H), polarized-light optical microscopy, and synchrotron X-ray diffraction have been used to determine the liquid-crystal structures (mesophases) formed at room temperature by four anionic cyanine dyes in water. A sharp, intense, bathochromically-shifted absorption band appeared in the UVvisible absorption spectra on increasing dye concentration for all the dyes, demonstrating that so-called J-aggregates are formed. The onset of J-aggregate formation is practically concomitant with the occurrence of mesophases. In some instances, these mesophases form at much less than 0.1% w/w. X-ray diffraction shows that three of the dyes form layer (lamellar) phases while one forms columnar nematic and hexagonal phases. These structures are consistent with the textures observed by optical microscopy. Unusually, the columns have a multimolecular cross section rather than being unimolecular. We propose a novel “hollow pipe” structure for these aggregates. The liquid-crystal type appears to be largely dependent on the precise molecular structure of the dye, presumably due to the short-range intermolecular interactions (electrostatic, steric, and van der Waals). Previously postulated stacking geometries for low-dimensional J-aggregates are compared to the liquid-crystal structures. Since the J-aggregates form a separate liquid-crystalline phase they consist of thousands of molecules or more, rather than the small aggregation numbers deduced by mass action considerations.
1. Introduction Cyanine dyes are employed extensively as spectral sensitizers in photographic systems.1-3 When adsorbed to silver halide crystals, they may induce sensitivity to longer wavelengths of the visible spectrum or to near-infrared radiation, selectively to radiation within their own absorption bands. The generic cyanine dye structure is composed of two basic (electron donating) heterocyclic rings usually containing nitrogen linked by a conjugated chain commonly having an odd number of methine carbons (Figure 1). The positive charge on one nitrogen is extensively delocalized between the two nitrogen atoms. Due to the strong intermolecular van der Waals attractive forces operating between these, usually planar, polycyclic molecules, aggregation or self-association occurs frequently both in aqueous solution and on solid surfaces.4,5 Consequently, the characteristic absorption spectrum exhibited by the monomeric state is modified. For the solutions, deviations from Beer’s law accompanied by distinct changes in band shape and large spectral shifts relative to the monomer M-band (typically (50 nm) are observed, which have been attributed to the formation of dye dimers, trimers, tetramers, and n-mers.1 Blue shifts to relatively shorter wavelengths are customarily referred to as H-bands (for hypsochromic) while bathochromic shifts to longer wavelengths may give rise to J-bands (probably named after Jelley6). Similar spectral transitions occur for H- and Jaggregates adsorbed to silver halide surfaces.2 The red-shifted J-aggregate is the most commercially important dye assembly for photographic spectral sensitization and is characterized by a narrow, intense excitonic absorption band. Recent interest has also focused on the ability of J-aggregates to exhibit coherent excitation phenomena with the potential for strongly enhanced nonlinear optical properties.7 * To whom correspondence should be addressed. X Abstract published in AdVance ACS Abstracts, January 1, 1996.
0022-3654/96/20100-2310$12.00/0
There appears to be a consensus that both the H- and J-aggregates formed at interfaces are composed of parallel dye molecules stacked plane-to-plane and end-to-end forming two-dimensional crystals (Figure 2a). In solution, analogous one-dimensional dye assemblies have been proposed (Figure 2b). According to theoretical point-dipole,8 extended-dipole,9 and transition density10 calculations, whether a dye aggregate exhibits a bathochromic or hypsochromic spectral absorption shift depends on the angle of slippage, R, between successive molecular planes (where R defines the angle between the line-of-centers of a column of dye molecules and the long axes of any one of the parallel molecules). Large molecular slippage (e.g. R < ∼32°) results in a bathochromic shift while small slippage (R > ∼32°) results in a hypsochromic shift.10 In bulk solution, the self-association of cyanine dyes displays a complex dependency upon factors such as dye structure, concentration, solvent polarity, pH, ionic strength, and temperature.4 Consequently, definitive dye aggregate structures have never been determined directly at the molecular level and remain the subject of much speculation and controversy. Many different, often conflicting, solution aggregate models have been proposed largely on the basis of indirect evidence. This includes, for example, X-ray crystallographic structure analyses of 3-D cyanine dye crystals;11,12 theoretical approximations of aggregate wavelength shifts relative to intermolecular orientations;8-10 low-temperature polarized-luminescence optical microscopy of J-aggregates on AgBr and AgCl emulsion grains;13,14 electron microscopy of solution-formed H-aggregates;15 electron diffraction,16 surface second-harmonic generation, and scanning force microscopy studies of LangmuirBlodgett dye monofilms;17,18 and scanning tunneling microscopy of a cyanine dye J-aggregate monolayer on a highly {111}oriented silver film.19 © 1996 American Chemical Society
J-Aggregates Formed by Aqueous Ionic Cyanine Dyes
Figure 1. Cyanine dye structures: dyes A, B, C, and D from top to bottom.
In a recent communication,20 we described the formation of liquid-crystalline layer (lamellar/ smectic) and columnar phases by a dilute aqueous cyanine and azo dye, respectively. This class of mesophases has been termed “chromonic” to distinguish it from amphiphillic lyotropic liquid crystals.21 In marked contrast to conventional surfactant lamellar and hexagonal lyotropic mesophases, these aggregated dye structures were found to possess unusually high short- and long-range molecular order. Here, the aqueous phase behavior and corresponding spectral absorption characteristics of several disparate anionic cyanine dyes have been quantified in detail using synchrotron X-ray diffraction, polarized-light optical microscopy, NMR, and UV-vis absorption spectroscopy. The equilibrium superstructures of the self-assembled solution J-aggregates have been determined, unequivocally, for the first time and their architectures compared with postulated aggregate models. 2. Experimental Section The dyes used in this study (Figure 1) were synthesized at the Research Laboratories of Eastman Kodak Co. with assays of g97.5% (by UV-vis spectroscopy). Aqueous dye samples
J. Phys. Chem., Vol. 100, No. 6, 1996 2311 were prepared both in normal water, which was deionized and purified by a Millipore Milli-RO 4 and Milli-Q SP reagent system (typical water resistivity, 18.3 MΩ cm at 19 °C), and deuterium oxide (2H2O) which was obtained from the Sigma Chemical Co. (99.9 atom % 2H) and used as received. No differences in phase behavior were observed for samples containing equal mole fractions of normal or heavy water. For consistency, all compositions were corrected for the difference in molecular weight between normal and heavy water and correspond to the 2H2O composition. Bulk aqueous dye samples were prepared gravimetrically to (0.05 mg in screw-topped glass vials. Typical sample sizes were 2.0-5.0 g, depending on composition. Homogenization of dilute dye samples was normally achieved by repeated agitation at ambient temperature over a period of days or weeks. For the more concentrated dye samples, dissolution and mixing was also facilitated by moderate heating (e50 °C) followed by prolonged equilibration (typically several months) at room temperature. In order to minimize photolytic degradation of the aqueous dyes during storage, all sample containers were covered with aluminum foil. Cyanine dyes containing benzoxazole moieties (i.e., dye B) are susceptible to irreversible hydrolytic degradation.1 Thus, measurements involving dye B were conducted generally within 72 h of preparation. Absorption spectra were recorded at ambient room temperatures of 24 ( 1 °C with a Perkin-Elmer Lambda 7 UV-vis spectrophotometer, using a slit width (effective bandpass) of 2 nm and a scan speed of 120 nm/min. Aqueous dye samples were generally contained in stoppered rectangular quartz cells of 1 or 10 mm path length. For concentrated dye samples (typically >1.0% w/w) wavelength-reproducible spectra with absorbance values below 3 were obtained from very thin sample films held between quartz plates by capillarity. Optical microscopy observations of aqueous dye films were performed with a Vickers M41 Photoplan polarizing microscope equipped with a Leitz 350 hot-stage. The specimen temperature was measured to ( 1 °C using mercury thermometers covering the ranges -30 to 80 °C and +60 to 170 °C. The samples were contained between standard glass microscope slides and cover slips or sealed in flat optical microslides of 200 µm path length (Camlab, U.K.). Mesophase formation was surveyed initially using the penetration technique of Lawrence.22 Here, water was allowed to diffuse into the solid dye, placed between a slide and cover slip, to establish a concentration gradient across the sample. Characteristic mesophase textures may then develop as separate bands around the solid dye. Alternatively, slow peripheral solvent evaporation of dilute dye preparations was similarly employed. 2H NMR spectra of dyes in 2H O were recorded at ambient 2 room temperature using a JEOL JNM-FX100 spectrometer operating on the deuterium lock channel at 15.26 MHz. Here, the magnetic field direction was perpendicular to the rotational long axis of the sample tube. Samples were contained in standard 10 mm o.d. NMR tubes which were stoppered and sealed with Nescofilm. Typically, 1-16 accumulations were sufficient to resolve low-noise spectra for all compositions studied. The sample temperatures were controlled by ambient temperature compressed air and measured directly in the NMR probe using a Comark 9001 digital thermometer fitted with a calibrated NiCr/ NiAl thermocouple and were constant to (0.1 °C. The reproducibility of the quadrupole splittings over a period of several months was, in most cases, better than (3%. 1H NMR spectra were recorded at 25.0 °C using a Varian UNITY plus spectrometer operating at 399.93 MHz under standard high-resolution conditions.
2312 J. Phys. Chem., Vol. 100, No. 6, 1996
Harrison et al.
Figure 2. Schematic representations of 2-D and 1-D cyanine dye aggregates with brickwork (A, A′), ladder (B, B′), and staircase (C, C′) molecular arrays: (a) dye molecules adsorbed on a mica surface (the surface is in the plane of the paper); (b) dye molecules in solution. Aggregate models reproduced from ref 9a.
TABLE 1: Spectral Absorption Data for Cyanine Dyes in 2H O at 24.5 °C 2 dye
λ-M, nm
λ-H, nm
λ-J, nm
J-aggregate onset, % w/w dye
A B C D
428 502 545 549
406 470 504 492
463 542 618 621
0.007 0.13 0.0007 0.74
3. Results
Figure 3. UV-visible absorption spectra of dye A in H2O at 24 °C: (a) M h H h J-band equilibria at 0.025% w/w dye (1 mm quartz cells); (b) sharp J-band at 0.80% w/w dye (quartz plates); (c) broad “J-band” at 3.53% w/w dye (quartz plates).
X-ray diffraction (XRD) experiments were conducted at the EPSRC Synchrotron Radiation Laboratory (Daresbury, U.K.) using Stations 8.2 (small angle) and 7.2 (wide angle). The aqueous dye samples were contained in flame-sealed Lindemann glass capillaries (1.0 mm diameter) which were mounted in a modified Linkam optical microscope hot-stage equipped with a Linkam TMS 90 temperature control system and maintained at 25.0 °C. The small-angle diffraction patterns were recorded by a 2-D multiwire proportional counter detector, while wideangle patterns were captured on photographic film or phosphor image plates. Small- and wide-angle calibrations were performed using oriented collagen fibres and powdered silicon, respectively.
3.1. UV-Visible Spectroscopy. Irrespective of molecular structure, each dye exhibited a similar concentration-dependent sequence of spectral transitions in aqueous solution at room temperature. Progressive increases in dye concentration resulted in an initial loss of M-band intensity at the expense of the coexisting H-band(s), followed by the sudden appearance of an additional, characteristically sharp, J-band above a critical dye concentration (Figure 3a). Subsequent dye additions shifted the M h H h J-band equilibria in favor of a single J-aggregate state (Figure 3b). Further increases in dye concentration generally produced a slight apex-broadening of the J-band and a gradual growth in intensity of the weak blue-shifted Jcomponent (thought to correspond to the forbidden transition of the J-state4), culminating in a broad “J-band” (Figure 3c). The pertinent data are summarized in Table 1. The J-band maxima were generally reproducible to better than (1 nm, though slightly greater fluctuations in peak positions were sometimes encountered for the broader and more concentrationdependent H-bands. The H-band wavelengths are those measured at the onset of J-aggregation. The tabulated values for the onset of J-aggregation are believed to be reproducible to (10%. The main sources of variance include lack of proper sample temperature control, dye solution stability allied with
J-Aggregates Formed by Aqueous Ionic Cyanine Dyes
J. Phys. Chem., Vol. 100, No. 6, 1996 2313
errors in gravimetric preparation, and dilution and sampling of heterogeneous two-phase dye dispersions. 3.2. Optical Microscopy. Polarized-light microscopy observations of thin-film aqueous preparations of all four dyes revealed characteristic birefringent mesophase textures at room temperature. Based on their mesophase structural elements (layers versus columns), two distinct types of phase behavior have been identified. For dye B, the so-called tiger-skin texture21 characterized the fluid, viscoelastic columnar nematic (N) mesophase formed at ∼0.13% w/w dye (Figure 4a). Nematic schlieren and homogeneous textures were also common features of these dilute preparations. The nematic mesophase onset was practically concomitant with the appearance of the spectroscopic Jaggregate absorption band. Subsequent peripheral evaporation produced a more viscous phase with a broken fan-shaped texture typical of a more concentrated hexagonal (M) mesophase. The N/M boundary was ill-defined, indicating a weak first- or higherorder N to M phase transition21 at ∼2.5% w/w dye. The aqueous phase behavior of dye B conformed to the archetypal columnar N/M mesophase sequence exhibited by similar drug and azo-dye molecules.20,21,23 These phase assignments were confirmed by deuterium NMR and X-ray measurements (see below). In marked contrast, dyes A, C, and D generally exhibited ill-defined grainy or “broken frond-like” birefringent textures at low concentrations. The dilute solutions also possessed a distinctive macroscopic turbid opalescence or lustrous sheen and exhibited spectroscopic J-bands. These optical properties could be ascribed specifically to lamellar/smectic (layered) mesophase formation.20 The more dilute solutions which contained Haggregates were optically transparent and invariably isotropic. Somewhat more concentrated mesophase preparations generally exhibited a non-geometric “frond-like” optical texture (Figure 4b) and possessed rather unusual rheological characteristics. Unlike conventional aqueous surfactant lamellar (LR) mesophases, the dye smectics tended to flow as viscous birefringent domains along well-defined boundaries when shear was applied to the cover slip. Furthermore, they did not readily adopt homeotropic orientations nor display the oily-streak or mosaic textures commonly associated with surfactant LR mesophases. 3.3. 2H NMR Measurements. These experiments pertain specifically to the measurement of deuterium quadrupole splittings from dye mesophase-associated 2H2O molecules. Comprehensive theoretical treatments of quadrupolar effects in aqueous surfactant mesophases have been reported previously.24 Only a brief outline is presented here. The deuterium nucleus has a spin quantum number (I) greater than 1/2 (2H, I ) 1) and hence possesses an electric quadrupole moment which interacts with electric field gradients (EFG’s). In an environment with noncubic symmetry, as in most liquid crystals, this results in a splitting of the 2H resonance into 2I peaks. In isotropic media, no splitting occurs. The frequency difference between adjacent maxima (for a uniaxial mesophase) is termed the quadrupole splitting, ∆. For a macroscopically aligned mesophase (where the director has a single orientation throughout the sample), ∆2H is given by
∆2H ) 3/4EqS[(3 cos2 θLD) - 1]
(1)
where, Eq is the characteristic quadrupole coupling constant; θLD is the angle between the mesophase director and the external magnetic field; S is an order parameter for the nuclear EFG which varies with the time-average angle, θDM, between the electric field gradient and the mesophase director, and is given by
S ) 1/2[(3 cos2 θDM) - 1]
(2)
For water in lyotropic mesophases, a simple two-site model25 involving the rapid exchange of free and bound molecules can be used to describe the dependence of the order parameter on composition. Here, we assume that the values of ∆2H are dominated by the fraction of aggregate-bound 2H2O molecules, Fb, where the anisotropy is greatest. The free 2H2O molecules are considered to have zero quadrupole splittings as a result of the vanishing net orientation of the EFG’s, while bound 2H2O molecules have a quadrupole splitting ∆b. Thus
∆2H ) Fb∆b
(3)
Assuming a constant number of bound water molecules per dye molecule, nb, over the entire mesophase concentration range (a valid approximation for many aqueous surfactant mesophases below ca. 70% surfactant), eq 3 becomes
∆2H ) nb(Xd/Xw)∆b
(4)
where, Xd and Xw are the mole fractions of dye and 2H2O respectively. Equation 4 may also be simplified by assuming that for a given mesophase structure, ∆b is also invariant with concentration. Thus:
∆2H ) k(Xd/Xw)
(5)
where k is a constant. A plot of ∆2H versus mole ratio (dye/2H2O) for dyes A-D at nominally 25 °C is presented in Figure 5. The linear concentration dependence of ∆2H and nonzero intercepts are consistent with a biphasic (mesophase plus isotropic solution) coexistence regimes in accordance with the lever rule. At higher dye concentrations, not reproduced in Figure 5, linear plots passing through the origin characterized the single-phase mesophase regimes.33 For each dye, the increases in ∆2H shown in Figure 5 were accompanied by increases in J-band absorbance (mesophase) at the expense of the coexisting M- and H-bands (isotropic dye solution). Within these heterogeneous biphasic regions, 2H spectra for the individual phases were not resolved due to fast solvent exchange between the isotropic and liquid-crystalline phases. Consequently, over the concentration ranges represented in Figure 5 (e2.5% w/w dye), the observed 2H spectrum was a single symmetrical doublet, with a quadrupole splitting corresponding to the weighted mean of those for the isotropic (0 Hz) and liquid crystal phases. In all cases, the 2H line shapes were indicative of monodomain (macroscopically aligned) mesophase samples.26 This is clearly illustrated in Figures 6a and 7a for dyes B and C, respectively. However, the exchange of dye molecules between the two phases is much slower, and hence they give superimposed proton NMR spectra. The isotropic phase has a high-resolution 1H NMR spectrum (Figure 8a), while, for the mesophase, lines due to individual dye protons are no longer resolved due to dipolar interactions. For dye A, the mesophase 1H spectrum was characterized, therefore, by three broad lines attributed to a 2HO1H resonance and single CH3 and CH2 resonances from the triethylammonium counterion (Figure 8b). As the mesophase concentration decreased, there was a concomitant decrease in ∆2H. Eventually, ∆2H became less than the natural 2H line width (∼4 Hz, nonspinning), so that dilute biphasic samples were characterized by a 2H singlet. However, by linearly extrapolating the ∆2H data to zero (linear correlation coefficients g0.998) an apparent onset concentration
2314 J. Phys. Chem., Vol. 100, No. 6, 1996
Harrison et al.
Figure 4. Optical microscope textures of cyanine dye lyotropic mesophases observed during H2O penetration scans (crossed polars, ×100): (a, top) tiger-skin texture of dye B columnar nematic mesophase at 80 °C. (b, bottom) Schlieren-like texture of dye A smectic mesophase at 22 °C.
of mesophase formation for each dye could be estimated. These values are presented in Table 2. In reality, the actual onset concentration was lower, being practically concomitant with the
formation of solution J-aggregates quantified by absorption spectroscopy (Table 1). This was evidenced by pronounced changes in the solution properties following J-aggregation, e.g.,
J-Aggregates Formed by Aqueous Ionic Cyanine Dyes
J. Phys. Chem., Vol. 100, No. 6, 1996 2315
Figure 5. Deuterium quadrupole splitting values (∆2H) for aqueous cyanine dye samples as a function of dye/2H2O mole ratio at 25 ( 1 °C.
Figure 6. Determination of the sign of the diamagnetic susceptibility anisotropy, ∆χ, for the aqueous columnar hexagonal mesophase of dye B (3.53% w/w dye in 2H2O) at 26.6 °C using 2H NMR: (a) after cooling stationary sample from 40.0 °C in NMR probe (B0 ) 2.3 T); (b) immediately after rotating sample by 90° with respect to B0.
onset of macroscopic heterogeneity, turbidity, birefringency, and increased viscosity. For dye B, the inconsistency between the 2H NMR and absorption spectroscopy derived concentrations may be related largely to the more temperature-sensitive stability
of the nematic phase, particularly at low dye concentrations. For dyes A and C, however, the orders of magnitude discrepancy in these values is probably due to an alteration in the composition of both the isotropic solution and the separated mesophase in very dilute mixtures. This arises because of the dye protonation equilibria commonly encountered with cyanine dyes in aqueous media.27 Depending upon dye structure, the stoichiometry of the protonation reaction (dye- + H+ h dye0), and hence the effective number of solution components and their relative concentrations, may vary in a nontrivial manner with the overall composition and aggregation state of the system. Dye-dye interactions, and particularly J-aggregation, can decrease the apparent dye basicity, sometimes by more than 4 pK units.27 Thus, for smectic mesophase-forming dyes, where relatively extensive mesophase (J-aggregates) plus isotropic dye solution (M h H-aggregates) coexistence regimes are commonplace, the composition and order parameter (and hence ∆2H) of the smectic J-aggregates may vary significantly at ultralow mesophase concentrations i.e. within the linearly-extrapolated concentration ranges. Thus, where the concentration of neutral (protonated) dye is greatest (at the very onset of J-aggregation), the resulting dilute smectic layers, composed of both protonated neutral and nonprotonated anionic molecules, experience reduced short-range dye-dye electrostatic repulsions and enhanced mesophase stability. This has the consequence that the fractions of each phase do not vary according to the lever rule. The situation can be compared to that of micelle formation for ionic surfactants in the presence of small concentrations of uncharged cosurfactants. The first aggregates to form are rich in the nonionic amphiphile, often forming a separate lamellar phase. At higher surfactant concentrations the ionic surfactant dominates the aggregate composition to form a micellar solution with a composition of almost pure ionic surfactant. The pronounced difference in gradient apparent for dye B in Figure 5 largely reflects differences in the diamagnetic susceptibility anisotropy (hence θLD and ∆2H) and, to a lesser extent,
2316 J. Phys. Chem., Vol. 100, No. 6, 1996
Figure 7. Determination of the sign of the diamagnetic susceptibility anisotropy, ∆χ, for the aqueous smectic mesophase of dye C (1.02% w/w dye in 2H2O) at 26.2 °C using 2H NMR: (a) 16 min after inserting stationary sample into NMR probe (B0 ) 2.3 T) following vigorous agitation; (b) immediately on cessation of spinning sample at 6 Hz about an axis orthogonal to B0; (c) after 67 h stationary in the NMR probe following spinning.
the order parameter, S, of the columnar nematic mesophase compared with the smectic layered mesophase of dyes A, C, and D. These fundamentally different J-aggregate structures (1-D columns versus 2-D layers) may be differentiated by the nature of the respective mesophase director response to the aligning spectrometer magnetic field, B0.26,28 For dyes A, C, and D, the smectic J-aggregate mesophase possesses a positive diamagnetic susceptibility anisotropy, ∆χ. Thus, for dilute, lowviscosity smectic samples, the director (normal to the plane of the layers) is spontaneously aligned parallel to B0. The aromatic dye moieties are thus aligned preferentially with their planes parallel to the magnetic field. The resulting homeotropic distribution of directors (θLD ) 0°) was characterized by a single symmetrical doublet spectrum. This is illustrated for dye C in Figure 7a. After spinning the monodomain sample about an
Harrison et al. axis perpendicular to B0, a 2H powder spectrum was observed (Figure 7b) resulting from a two-dimensional (planar) distribution of directors in a plane perpendicular to the rotation axis and parallel to B0. The powder splitting (42.0 Hz) clearly corresponded to half the original aligned splitting (84.0 Hz), behavior consistent with a uniaxial smectic structure.26 Following a further 67 h in the magnetic field, the intensity of the signals from the directors at θLD ) 0° (the outer singularities of the powder spectrum) became strongly accentuated as slow director re-alignment proceeded (Figure 7c). These evolutionary line shapes confirmed the positive value of ∆χ and supported the smectic structural model derived from X-ray diffraction studies (see below). For dye B, the columnar J-aggregate mesophases (nematic and hexagonal) possess a negative ∆χ and align spontaneously with the director (which is parallel to the column long axis and normal to the molecular planes) perpendicular to B0.28 This is illustrated for a hexagonal mesophase sample in Figure 6a. The initial well-resolved doublet (∆2H ) 59.6 Hz) corresponds to an isotropic distribution of directors in the plane perpendicular to B0 (θLD ) 90°). This was confirmed by subsequently rotating the aligned sample by 90° with respect to B0 and immediately rerecording the spectrum prior to realignment (Figure 6b). The resulting powder line shape, with an essentially invariant quadrupole splitting (∆2H ) 56.9 Hz), was then characteristic of an uniaxial mesophase possessing a distribution of directors in a plane containing the B0 vector. The outer shoulders corresponding to director orientation along B0 are clearly visible, although a full 2-D powder pattern is not observed due to partial reorientation. 3.4. X-ray Diffraction. In aqueous solution, dyes A, C, and D exhibited concentration-dependent small-angle XRD patterns with sharp Bragg reflections in positions corresponding to lattice spacings of d, d/2, d/3, d/4, d/5, d/6, etc. These spacings are characteristic of {00l} reflections from a stratified smectic mesophase possessing long-range one-dimensional periodicity. Here, the stacked and alternating layers are composed of dye and water, respectively (Figure 9). The position of the {001} reflection gives the total repeat spacing d0, (the unit cell dimension) equivalent to the thickness of one dye layer, blt plus one water layer, dw. The mesophase structural elements (J-aggregates) are, by inference, 2-D dye layers, assumed to be rigid, well-ordered, and continuous (infinite). These assumptions are consistent with the molecular architecture, the observation of an unusually large number of both smalland wide-angle Bragg reflections20 and the relatively large 2H NMR quadrupole splittings of mesophase-bound 2H2O molecules. In reality, the smectic domain size will be determined by both the free energy of defects and the sample history (e.g., shear and temperature effects). In marked contrast, dye B (at concentrations g3.0% w/w dye) exhibited a series of sharp, concentration-dependent Bragg reflections with spacings of d, d/x3, d/x4, d/x7, d/x9, etc. These XRD patterns are characteristic of a hexagonal mesophase possessing long-range two-dimensional periodicity. The mesophase structural elements in this case are, by inference, 1-D columns of dye molecules stacked plane-toplane, with the columnar long axis perpendicular to the molecular planes (consistent with 2H NMR mesophase-alignment studies). In order to extract structural dimensions from the XRD data, we have employed the Luzzati29,30 approach for surfactant systems. Here, it is assumed that the dye aggregates consist of flat, infinite layers in the smectic phase and indefinitely long, circular cylinders in the hexagonal phase. Secondly, the water
J-Aggregates Formed by Aqueous Ionic Cyanine Dyes
J. Phys. Chem., Vol. 100, No. 6, 1996 2317
Figure 8. 1H NMR spectra for dye A in 2H2O at 30.0 °C: (a) biphasic dispersion of smectic dye mesophase in isotropic dye solution (0.30% w/w dye); high-resolution spectrum observed for dye solution component only; (b) homogeneous smectic dye mesophase sample (3.53% w/w dye). Dye protons not resolved under high-resolution conditions due to dipolar broadening. In both spectra 1HO2H signal arbitrarily set to δ ) 5.0 ppm.
and dye components occupy distinct regions within each mesophase; i.e., the interface is smooth and sharp. All the water,
dye SO3- hydrophilic headgroups and counterions together constitute a polar region, and the remaining alkyl group/benzene
2318 J. Phys. Chem., Vol. 100, No. 6, 1996
Harrison et al.
TABLE 2: Apparent Concentration Onset of Dye Mesophase in 2H2O by 2H NMR dye A B C D
mesophase type
mesophase onset, % w/w dye
temp, °C
smectic columnar nematic smectic smectic
0.087 ( 0.004 0.42 ( 0.02 0.047 ( 0.005 0.81 ( 0.02
25.5 26.0 25.0 24.0
moiety, a nonpolar region. While this scenario is reasonably accurate for flexible amphiphilic molecules self-assembled into cylindrical micelles (e.g., the hexagonal H1 mesophase), its applicability to rigid polyaromatic mesogens aggregated in the hexagonal M mesophase has been questioned23 and is currently a matter of speculation.20,31 We use the model here for simplicity, but recognize that there are several approximations (e.g., the aggregates are not smooth and of uniform thickness). Finally, the densities of the components within the mesophase are assumed equal to their bulk values at the same temperature and pressure. For the benzene layer we employ a density of 1.200 kg dm-3, while that of the aqueous layer is 1.107 kg dm-3. We have calculated the benzene layer thickness (blt) and the area per molecule (Sbl) for the smectic phase, and the column cross section area for the columnar phase. The calculated value of blt for each smectic system (Table 3) is consistent with monomolecular layers where the molecular long axes lie orthogonal to the mesophase director and parallel to the layer plane (Figure 9). Furthermore, the calculated values of Sbl (Table 3) for these infinite dye monolayers, composed of parallel, close-packed molecules, are in reasonable agreement with the product of the respective molecular length and the intramolecular separation (∼3.5 Å, determined by wide-angle XRD). Within these layers, the dye molecules are arranged a priori with their sulfonate moieties uniformly distributed above and below the layer plane. Such an arrangement would provide the net negative charge distribution necessary for the electrostatic repulsive stabilization of the long-range smectic order. In order to produce the characteristic J-aggregate spectral absorption band, the molecules are probably arranged in a brickwork-like array.9 Indeed, our observations of multiple, often sharp and intense XRD reflections in the approximate range 3-50 Å, with concentration-invariant positions, are congruent with an unusually high degree of intralayer molecular order.20 For these rigid, biaxial molecules, this introduces a potential for complex intralayer structural polymorphism, reminiscent of the thermotropic smectic mesophases.32 In marked contrast to aqueous single surfactant systems, where the lamellar (smectic) phase usually has a limited swelling range, the dye smectic phases were stable over a considerably larger dilution range. This is exemplified in plots of the XRD spacing, log d0, versus reciprocal volume fraction of dye, log(1/φd), illustrated for dyes A and D in Figure 10, a and b, respectively. Here, the maximum interlayer spacing of dyes A and D coincides with a pronounced discontinuity in slope (typically from a gradient of ∼1 to, essentially, zero) at approximately 2.2 and 8.2% w/w dye, respectively. The water layer thickness is 1-2 orders of magnitude larger than that of the dye layers. For all three dye systems, the gradient of ∼1 is characteristic of a one-dimensionally swelling smectic mesophase, where the dye exhibits a negligible solubility in the interleaving water layers. Thus, upon dilution, additional water is incorporated solely in the region dw between the dye layers and is not intercalated between the dye molecules. Consequently, the composition of the dye layers (and hence the parameters blt and
Figure 9. Schematic representation of an aqueous cyanine dye smectic mesophase. The rectangles represent individual dye molecules which are close-packed into infinite 2-D brickwork-like monolayers (Jaggregates). The biaxial dye monolayers are oriented to produce a uniaxial mesophase structure defined by a single director, n.
TABLE 3: Small-Angle XRD Data for Dye-2H2O Smectic Mesophases at 25 °C dye A C D a
blt, Å
max dw, Å
Sbl, Å2
gradienta
9.2 ( 0.5 8.8 ( 0.9 9.4 ( 0.3
688 g717 197
66 ( 4 85 ( 9 66 ( 2
0.98 0.89 0.93
Gradient of log d0 versus log (1/φd) plot.
Sbl; Table 3) remains essentially independent of the water content throughout the homogeneous (single-phase) liquid-crystal regime. For dyes D and C the slope is slightly smaller than unity (Table 3) producing an apparent increase in blt as the dye concentration is increased. This is reflected in their relatively large standard deviations. The trend may result simply from a systematic error related to the dye purity, e.g., an undefined solvent content or other nonmesogenic impurity, but we are unable to locate its origin. However, it could indicate that the short molecular axis is becoming tilted on dilution. We note that such changes in the molecular area with concentration are common in surfactant mesophases. For dye D, the pronounced discontinuity in slope (at ∼8.2% w/w) results from a first-order phase transition between the homogeneous (one-phase) smectic liquid crystal and a heterogeneous (two-phase) dispersion, composed of swollen smectic mesophase (J-aggregates) in equilibrium with dilute isotropic dye solution (composed of monomer and H-aggregates). Here, the composition and structure of the J-aggregate mesophase become fixed over a wide concentration range, with dw ∼197 Å and blt ∼9.4 Å, and its relative abundance decreases monotonically toward zero at ∼0.8% w/w dye. On prolonged standing, dilute biphasic samples of dye D in 2H2O (1.20 and 2.01% w/w dye) separated macroscopically into distinct smectic (bottom) and isotropic solution (top) layers. For dyes A and C the phase equilibria are more complex, as discussed above. For dye B the calculated values for the cross-sectional area of these columns (1310 ( 51 Å2), of estimated diameter, ddc, 40.8 ( 0.8 Å, exceeds, by far, the dye molecular area of ∼151 Å2. Clearly, the individual dye stacks possess a multimolecular cross section, and not the unimolecular structure reported
J-Aggregates Formed by Aqueous Ionic Cyanine Dyes
J. Phys. Chem., Vol. 100, No. 6, 1996 2319 linearity of this plot indicates a concentration-invariant aggregate composition (size and anisotropy) throughout this region. The columnar hexagonal mesophase of dye B is thus stable at φd values ,0.48, the lower stability limit predicted theoretically by Taylor and Herzfeld34 for a linearly-aggregating system of rectangular rodlike aggregates. Further dilution to ∼2.5% w/w dye (defined by optical microscopy and 2H NMR), induced a transition from the columnar hexagonal to a columnar nematic mesophase (not represented in Figure 12).33 This transition is accompanied by a very slight diminution of the 2H NMR quadrupole splitting, most probably resulting from a decrease in the mesophase order parameter. Continued dilution, to ∼0.42% w/w dye at 26.0 °C (as determined by 2H NMR), brings about a second distinct (first-) order transition between the columnar nematic mesophase (J-aggregates) and isotropic dye solution (M h H-aggregates), via a narrow intervening twophase zone. 4. Discussion
Figure 10. Variation of the X-ray diffraction first-order Bragg spacing, d0, with reciprocal dye volume fraction, 1/φd, for aqueous cyanine dyes at 25 °C: (a, top) dye A in 2H2O; (b, bottom) dye D in 2H2O. The notations S and I represent smectic mesophase and isotropic dye solution, respectively.
previously for similar systems.22,23 Qualitatively similar results have also been obtained for the aqueous columnar aggregates of Nylomine Red A-2B 20 and the columnar J-aggregates of 1,1′-diethyl-2,2′-cyanine chloride.33 The precise molecular architecture of these highly-ordered 1-D columns is currently a matter of speculation.20 Postulated models include the uniaxial hollow cylinder and the biaxial rectangular stack (Figure 11). With the “hollow cylinder” model we expect the majority of ionic groups to reside on the outside of the column. It is difficult to envisage that a biaxial rectangular stack of width ca. 200 Å (from the column area) could pack into a hexagonal array with center-center separation of 100 Å (at the highest concentration) rather than forming an orthorhombic structure. Also, the swelling behavior (see below) indicates a constant aggregate structure, which again seems improbable with this model. Hence we prefer the hollow cylinder structure, despite our initial prejudice against this model. Upon dilution, the hexagonal mesophase of dye B swells progressively in two dimensions i.e., the dye stacks simply separate in directions normal to their long-axes. This is reflected in the log d0 versus log(1/φd) slope of 0.51 (Figure 12). Across this composition range, the center-to-center separation of the columns, a, increases monotonically from ∼94 Å at 24.80% w/w dye, to ∼282 Å at 3.04% w/w dye. Furthermore, the
The J-aggregate is the most commercially important dye assembly for the spectral sensitization of photographic silver halide grains. Their discovery dates back over half a century, though their detailed solution structures have remained the subject of some speculation and controversy. Here, the equilibrium supramolecular architectures of free-standing J-aggregates have been elucidated and quantified for the first time using a combination of nonintrusive experimental techniques. It has been established, unequivocally, that the solution J-aggregate state of several disparate ionic cyanine sensitizing dyes is largely liquid crystalline in nature. Indeed, the pioneering J-aggregation studies of the aqueous cationic sensitizing dye 1,1′-diethyl-2,2′-cyanine chloride, conducted by Scheibe et al.35 and Jelley6 some 60 years earlier, had intimated as much. The J-aggregate mesophase properties (structure, order, dimensions, and stability), however, are governed largely by the molecular structure of the dye, and Vide infra the short-range intermolecular interactions (electrostatic, steric, and van der Waals). For example, dilute liquid crystals possessing long-range translational smectic (dyes A, C, and D) or hexagonal periodicity (dye B) and long-range orientational nematic order (dye B) may all exhibit characteristic spectroscopic J-bands. The individual mesogenic J-aggregates (2-D monolayers or 1-D columns) must, by inference, be composed of many thousands or more of dye monomers, depending on the mesophase structure and concentration. Thus, typical association numbers of four molecules,4 estimated for the polymer-like J-aggregates of aqueous ionic cyanine dyes from mass action considerations of absorption spectra (M h J-band equilibrium), do not correspond to the number-molecular weight of the aggregate itself. Clearly the simple mass action model used does not apply to equilibria involving phase separation and protonation. Polarographic diffusion coefficient measurements36 have furnished more realistic J-aggregate association numbers equivalent to an assembly of hundreds of such repeating units. Conversely, the H-aggregate precursors of dyes A to D are comparatively small, probably anisometric, dye assemblies such as dimers, trimers, etc. They constitute both structurally and optically, isotropic fluid systems. The M h H h J-aggregate equilibria of aqueous ionic cyanine dyes may be considered similar, though not strictly analogous, to the monomer h micelle h mesophase equilibria of aqueous surfactant systems. Thus, for dilute smectic dye systems in particular, where extensive heterogeneous two-phase coexistence regions may exist, the J-aggregate mesophase may be isolated from dilute dye solution simply by centrifugation.5
2320 J. Phys. Chem., Vol. 100, No. 6, 1996
Harrison et al.
Figure 11. Postulated models for the 1-D dye columns (J-aggregates) of the aqueous nematic and hexagonal mesophases of dye B: (a, left) uniaxial hollow cylinder; (b, right) biaxial stack of rectangular cross section. A random orientation of biaxial dye stacks about the director, n, confers a uniaxial mesophase superstructure.
Figure 12. Variation of the X-ray diffraction first-order Bragg spacing, d0, with reciprocal dye volume fraction, 1/φd, for the aqueous hexagonal M mesophase of dye B at 25 °C.
In marked contast to aqueous amphiphilic systems such as surfactants,38 the driving force for dye self-association is believed to result, primarily, from short-range intermolecular attractive forces involving both σ and π electrons37 and not the hydrophobic effect. Thus, in a similar fashion to thermotropic mesogens,32 seemingly minor changes to the generic cyanine dye structure can have a profound effect on the number, type, and stability of the mesophases observed.33 The evolutionary phase behavior of these rigid, planar molecules also appears less complex than conventional micellar lyotropics38 and is seemingly not driven by changes in aggregate interfacial curvature.39 For example, no single aqueous chromonic (dye or drug) system exhibiting both a truly cylindrical hexagonal and a planar smectic mesophase has been identified experimentally, to date. Furthermore, isotropic cubic mesophases analogous to the amphiphilic I and V phases remain conspicuously absent from the phase diagrams of chromonic molecules. The phase progression of aqueous ionic cyanine dyes with concentration may be considered to result, simply, from disorderorder transitions between topologically invariant aggregate
structures, driven primarily by long-range interaggregate electrostatic and excluded-volume interactions. In reality, the 2-D monomolecular J-aggregate dye layers composing the smectic mesophase of dyes A, C, and D bear more than a passing resemblance to Kuhn’s9 proposed 2-D brickwork model for mica-adsorbed cyanine dyes (Figure 2A). In both cases, the planar molecules are parallel and close-packed edge-to-edge and plane-to-plane, with the in-plane molecular short-axis perpendicular to the layer plane, and neighboring molecular rows mutually staggered. The incongruous absorption wavelengths of the solution and AgX-adsorbed (0.2 and 0.3 µm AgBrI cubic and octahedral emulsions, respectively40) Jaggregates of dyes A, C, and D are, however, indicative of significantly different intermolecular correlations (slip angles, R) within the respective dye layers. For cyanine dye B, which forms aqueous columnar nematic and hexagonal mesophases, the precise molecular architecture of the one-dimensional dye columns (J-aggregates) remains equivocal. In direct contrast to the uni- and bimolecular dye stacks (Figure 2b) previously proposed for solution J-aggregates,9 the X-ray data presented here are consistent with columns of multimolecular cross section. Furthermore, 2H NMR and XRD results for dye B are also incompatible with the 1-D brickwork models proposed for the aqueous J-aggregates of 1,1′-diethyl-2,2′-cyanine chloride (PIC),9,42 where the columnar long-axis is coincident with the molecular long-axes (e.g. Figure 2A′). Irrespective of the detailed topology of their solution J-aggregates, both dye B and PIC, perhaps significantly, exhibited wavelength-invariant J-bands both in solution and adsorbed40,41 to AgBrI cubic and octahedral microcrystals, respectively. Recently we have become aware of further studies on the solution aggregates of PIC.43 Stegemeyer and Sto¨ckel have employed optical microscopy and UV-vis spectroscopy to demonstrate that the onset of J-aggregation is almost coincident with the formation of a nematic phase. (They use the term “Scheibe aggregates” rather than “J-aggregates” because of Scheibe’s many contributions to this area.) This is clearly in good agreement with our conclusions. It is necessary for many other J-aggregating dyes to be examined before this identity
J-Aggregates Formed by Aqueous Ionic Cyanine Dyes between J-aggregates and mesophases can be taken to be a general phenomenon. Experimental studies are continuing to elucidate the precise molecular ordering within the solution J-aggregates of both smectic and nematic/hexagonal mesophaseforming cyanine dyes. Acknowledgment. We thank Kodak Ltd. and the EPSRC for financial support (CASE Award) and the EPSRC Synchrotron Radiation Laboratory (Daresbury, U.K.) for the use of X-ray diffraction facilities. We are grateful to Professor Stegemeyer for the commumication of his results (ref 43) prior to publication. We also thank an anonymous referee for helpful comments. References and Notes (1) Sturmer, D. M.; Heseltine, D. W. In The Theory of the Photographic Process, 4th ed.; James, T. H., Ed.; Macmillan: New York, 1977; Chapter 8, and references therein. (2) Herz, A. H. The Theory of the Photographic Process, 4th ed.; James, T. H., Ed.; Macmillan: New York, 1977; Chapter 9 and references therein. (3) West, W.; Gilman, P. B. The Theory of the Photographic Process, 4th ed.; James, T. H., Ed.; Macmillan: New York, 1977; Chapter 10 and references therein. (4) Herz, A. H. AdV. Colloid Interface Sci. 1977, 8, 237. (5) Herz, A. H. Phot. Sci. Eng. 1974, 18, 323. (6) Jelley, E. E. Nature 1936, 138, 1009; 1937, 139, 631. (7) Spano, F. C.; Mukamel, S. Phys. ReV. A 1989, 40, 5783. (8) McRae, E. G.; Kasha, M. J. Chem. Phys. 1958, 28, 721. (9) Czikkely, V.; Fo¨rsertling, H. D.; Kuhn, H. Chem. Phys. Lett. 1970, 6, 11, 207. (10) Norland, K.; Ames, A.; Taylor, T. Photogr. Sci. Eng. 1970, 14, 296. (11) Smith, D. L. Photogr. Sci. Eng. 1972, 16, 329; 1974, 18, 309. (12) Nakatsu, K.; Yoshioka, H.; Nishigaki, S. Nippon Shashin Gakkaishi 1983, 46, 89, T9158. (13) Maskasky, J. E. Langmuir 1991, 7, 407. (14) Maskasky, J. E. J. Imaging Sci. 1991, 35, 29. (15) Emerson, E. S.; Conlin, M. A.; Rosenoff, A. E.; Norland, K. S.; Rodriguez, H.; Chin, D.; Bird, G. R. J. Phys.Chem. 1967, 71, 2396. (16) Bliznyuk, V. N.; Kirstein, S.; Mo¨hwald, H. J. Phys. Chem. 1993, 97, 569. (17) Kajikawa, K.; Takezoe, H.; Fukuda, A. Chem. Phys. Lett. 1993, 205, 225.
J. Phys. Chem., Vol. 100, No. 6, 1996 2321 (18) Wolthaus, L.; Schaper, A.; Mo¨bius, D. Chem. Phys. Lett. 1994, 225, 322. (19) Kawasaki, M.; Ishii, H. Chem. Lett. Jpn. 1994, 1079. (20) Tiddy, G. J. T.; Mateer, D. L.; Ormerod, A. P.; Harrison, W. J.; Edwards, D. J. Langmuir 1995, 11, 390. (21) Turner, J. E. Lyotropic Discotic Dye/Water Systems; Ph.D. Thesis, Leeds University, U.K., 1988. Attwood T. K.; Lydon J. E. Mol. Cryst. Liq. Cryst. 1984, 108, 345. Turner J. E.; Lydon J. E. Mol. Cryst. Liq. Cryst. Lett. 1988, 5, 93. (22) Lawrence, A. S. C. In Liquid Crystals; Brown, G. H., Ed.; Gordon & Breach: London, 1969, Vol. 1, p 1. (23) Hartshorne, N. H.; Woodard, G. D. Mol. Cryst. Liq. Cryst. 1973, 23, 343. (24) Halle, B.; Wennerstro¨m, H. J. Chem. Phys. 1981, 75, 1928. (25) Rendall, K.; Tiddy, G. J. T.; Trevethan, M. A. J. Chem. Soc., Faraday Trans. 1 1983, 79, 637. (26) Kustanovich, I.; Poupko, R.; Zimmermann, H.; Luz, Z.; Labes, M. M. J. Am. Chem. Soc. 1985, 107, 3501. (27) Herz, A. H. Photogr. Sci. Eng. 1974, 18, 207. (28) Goldfarb, D.; Labes, M. M.; Luz, Z.; Poupko, R. Mol. Cryst. Liq. Cryst. 1982, 87, 259. (29) Luzzati, V.; Mustacchi, H.; Skoulios, A.; Husson, F. Acta Crystallogr. 1960, 13, 660. (30) Luzzat, V. In Biological Membranes, Physical Fact and Function; Chapman, D., Ed.; Academic Press: London, 1968; p 71. (31) Lydon, J. E. Mol. Cryst. Liq. Cryst. Lett. 1980, 64, 19. (32) Gray, G. W.; Goodby, J. W. G. In Smectic Liquid Crystals Textures and Structures; Leonard Hill: London, 1984. (33) Harrison, W. J.; Mateer, D. L.; Tiddy, G. J. T. Unpublished data, 1993-1994. (34) Taylor, M. P.; Herzfeld, J. Langmuir 1990, 6, 911. (35) Scheibe, G.; Kandler, L.; Ecker, H. Natu¨ rwissenschaften 1937, 25, 75. (36) Hilson, P. J.; McKay, R. B. Trans. Faraday Soc. 1965, 61, 3711. (37) Hunter, C. A.; Sanders, J. K. M. J. Am. Chem. Soc. 1990, 112, 5525. (38) Tiddy, G. J. T. Phys. Rep. 1980, 57, 1. (39) Ke´kicheff, P.; Cabane, B. J. Phys. 1987, 48, 1571. (40) Dobles, T. R.; Cravatta, C. A. Eastman Kodak, unpublished data. (41) Muenter, A. A.; Brumbaugh, D. V.; Apolito, J.; Horn, L. A.; Spano, F. C.; Mukamel, S. J. Phys. Chem. 1992, 96, 2783. (42) Daltrozzo, E.; Scheibe, G.; Gschwind, K.; Haimerl, F. Photogr. Sci. Eng. 1974, 18, 441. (43) Stegemeyer, H.; Sto¨ckel, F. Ber. Bunsen-Ges. Phys. Chem., in press.
JP952532L
