Ultrafast Electron Crystallography. 2. Surface Adsorbates of Crystalline

4920

J. Phys. Chem. C 2007, 111, 4920-4938

Ultrafast Electron Crystallography. 2. Surface Adsorbates of Crystalline Fatty Acids and Phospholipids Marco T. Seidel, Songye Chen, and Ahmed H. Zewail* Physical Biology Center for Ultrafast Science and Technology, and Laboratory for Molecular Sciences, California Institute of Technology, Pasadena, California 91125 ReceiVed: NoVember 10, 2006

In this account, the second in this series, we report our detailed studies of the structures and dynamics of adsorbates of crystalline fatty acids and phospholipids, using ultrafast electron crystallography (UEC). These macromolecules serve as model systems for biomembranes and allow for systematic studies under controlled conditions. The systems investigated are arachidic (eicosanoic) acid and dimyristoyl phosphatidic acid (DMPA), deposited on a substrate by the Langmuir-Blodgett technique. We studied these systems in monolayer, bilayer, and multilayer structures, and under different conditions (pH, pressure, and temperature) and on different substrates (hydrophobic and hydrophilic). The subunit cell -CH2-CH2-CH2- distances were determined for all structures. For fatty acid samples, the structure is orthorhombic, and a0 and b0 have the values, depending on conditions, of 4.7-4.9 Å and 8.0-8.9 Å, respectively. The c0 value is 2.54 Å and for a given sample the accuracy is milliangstrom. Structural dynamics, after an ultrafast temperature jump in the underlying substrate, were studied by observing changes of the diffraction patterns: Bragg spot position, intensity, width, and the rocking curves. All structures exhibit a coherent anisotropic nonequilibrium expansion along the aliphatic chains, accompanied by transient structural ordering on the ultrafast time-scale. This wave-type, not diffusive, motion is followed by contraction and restructuring at longer times due to energy redistribution and diffusion. The transient behavior is entirely different from that reported here at equilibrium temperatures, in the range of 100-380 K. From these results, we are able to draw a general picture for the structural dynamics of amphiphilic chain molecules and elucidate the important role of nonequilibrium behavior at short times.

I. Introduction series,1

In the preceding paper, 1 in this we have discussed the principles of ultrafast electron crystallography (UEC) with applications to studies of surfaces and bulk crystals, such as silicon,2 GaAs,3 and gold.4 The combined determination of structures and dynamics on the relevant atomic-scale spatial and ultrafast temporal resolutions provides the unique approach for examining the nature of the atomic motions and structural changes. These changes are induced by a femtosecond laser excitation, which establishes a temperature jump. Basically, the carriers are initially created with a temperature that far exceeds that of the lattice, and, through electron-phonon coupling, the energy is transferred into the lattice giving rise to a temperature of tens of degrees and in a few picoseconds. As detailed in paper 1, this ultrafast heating provides an opportunity for studying the nonequilibrium structures at ultrashort times and the equilibration at longer times, when the material acquires incoherent, thermal heating. Such a temperature jump is ideally suited for studies of adsorbates, and the strong electron-matter interactions make this possible even for monolayers on the substrate. As with the substrate itself, we can study the nonequilibrium structural changes of the adsorbate. Earlier communications have focused on UEC studies of nanometer-scale water (adsorbate) on silicon (substrate)5 and self-assembled monolayers of organic molecules on gold. More recently, we reported preliminary results of UEC of larger assemblies of fatty acids6 and phospholipids.7 The * To whom correspondence should be addressed. E-mail: zewail@ caltech.edu.

dynamics of structural change, besides being interesting in their own right, may be of significance to biomembranes and their transport phenomena. In this paper, we provide a full account of our studies of fatty acid and phospholipid adsorbates on a hydrophobic or hydrophilic substrate. Their common structural features are the long aliphatic chains and amphiphilic character (see Figure 1). Such characteristics allow for controlled layer-by-layer deposition by means of the well-known Langmuir-Blodgett (LB) technique.8,9 Furthermore, LB films exhibit a high order of crystallinity, in comparison with self-assembly, and such interfaces with welldefined structures are ideal for examination with UEC. The length scale of a monolayer extends over ∼25 Å, and bi(multiple)-layers provide structures of 5 nm or more. The interfacial assemblies are unique in their physical properties, and the questions of interest are fundamental: upon heating, does the assembly involve collective nuclear motions or does it change by incoherent movements of atoms? Also, what is the nature of structural changes in the nonequilibrium regime, and to what extent does structural change reflect inhomogenity at the interface and near phase transitions of the assembly? Here, we will address these questions and present evidence for a coherent, nonequilibrium dynamics in the chains of the adsorbate, following the ultrafast temperature jump. The chains rearrange to give enhanced transient structural ordering. At longer times, contraction and restructuring follows with energy redistribution and diffusion being the dominant effects. Such a transient behavior is in vast contrast to structural changes we studied, and reported here, at thermal equilibrium, and would

10.1021/jp0674672 CCC: $37.00 © 2007 American Chemical Society Published on Web 03/13/2007

Ultrafast Electron Crystallography. 2

J. Phys. Chem. C, Vol. 111, No. 13, 2007 4921

Figure 1. Structure of crystalline adsorbates of fatty acids and phospholipids. (A) The molecular structure of arachidic acid and DMPA. (B) Schematic side view of the two adsorbates on a hydrogen-terminated silicon(111) substrate. For the phospholipid and fatty acid, we show the bridging through the ions (salt); for the fatty acid, depending on pH, both the neutral and salt forms are present in the film. (C) The orthorhombic CH2 subunit cell of aliphatic chains with lattice parameters a0, b0, and c0. The chain at the center was omitted for clarity.

remain unrevealed without the spatiotemporal resolutions of UEC. Also here, we present detailed studies of structural dynamics upon going from the two-dimensional case (monolayer and bilayer) to the three-dimensional case (multilayer), and the influence of the underlying substrate. The paper is outlined as follows. section II gives a brief introduction to the relevant work on LB films of fatty acids and phospholipids, and section III introduces the methodology: the UEC technique (A) and details for preparation of fatty acid and phospholipid interfaces (B). Section IV is devoted to results and discussions, with focus on the structures of fatty acid bilayers (A), the transition from bilayers to 8-layer interfaces (B), and the temperature dependencies (C). In section IV.D, we discuss the structural dynamics of fatty acid adsorbates, and in section IV.E, the mono- and bilayer structures and dynamics of phospholipids are presented. The emerging general picture, as viewed by UEC, together with some concluding remarks are given in section V. II. Layers of Fatty Acids and Phospholipids Since their introduction by Irving Langmuir and Katharine B. Blodgett in the 1930s,10,11,12 the LB methodology has proven powerful in the preparation of two-dimensional crystalline films and in many other applications of technology developments such as molecular electronics,13 biological sensors,14 and nonlinear optics.15 However, the selective preparation of adsorbates with well-defined structures by means of the LB technique is not a trivial task.16,17 There was and still is considerable effort in understanding the structure of even the “simplest” of all LB films, fatty acids and fatty acid salts.18-22 A plethora of powerful techniques (e.g., electron diffraction, neutron scattering, X-ray techniques, AFM, FTIR, Raman, and NMR) have been utilized to study the effect of the substrate,23-25 dipping conditions (pH value17,26,27 pressure,21,26,28-31 counterion,23,25 etc.) and the effect

of the surfactant itself on the resulting film structure. Of particular interest are the tilting of fatty acids29-31 structural differences in mono- and multilayers,19,32,33 epitaxal growth of fatty acids,34-36 thermal annealing and aging effects.35-37 Reviews are given by Schwartz,38 Peng et al.,39 and Roberts,8 and Petty.9 Phospholipids, which are the main building blocks of biomembranes, are even less understood because of their more complex structure. Layers of phospholipids often serve as model systems for studying membrane structures and properties, such as head group organization and hydration,40 phase transitions,41 interactions with membrane proteins,42,43 and proton diffusion.44 Their structures have been probed by a variety of techniques, including electron diffraction,45-47 X-ray diffraction,40,48 infrared spectroscopy,49,50 scanning tunneling microscopy,49,51 and atomic force microscopy.52 The thermal behavior of such films is of particular interest, especially when considering their two-dimensional structure on the substrate. Early studies of fatty acids have shown the onset of melting far below the bulk melting temperature and in multiple steps.53-59 AFM has elucidated the two-dimensional melting and the structures involved.60,61 Multiple step melting was also observed in X-ray diffraction studies.62 The transition from two-dimensional to three-dimensional melting in these fatty acid films, upon increasing the number of deposited layers, has been examined in a number of studies.63-65 In all of these investigations, however, direct knowledge of the dynamics is not possible because the applied techniques are not endowed with the ability to resolve structural changes on the relevant time scales, down to picoseconds. Femtosecond spectroscopies have provided valuable information about the dynamics on the picosecond time scale but could not determine the structure.66,67 Time-resolved X-ray diffraction studies showed a laser-induced ultrafast melting in fatty acid films above the

4922 J. Phys. Chem. C, Vol. 111, No. 13, 2007

Seidel et al.

Figure 2. Methodology of UEC. (A) Schematic view of the experiment, showing the heating and electron pulses, together with a typical diffraction frame. The geometry of the experiment is also indicated in (B). B k0 is the wave vector of the incident electron beam, B k is that of the diffracted beam defining Bragg spots; k ) |k B0| ) |k B| ) 1/λ. The direction of the incident electron beam is defined with respect to a specific crystal orientation (zone axis; here the [112h] direction of silicon) and φ is the angle between the projection of the electron beam on the sample surface and this zone axis. The incidence angle between the electron beam and the sample surface is θi. The momentum transfer b s ≡B k -B k0 in reciprocal space satisfies the Laue condition b s ) ha b* + kb B* + lc b*, with the integers h, k and l being the Miller indices and b a*, B b*, and b c* are the reciprocal lattice constants. The value of s ) |s b| is given by s ) 2k sin(θ), and tan(2θ) ) R/L, where R is the distance between the diffraction spot and the direct beam position on the screen and L is the distance between the sample and the screen, i.e., the camera length. When the diffracted angle θ is very small, sin (θ) ≈ θ and tan (2θ) ≈ 2θ, and the momentum transfer is simply s ) kR/L.

damage threshold.68 However, the results were not for monolayers and bilayers, and for the sample used (83 layers), a drawback is the inevitable destruction of the films. To our knowledge, the only reported nondestructive technique used to resolve the structure and dynamics is UEC. In what follows, we describe the methodology and results of these studies. III. Methodology A. Ultrafast Electron Crystallography. All experiments were carried out in our previously described UEC apparatus (see Figure 2; also paper 1). An amplified femtosecond laser

system is used to generate the ultrafast laser pump and electron probe pulses. The IR laser pulses (800 nm, ∼100 fs, ∼1 mJ, 1 kHz repetition rate) serve as the heating pulses and are focused onto the sample surface at an incidence angle of 30°. Part of the IR beam was frequency tripled (266 nm, ∼10 nJ) and focused onto a back-illuminated silver photocathode after an adjustable time delay∆t to generate the electron pulses (electron energy 30 keV, de Broglie wavelength λde Broglie ) 0.07 Å) via the photoelectric effect. A series of deflectors and apertures guide the electron beam to a grazing incidence, adjustable from θi ) 0 - 5°. The overlap between the electron and light pulses

Ultrafast Electron Crystallography. 2 is important for time resolution, and here, it is effectively several hundred microns, giving the geometrical resolution to be ∼1-2 ps. The diffraction patterns are recorded with a low-noise, image-intensified, charge-coupled device (CCD) camera assembly capable of single electron detection. Conceptually, the methodology is as follows. A heating IR laser pulse initiates a structural change, which is subsequently probed by the electron pulse. The electrons impinge the sample surface at the incidence angle θi and scatter to form the diffraction pattern in the far field. A series of diffraction patterns are recorded at specific times determined by the delay between the laser and electron pulses. When combined, these frames give the real-time change of the structure. The zero of time was determined in situ. The samples were mounted on a goniometer, allowing the positioning in three translational and two rotational degrees of freedom (precision in x, y, z is 10 µm and in θ, φ is 0.01°, respectively). The goniometer is equipped with a heating/cooling system, which provides control of the static temperature in the range from T ) 90 to 500 K with a precision of (2 K. All experiments were carried out under ultrahigh vacuum conditions (∼10-10 Torr). It is important to emphasize that the LB films are extremely stable under these conditions, and we observed no change or deterioration in the quality of the diffraction patterns due to the pulsed low-electron flux irradiation. However, prolonged laser irradiation and increased static sample temperature can lead to loss of the diffraction patterns, as described below. B. Preparation of Layers by LB Deposition. 1. Fatty Acid Samples. The substrate for all investigated samples was hydrophobic silicon(111), which was hydrogen-terminated prior to deposition, as discussed below. The deposition of the fatty acids is in the so-called Y geometry (i.e., head-to-head and tailto-tail). The first layer is attached to the surface by its hydrophobic tail. Since the film is always terminated by a hydrophobic tail, only even numbered multilayers can be obtained for a hydrophobic substrate. In order to achieve good film quality, it was mandatory to prepare the LB films immediately after hydrogen termination, because the hydrophobic silicon surface is unstable in air. Arachidic acid (eicosanoic acid, C19H39COOH) was purchased from Aldrich and used without further purification (see Figure 1A). The fatty acids were spread from a chloroform solution. Prior to compression to the final deposition pressure of π ) 29 mN/m, we allowed 20 min for complete evaporation of the solvent. The deposition pressure was chosen to be considerably larger compared to that used in our previous study, of which some results are reported as well, because deposition of fatty acid films with more than two layers failed at π ) 10 mN/m. For all samples, the deposition took place in a NIMA LB trough at a dipping speed of 1 mm/min. The used subphase was an aqueous CaCl2 solution (c[CaCl2] ) 0.5 mmol/L, Millipore Water), which was adjusted to the desired pH-value with aqueous diluted NaOH. To study the influence of film thickness and dipping conditions, we prepared multiple samples of LB films at different pH values. Here, we present selected samples of 2-, 4-, and 8-layer films at pH ) 6.4, ∼7, and 9. At the pH values of 6.4 and 7, about 60-80% of the arachidic acid is deposited as the corresponding calcium salt, whereas at pH ) 9, 100% of the arachidic acid is deposited as the salt.69 To ensure a high sample quality, the isotherms for the precursor Langmuir film and the transfer ratios upon deposition were routinely recorded. Furthermore, the morphology of the


Figure 3. Subunit cell orientation and directions of observation and dipping. (A) For most cases here, electron incidence was either along the [112h] direction of the hydrogen-terminated silicon(111) surface, corresponding to φ ) 0°, or along the [1h10] direction (φ ) 90°). The orientation of the fatty acid subunit cell on the surface and the alignment of the b0 axis along the [1h10] direction of silicon are indicated. Note the near match between the adsorbates and substrate along b0. (B) Deposition of a LB-film results in two distinguishable directions; parallel (|) and perpendicular (⊥) to the dipping direction.

Langmuir film at the air-water interface was monitored with a Brewster angle microscope. For our dipping conditions, the Langmuir films were very uniform and did not show any (macroscopic) holes and/or multilayer islands. In general, we recorded diffraction patterns of the LB films along two directions, which are perpendicular to each other. These correspond to the [112h] and [1h10] zone axes of the underlying silicon substrate (see Figure 3A), as evidenced from the diffraction patterns at high θi’s. We denote these two directions with the azimuthal angles φ ) 0° and 90°, respectively. Note, however, that we also have the ability to record diffraction patterns around these φ values in a range of at least (30°. The deposition of the LB films was along either of these fundamental axes and accordingly we can distinguish a parallel (|) or a perpendicular (⊥) dipping direction (see Figure 3 B). However, it is the silicon surface, rather than the dipping direction, which determines the structure and orientation of the fatty acid subunit cell on the substrate, as discussed below.

4924 J. Phys. Chem. C, Vol. 111, No. 13, 2007 2. Phospholipid Samples. We prepared monolayers of dimyristoyl phosphatidic acid (DMPA) on a hydrophilic surface; the polar head groups are involved in bonding to the substrate. For the bilayers of DMPA, a hydrophobic substrate was utilized, in a similar fashion to the fatty acid samples. We used a hydrophilic oxide-terminated silicon(111) surface for the deposition of the monolayer and hydrophobic hydrogen-terminated silicon(111) surfaces for the bilayer. The preparation of these surfaces is given below. As in the case of the fatty acids, the deposition is in the Y geometry. 1,2-Dimyristoyl-sn-glycero-3-phosphate monosodium salt (dimyristoyl phosphatidic acid, DMPA) was purchased from Aldrich and used without further purification (see Figure 1A). The lipids were spread from a chloroform/methanol (3:1) solution, and the subphase used was Millipore water containing sodium ions at pH ) 5.5. The deposition took place in a NIMA LB trough at a dipping pressure of π ) 29 mN/m and a dipping speed of 1 mm/min. Film quality measurements were routinely recorded, as described above. 3. Silicon Functionalization. The silicon(111) surfaces used were functionalized according to standard methods. Hydrophilic silicon surfaces were prepared just before deposition by cleaning and oxidation with an RCA-1 solution. For the hydrophobic surfaces, subsequent etching of the oxide surface was made in a 40% NH4F solution for 15-20 min in order to obtain the Si(111):H surface. IV. Results and Discussion A. Fatty Acid Bilayers: Structures. In general, fatty acid molecules in LB films arrange in the following fashion: the long aliphatic chains pack with their axes parallel to each other but not necessarily perpendicular to the substrate surface, as discussed in several articles,20,21 reviews,38,39 and monographs.8,9,70 The C2H4 units form a sub-lattice with different possible symmetries and variations depending on the relative displacement of adjacent molecules. Moreover, depending on the nature of the substrate (hydrophobic or hydrophilic), the pH, and methods of interfacial preparation, the structure can vary. Our observed diffraction patterns are directly the result of the C2H4 subunit cell because of the covered s range. The magnitude of the momentum transfer vector (s of Figure 2) between the incident electron and the scattered electron is given by s ) kR/L where k ) 1/λdeBroglie, R is the distance between a diffraction spot and the direct beam position on the screen, and L is the distance between the sample and the screen (so-called camera length; see Figure 2). Different arrangements of the subunit cell can be uniquely distinguished by the measured lattice parameters and systematic absentees of certain Bragg spots. In this way we are able to establish the subunit cell structure, the in-plane orientation of the subunit cell with respect to the silicon surface and the out-of-plane direction of the aliphatic chains (i.e., inclination). Figure 1C depicts a schematic view of an orthorhombic C2H4 subunit cell; the depicted structure is from our diffraction analyses given below. The static diffraction patterns of bilayer films are shown in Figure 4A-C. The panels are for LB films prepared at different conditions but all observed by UEC (i.e., recorded with ultrashort electron pulses, but without the laser heating pulse). The diffraction patterns are recorded at φ ) 0° and 90°, at low incidence angles θi < 1°; (see Figure 2). The diffraction patterns are composed of Bragg spots, which are well resolved. This is the result of a high quality crystalline structure of the bilayers. From the positions of the Bragg spots (including systematic absentees), the subunit cells were determined to be orthorhom-

Seidel et al.

Figure 4. Experimentally observed static diffraction patterns of the calcium arachidate LB bilayers at T ) 295 K. The indexing for an orthorhombic R(001) structure of the fatty acids is indicated; Bragg spots not indexed result from the underlying silicon substrate as shown by the rocking curves (Figure 5). The top panels show the diffraction patterns observed at φ ) 0°, while the bottom ones show diffraction patterns observed at φ ) 90°. The s range for all diffraction patterns is from -0.6 to +0.6 Å-1 in the horizontal direction and from 0 to +1.2 Å-1 in the vertical direction. (A) Diffraction patterns for the bilayer deposited at pH ) 9 and π ) 10 mN/m. The subunit cell parameters obtained from the position of the Bragg spots and according to the indexing indicated are a0 ) 4.7 Å, b0 ) 8.0 Å, and c0 ) 2.54 Å. Dipping occurred along the [112h] direction of silicon. The incidence angle is θi ) 0.8° for φ ) 0° and φ ) 90°. (B) Diffraction patterns for the bilayer deposited at pH ) 6.4 and π ) 29 mN/m. The obtained subunit cell parameters are a0 ) 4.9 Å, b0 ) 8.3 Å, and c0 ) 2.59 Å. Dipping occurred along the [1h10] direction of silicon. The incidence angle is θi ) 0.8° for φ ) 0° and θi ) 0.9° for φ ) 90°. (C) Diffraction patterns for the bilayer deposited at pH ∼ 7 and π ) 29 mN/m. The subunit cell parameters obtained are a0 ) 4.7 Å, b0 ) 8.5 Å, and c0 ) 2.57 Å. Dipping occurred along the [1h10] direction of silicon. The incidence angle is θi ) 0.4° for φ ) 0° and φ ) 90°.

bic, with short notation of R(001). Accordingly, “R” represents the orthorhombic symmetry and (001) is for denoting the (a b) plane of the C2H4 subunit cell which is parallel to the substrate silicon(111) surface. It follows that the aliphatic chains pack with their chain axes parallel to each other and perpendicular to the substrate surface. In Figure 4, the indexing of Bragg spots are in accord with reciprocal space indices (see paper 1). Basically, the vertical s spacing gives the c0 values while side diffractions give a0 and b0 for the two φ’s, respectively. We note two points. First, the bright spots above the shadow edge are due to the substrate silicon. This was confirmed from the θi dependence (rocking curve) which shows large changes for these bright spots but not for the bilayer adsorbate (Figure 5 depicts this behavior for the substrate and for the bilayer deposited at π ) 10 mN/m and pH ) 9, and whose static diffraction patterns are depicted in Figure 4A). Second, in s space the vertical distance from the (101) or (1h01) spots to the (002) spot gives c0, and twice such distance is where the direct beam is located (below the shadow edge). The determined lattice parameters of the different bilayer structures are summarized in Table 1. In order to compare with the diffraction from a 3-D crystal, we have simulated the diffraction patterns of an orthorhombic crystal for the two directions studied here (see Figure 6). Although the intensities of the calculated Bragg spots (2h02) and (202) do not reproduce those measured, because the calculation was made for an infinite crystal and the underlying substrate and the headgroups are not taken



Figure 6. Calculated diffraction patterns of an orthorhombic, infinite crystal of aliphatic carbon chains observed along the [100] and [010] direction, respectively. The subunit cell lattice parameters are a0 ) 4.93 Å, b0 ) 7.4 Å, and c0 ) 2.534 Å. The space group is Vh16 (Pnam). Indeed the patterns show the characteristic spots, but with differences in intensity, due to the infinite array of chains in the crystal (see text).

TABLE 1: Structural Dimensions for the Bilayers of Fatty Acids Studied on the Hydrophobic Substrate of Silicon(111) Surface

Figure 5. Experimental rocking curve for the z-component of the scattering vector s (from +0.25 to +3.0 Å-1) vs incidence angle θi (0 to 5.85°) at φ ) 90°, showing the (002) peak (see diffraction near s ) 0.8 Å-1) of the bilayer and (00) peak of the Si(111) surface (at higher s values). Note that for small θi the diffraction is that of the bilayer (see Figure 4A).

into account, the absentees of first-order diffraction, spacings, and symmetry of patterns are all consistent with observations made for the layers. The calculations were not scaled with the s decay of diffraction. The lattice parameters c0, which result

a0/Å b0/Å c0/Å

pH ) 9, π ) 10 mN/m (Figure 4 A)

pH ) 6.4, π ) 29 mN/m (Figure 4 B)

pH ∼ 7, π ) 29 mN/m (Figure 4 C)

4.7 8.0 2.54

4.9 8.3 2.59

4.7 8.5 2.57

from the vertical s spacing of the Bragg spots, have the same magnitude for all three samples. They agree well with the theoretical value of 2.54 Å, which follows from a simple geometrical consideration for an aliphatic chain with bond distances rc-c ) 1.53 Å and angle ∠c-c-c ) 112°. The c0 value represents the distance between CH2 planes (see Table 1), and it holds true for the 4- and 8-layer samples, as shown below.

4926 J. Phys. Chem. C, Vol. 111, No. 13, 2007 The determined a0 and b0 values vary for different conditions and range for a0 from 4.7 to 4.9 Å and for b0 from 8.0 to 8.5 Å. These values differ somewhat from the theoretical values of a0 ) 4.96 Å and b0 ) 7.4 Å,70 which are estimated, by purely geometrical considerations, for the ideal case of closest packed arrays of infinite aliphatic chains. The difference can be explained by the fact that the theoretical values do not take into account the carboxylic end groups of fatty acids. Moreover the substrate and the deposition conditions all have an important role in the order at the interface, as shown below. Although the parameter a0 has a similar value for all bilayer samples, b0 is slightly larger for the samples prepared at lower pH values (pH ) 6.4 and ∼7 vs pH ) 9) and higher pressures (π ) 29 mN/m vs π ) 10 mN/m). Though the pressure has little effect on the molecular packing in this range, as observed in the π-area isotherm (not shown), it has been observed that the pH value has a rather strong influence on the overall packing of the LB film. An increase of the pH value is believed to pull the head groups closer together and support a nontilted LBfilm structure. In the sample deposited at pH ) 9, virtually all fatty acid molecules are in their salt form, which leads to a slightly closer packing compared to the deposited structures at lower pH values, where only 60-80% of the fatty acids are in their salt form.69 By comparing panels A and B in Figure 4, we notice that the orthorhombic subunit cell is always aligned with its b0 axis along the [1h10] direction of the silicon(111) surface, and thus, the a0 axis is along the [112h] direction (regardless of the dipping direction). This influence of the substrate is expected and has been noted for hydrogen-terminated silicon(111).71 In this context, it is worth noting that the structure of the precursor Langmuir film (i.e., the monolayer on water) has relatively less influence on the structure of deposited LB films34,39,72 The influence of the substrate explains the orientational ordering of the adsorbate on the substrate (see Figure 3). The substrate values of the length of one repetitive unit along the [1h10] and [112h] direction of silicon, which are 3.84 and 6.65 Å in real space, respectively, are to be compared with b0 and a0 of the adsorbate layer. Two repetitive units along the [1h10] direction correspond to 7.68 Å, a value which is close to the observed b0 ) 8.0 Å (and 8.3 Å). The length mismatch is only 4% (and 8%). If b0 was oriented along the [112h] direction, then, because the distance of repetitive units is 6.65 Å, the mismatch would be 20-25%. The disagreement between a0 of the orthorhombic subunit cell and the lattice parameters of silicon is >20% in either orientation. It is therefore reasonable to explain, because of this length correlation, the b0 axis alignment along the [1h10] direction of the substrate, and thus, the orientation is independent of the dipping direction and precursor Langmuir structure. The orientation is also sensitive to substrate structure. In Figure 4C, the observed diffraction patterns form no “perfect” rectangular array of Bragg spots and the diffraction patterns in the two distinct directions of φ ) 0° and 90° are similar and differ only in intensity and sharpness of the spots. This is in stark contrast to the distinct diffraction patterns in the two directions for the samples shown in Figure 4 A and B and the predicted patterns in Figure 6. Close inspection of the diffraction patterns in Figure 4 C reveals that they result from a superposition of the two diffraction patterns (φ ) 0° and 90°) displayed in Figure 6. This loss of preferred orientation is consistent with the fact that the diffraction patterns do not change their general form upon rotation in a range of at least φ ) (30° in both directions. In this particular sample the absence of the alignment

Seidel et al.

Figure 7. Diffraction patterns obtained for the calcium arachidate LBfilms of 2, 4, and 8 layers deposited at pH ) 6.4 and π ) 29 mN/m (from left to right). The top panels show the diffraction patterns observed at φ ) 0°, while the bottom ones show the patterns observed at φ ) 90°. Deposition occurred along the [1h10] direction of silicon. The indexing for the bilayer is for an orthorhombic R(001) structure of the fatty acid sublattice, and is the same for the 4- and 8-layer sample (omitted for clarity). The incidence angles for φ ) 0° are θi ) 0.8, 0.8, 0.5° for 2, 4, and 8 layers, respectively, and for φ ) 90° they are θi ) 0.9, 0.4, and 0.4°. The s range for all diffraction patterns is from -0.6 to +0.6 Å-1 in the horizontal direction and from 0 to +1.2 Å-1 in the vertical direction.

is believed to originate from a partial oxidization of the surface prior to dipping; hydrogen-terminated silicon surfaces are known to be easily oxidized in air. We should note that this has the advantage of simultaneously extracting a0 and b0 from one diffraction pattern, and we used this feature to study the static temperature dependencies of the fatty acid bilayer (see section IV.C). In summary, calcium arachidate bilayers form orthorhombic LB films on the hydrophobic silicon(111) substrate, with the fatty acid chains aligned perpendicular to the surface. Samples prepared at higher pH values showed a closer packing of the fatty acid. The substrate surface has a large influence on orientation, and in this case, the b0 axis is along the [1h10] direction of the hydrogen-terminated silicon (111) surface. The a0 axis is along the [112h] direction, and we determined these values to be a0 ) 4.7 Å, b0 ) 8.0 Å, and c0 ) 2.54 Å at pH ) 9; a0 ) 4.9 Å, b0 ) 8.3 Å, and c0 ) 2.59 Å at pH ) 6.4; a0 ) 4.7 Å, b0 ) 8.5 Å, and c0 ) 2.57 Å at pH ) ∼7. B. Fatty Acid Multilayers: Structures. The static diffraction patterns recorded at room temperature for 2-, 4-, and 8-layer samples deposited at pH ) 6.4 and π ) 29 mN/m are shown in Figure 7. The upper panels show the diffraction patterns observed along the silicon [112h] direction (φ ) 0°, ⊥), and the lower panels show the patterns along the silicon [1h10] direction (φ ) 90°, |). All diffraction patterns take the form of rectangular arrays of spots and are indexed from the positions of the Bragg spots, as shown for the bilayer (see Figure 4B and also Figure 6). The indexing is the same for the 4- and 8-layer samples but omitted in the figure for clarity. The fatty acid molecules arrange in an orthorhombic R(001) packing with the subunit cells b0 axis aligned along the silicon [1h10] direction for all samples independent of layer thickness. The extracted subunit cell parameters are summarized in Table 2. Within the error bars, the 2- and 4-layer samples have the same lattice parameters, possibly because of the strong influence of the underlying substrate. With more deposited layers, this influence becomes weaker and results in the slight expansion of b0 in the 8-layer sample by ∼8% compared to the 2-layer



Figure 8. Inclined diffraction patterns for 8-layer samples at φ ) 0°. (A) The 8-layer sample deposited at pH ) 6.4 and π ) 29 mN/m shows a partial inclination of ∼20° to the left (dotted line), which vanishes after laser irradiation. The incidence angle is θi ) 0.4°. The s range is from -0.6 to +0.6 Å-1 in the horizontal direction and from 0 to +1.2 Å-1 in the vertical direction. (B) The 8-layer sample deposited at pH ∼ 7 and π ) 29 mN/m shows a partial inclination of ∼10° to the right, which for this sample is stable with laser irradiation. The incidence angle is θi ) 0.5°. The s range is from -0.6 to +0.6 Å-1 in the horizontal direction and from 0 to +1.2 Å-1 in the vertical direction.

TABLE 2: Structural Dimensions for the 2, 4, and 8 Layers Studied on the Hydrophobic Substrate of Silicon(111) Surface: pH ) 6.4 and π ) 29 mN/m a0/Å b0/Å c0/Å

2 layer

4 layer

8 layer

4.86 8.28 2.59

4.87 8.33 2.54

4.85 8.91 2.54

sample; a0 remains unaltered in the 8-layer sample. Because b0 is the direction of better match (stronger interaction) with the substrate the change is larger in this direction as layers become further away from the surface. An increase of “defects” and holes, when multiple layers are deposited,39 can also lead to a relaxation of the observed film structure. The Bragg spots are not as sharp for the 8-layer sample as they are for the 2- and 4-layer ones, and this behavior is consistent with some loss of order of crystallinity to randomly oriented crystallites, as has been observed for behenic acid multilayers.35,36 However, the aliphatic chains are still aligned nearly perpendicular to the sample surface and keep their stacked arrangement. Before laser irradiation, the 8-layer sample gives the diffraction pattern shown in Figure 8A for φ ) 0°. Upon laser annealing, this sample recovers the diffraction pattern shown in Figure 8 and is that of Figure 7. The pattern before annealing is basically characteristic of the film but has an additional “inclined line” (by ∼20°), reflecting partial inclination of the chains; the annealing leads to the disappearance of such a line, and the structure becomes that of a perpendicular chain geometry. However, another 8-layer LB film prepared at pH ∼ 7 and π ) 29 mN/m, although showed clearly the partial inclination, it did not change upon laser irradiation, as is apparent in the static diffraction pattern shown in Figure 8B, which shows the persistence of the inclined feature. Two parallel lines with a tilt angle of ∼10° to the right were observed parallel and perpendicular to the dipping direction. The spacing between the inclined lines along the tilted direction gives again c0 ) 2.54 Å, consistent with results to all samples studied. Besides the smeared lines, some Bragg spots are apparent, most notably the (002) reflection; however, lateral order could not be extracted. This change into an inclined chain structure is consistent with results showing that LB films composed of a few layers of fatty acid salts have an upright arrangement, whereas multilayers possess an inclined carbon chain geometry.39 The panel of Figure 8B is of value for dynamical studies, as discussed below. In summary, the static structures of 2-, 4-, and 8-layer samples of calcium arachidate are rather similar, though inclination of

Figure 9. Static temperature-dependent diffraction patterns for the calcium arachidate bilayer deposited at pH ∼ 7 and π ) 29 mN/m at T ) 100, 295, 333 and 370 K. The patterns are observed perpendicular at φ ) 0°and θi ) 0.4°. The indexation for 100 K is for an orthorhombic R(001) symmetry, and is the same for 295 and 333 K. The diffraction pattern at 370 K shows virtually no structure. The s range for all diffraction patterns is from -0.6 to +0.6 Å-1 in the horizontal direction and from 0 to +1.2 Å-1 in the vertical direction. Note the increased diffusiveness as the temperature increases.

the fatty acids together with an increase in disorder begins to show in the 8-layer samples. The influences of substrate orientation and interaction decrease as more layers are formed in the structure. C. Fatty Acid Bilayers: Temperature Dependence. Following characterization of structures, we studied the effect of substrate temperature. For the bilayer sample deposited at π ) 29 mN/m and pH ∼7, diffraction patterns at φ ) 0° were recorded at different temperatures in the range from T ) 100 to 380 K; representative examples of diffraction patterns are shown in Figure 9 for T ) 100, 295, 333, and 370 K. The patterns are indexed as described above and displayed on the T ) 100 K pattern. The diffraction pattern shows Bragg spots and is very clear at 100 K. The patterns become more diffuse with increasing temperature; the intensity of Bragg spots decreases while the background gains intensity. At 370 K, just below the melting temperature of arachidic acid salt LB films of ∼383 K,57 the characteristic features of spots are barely visible in the patterns. This shows that the temperature increase is responsible for the enhanced thermal and inhomogeneous disorder, leading ultimately to a total randomness and breakdown of the bilayer structure at the melting point. The relative (integrated) intensity change I/I0 of the (002) Bragg spot as a function of temperature is shown in Figure 10A. An intensity drop is expected by the Debye-Waller effect due to increased thermal motions of the scattering atoms with increasing temperature. These motions have direct influence on structural factors (see paper 1). However, it is remarkable that the intensity drop shown in Figure 10A features certain temperature points at which the slope abruptly changes (shown in the inset). Such a thermal behavior is different from conventional melting of isotropic bulk materials and could be related to phase transitions, as observed with AFM,60,61 of twodimensional, anisotropic LB films. In diffraction, besides the Debye-Waller intensity drop, one expects a sharp decrease in intensity if a phase transition is present, such as in conventional melting. However, this transition is only sharp if the layers are thick enough to define a threedimensional collective behavior. For two-dimensional melting,


Seidel et al.

Figure 11. Schematic view of the static thermal behavior for changes of a bilayer with temperature increase. The fatty acid structure gets more and more disordered along the carbon chains as well as in the lateral plane. Furthermore, we observe lateral expansion of the subunit cell and loss of Bragg spot intensity. The process is partially reversible as long as the structure does not reach the main melting transition, which results in the collapse of the headgroup region (see text).

Figure 10. Static temperature dependence of diffraction intensity and subunit cell dimensions (Å) for the calcium arachidate bilayer deposited at pH ∼ 7 and π ) 29 mN/m. (A) Relative intensity change I/I0 of the (002) spot in the temperature range from T ) 140 to 380 K observed at φ ) 0° and θi ) 0.4°. The red line is a guide to the eye. The inset emphasizes the abrupt slope changes in a plot of ln(I/I0) against temperature in the range from 280 to 380 K. (B) Change of the subunit cell dimensions a0, b0, and c0 observed at θi ) 0.4°. The black symbols represent cell parameters observed at φ ) 0°, whereas the open symbols denote cell parameters observed at φ ) 0° after cooling down from T ) 380 K. The green symbols stand for the subunit cell parameters obtained at φ ) 90° from T ) 95 to 290 K.

pre-transitional disordering in the form of thermally induced random tilt and/or bending of the aliphatic chains57 erodes the sharpness and renders the transition to occur over a wider temperature range. This picture of disorder involves several steps, first disordering of the hydrocarbon tails and then the breakdown of the headgroup region, the latter occurs at the main melting point.53,54 Raman studies support the presence of pretransitional regions.73 The effect of static temperature on the subunit cell parameters is shown in Figure 10B. With increasing temperature, the film expands along a0 and b0, first slowly then more rapidly with an onset of ∼50 K below the bulk melting point. A similar behavior has been reported for a fatty acid salt LB film;58 however, the expansion reported here is significantly more pronounced. It should be noted that the guide for the eye (red line) has the same shape for a0 and b0 but is only shifted along the y axis, indicating a similar expansion in a0 and b0 of ∼25% in the range from 100 to 380 K.

The expansion of the subunit cell in the plane parallel to the substrate surface is fully reversible as long as the sample is heated to temperatures below or near the bulk melting point and for a short time (minutes). In Figure 10B, the open circles and squares indicate that the initial cell parameters are retrieved after cooling down. However, the intensity of the Bragg spots recovers only partially (not shown), indicating some loss of crystalline order in the LB film. Within our accuracy, no expansion is observed along c0 (i.e., along the fatty acid chains). As shown below, this static picture is very different from that of the dynamical expansion observed under nonequilibrium conditions. In summary, with increasing temperature, we observe a pronounced loss of Bragg spot intensity, but with behavior much different from what is expected from the Debye-Waller effect for a bulk material. Rather, the behavior reflects the twodimensional pre-transitional structural changes. We also observe an in-plane expansion of the subunit cell of the chains, together with increased lateral (in-plane) disorder, as evidenced in the buildup of background scattering. At the same time, the degree of order in the carbon chains decreases. These processes are partially reversible as long as the temperature is below the main melting point. Beyond the main melting point, the headgroup region breaks down and the crystalline order of the adsorbate is lost. A structural picture is depicted in Figure 11. D. Structural Dynamics: Fatty Acids. To study the dynamics of the surface adsorbates of crystalline fatty acids, the diffraction patterns were followed as a function of time after ultrafast laser-pulse heating of the substrate. The fatty acids have no absorption resonance at the excitation wavelength of 800 nm and thus appear transparent, with the energy being absorbed solely by the silicon substrate. The mechanism of heating is discussed in paper 1. Following the initial formation of electron hole pairs, the energy is transferred in a few picoseconds to the lattice through electron-phonon coupling, thereby virtually creating a temperature jump in the silicon substrate. Through coupling of the substrate to the fatty acid molecules the energy is also transferred into the adsorbate, where unique structural changes are induced at far-from-equilibrium temperatures. At negative time delays, i.e., when the electrons arrive before the laser pulses, the observed diffraction frames are the same as those obtained at steady-state (i.e., the static diffraction patterns). The frames at negative times serve as reference for the subsequent changes with time. At positive time delays, we probe the structural changes of the adsorbate as a result of the



Figure 12. Full analysis of the (002) Bragg spot of the 4-layer calcium arachidate sample deposited at pH ) 6.4 and π ) 29 mN/m for φ ) 90° and θi ) 0.4°. (A) Representative Voigt fit of the normalized (002) Bragg spot in the vertical direction at t ) 1000 ps, after background removal. (B) The momentum transfer change ∆s in the vertical direction as a function of time. (C) Relative intensity change I/I0 as a function of time. (D) Full width at half-maximum (fwhm) for the (002) Bragg spot as a function of time. Note the (almost) mirror image of the behavior in (C) and (D).

temperature jump in the underlying substrate. For every time frame, each Bragg spot was fitted with a Voigt line shape function in the vertical direction after careful removal of the background, which results mainly from incoherent, inelastic, and other scatterings. This fitting procedure provides with precision the center position and the width as a function of time. The determination of the center position proved to be insensitive to background subtraction, because of the large intensity of diffraction spots relative to the background; for the line width, accurate background removal and good signal-to-noise ratio determine the accuracy of the values obtained. The intensities of Bragg spots were obtained by gating with a window of pixels on the CCD and integrate it to give the changes with time; the gate remains the same for all frames. In Figure 12, we show examples of the detailed analysis of the (002) Bragg spot for the 4-layer sample (φ ) 90°). As depicted in Figure 12A, for a delay time as long as t ) 1000 ps the fit of the Bragg spot with a Voigt line shape function in the vertical direction is very good; the same is true for all other time delays. From such excellent data sets, the exact center position change, the relative integrated intensity, and line width at half-maximum were obtained as a function of time (shown in Figure 12B-D, respectively). We observed similar trends for all Bragg spots in each sample under investigation. Note, however, that for the data sets with less than optimal quality

samples the line width changes with time were sometimes obscured by the noise. 1. Atomic Motions of the Chains. Before considering the evolution with time, it is useful to define the coherence lengths involved and s ranges of diffraction. For 4 layers (see Figure 7) and 2 layers (see Figure 7 and Figure 4A) depending on film preparation and condition, the spot size provides information on coherence length parallel and perpendicular to the surface and also on inhomogenity. For example, the horizontal width of the observed spots in Figures 4 and 7 is ∆s ∼ 0.06 Å-1, indicating a coherence length of a nanometer scale. On the other hand, the vertical width reflects an essential feature of 2-D/3-D diffraction. In 2-D, the rods of reciprocal space (paper 1) are continuous and the intersection with Ewald’s sphere can result in elongated spots, especially at small incident angles. In this case, the rocking curve will not display the spot structure characteristics of different orders, when the incident angle is changed, for the 3-D case. Basically, as the system changes from 2-D to 3-D, the rods are modulated by the vertical lattice spacing s value, and when penetration is sufficient, the rods become spots. This behavior can clearly be seen in the rocking curve, when θi changes (see Figure 5). For the substrate, the higher orders of diffraction were observed, but for the bilayer film, the intensity remains independent of θi in the range indicated. Thus,


Figure 13. The time-dependent diffraction of the bilayer deposited at pH ) 9 and π ) 10 mN/m. The diffraction frame difference at various time delays obtained for the (002) Bragg spot at θi ) 0.8Ε and φ ) 0°. Note the absence of diffraction in the frame difference at negative time relative to the static diffraction; the appearance at 1 ps, and fading again at long times. The nonequilibrium change at short time and restructuring at long time is evident in the frames.

the vertical width of, e.g., the (002) spot (Figure 4 A), reflects this feature of 2-D systems and also the inhomogenity in the distribution of the -CH2-CH2-CH2- distances as a result of film packing and condition. This width, relative to the s value of a -CH2-CH2-CH2- chain (∼0.78 Å-1), gives at most a ∼20% distribution of distances. Finally, we note that, even for the substrate, the rocking curve provides the correct lattice spacing only from higher order diffraction spots (large θi’s), which have small widths, and therefore, care has to be taken at small θi’s when considering changes in ∆s and the associated larger uncertainty (see paper 1). The most pronounced dynamical features of the diffraction patterns after the ultrafast temperature jump are as follows: the collective downward movement of the Bragg spots and subsequent recovery at longer times (see Figure 12B), the unusual intensity changes (Figure 12C), and width changes (Figure 12D) as a function of time. A downshift of Bragg spots in reciprocal space corresponds to an expansion of the orthorhombic R(001) subunit cell along c0 in real space. There is no measurable change in any of the in-plane directions, that is along a0 and b0. Note that this is in contrast to the observations made in the steady-state regime, consistent with the nonthermal behavior of the LB films on the ultrafast time scale, as discussed below. A typical change of the (002) spot for the bilayer deposited at π ) 10 mN/m and pH ) 9 as a function of time is shown in the frames of Figure 13. As it should, the diffraction difference frame at negative time delay (-20 ps) does not show any features; that is, there is no change before time zero. Immediately after the heating pulse (0 and 1 ps), an intensity loss of the Bragg spots is observed, as dark spots appear in the difference frames. The change in the Bragg spots becomes more prominent over time (10-100 ps). The lower part of the peaks becomes brighter, whereas the upper part becomes darker, showing a downward shift of the Bragg spots. With extended time delays (200-1110 ps), the difference patterns get fainter again, indicating that the peaks are moving back to their original position. This behavior confirms that both the heating and electron pulses, because of their ultrashort durations and fluxes, do not damage the bilayer.

Seidel et al. The resulting change, ∆c0, which is the CH2-CH2-CH2 distances of chains for every single Bragg spot (φ ) 90°), is plotted in Figure 14A at different times for the 2, 4, and 8-layer samples. For φ ) 0° (not shown), the dynamical behavior is similar. We note that the absolute ∆s change is given as observed on the CCD, but to obtain the change in real space ∆c0, we use the s value for the higher order diffraction, i.e., l ) 2 minus l ) 1 (steady-state), as we did for static structural determination, in order to avoid uncertainty at the lowest θi, as discussed above. It is further noted that, under identical conditions, the 2-, 4-, and 8-layer samples exhibit a larger amplitude of ∆c0 change with increasing film thickness, a point that we shall discuss below. As a result of energy transfer from the heated silicon substrate into the adsorbate, the observed ultrafast increase of the axial change ∆c0 is consistent with expansion of the fatty acids solely along their aliphatic chains (20-60 ps). This efficient and directional energy transfer within the fatty acid assembly is facilitated by the unique substrate heating. As discussed in paper 1, the initial ultrafast heating induces a potential-driven lattice change in the material, and this results in a surface displacement which for the case discussed here acts as a directional force. It should be mentioned that the surface heating area by the laser is much larger than the absorption depth in the substrate, and this gradient of stress (temperature) facilitates the c-direction change. Moreover, the transfer within the chain is much more efficient than across the chains because of covalent bonding, as opposed to the relatively weak interactions across the chains. The elongation is followed by nonequilibrium contraction and restructuring due to energy redistribution and diffusion at longer times (60-230 ps and 1.2-1.5 ns). As discussed in section V, the behavior is indicative of two regimes of structural dynamics: that of nonequilibrium coherent motion of atoms at short time and the evolution toward equilibrium at longer times. 2. Effect of Multilayers. The effect of multilayers on the expansion is shown in Figure 14. The amplitude increases with overall chain length which as discussed in section V is consistent with dynamics in the nonequilibrium regime. However, we must also take into account that LB films show an increased density of “defects” and holes upon increasing thickness (mainly in the outer layers).39 As mentioned above, under static structure conditions, the decreased influence of substrate has the consequence of in-plane expansion of the subunit cell along a0 and b0, the onset of chain inclination, and the loss of crystalline order. The decreased density in LB films upon going from single to multiple layers results in a less confined environment, with the expansion along the carbon chains becoming freer. Spatial confinement indeed decreases the mobility of say a chromophore in the membrane of vesicles74 and even water in reverse micelles.75 The inclined chains provide an opportunity to test the nature of the expansion in directions other than perpendicular. Their inclined diffraction patterns are shown in Figure 15 for the 8-layer sample deposited at π ) 29 mN/m and pH ∼ 7. After the temperature jump, the smeared diffraction lines move down along the direction perpendicular to their longitudinal axis on the CCD, as shown in Figure 15, panels A and B, for various diffraction difference images. The amplitude and time behavior of the change in the lattice spacing is comparable to that of a nontilted pattern (see Figure 15 C) and is calculated from the s change for the (hk2) line along the inclined direction. The dynamics of the nontilted pattern is shown for the (002) Bragg spot of the 8-layer calcium arachidate sample deposited at pH ) 6.4 and π ) 29 mN/m.



Figure 14. Structural dynamics for the calcium arachidate samples of 2, 4 and 8 layers (from left to right) for φ ) 90°, as shown in Figure 7, deposited at pH ) 6.4 and π ) 29 mN/m. (A) The molecular axial length change ∆c0 of the CH2 subunit cell as a function of time showing the behavior for the different diffraction spots as indexed. (B) Relative intensity change I/I0 of the (002) Bragg spot as a function of time. Note the increase in amplitude with increase in layer thickness. The distances are referenced to l ) 1 (see text).

This observation of tilted line movement is in total agreement with the heating pulse initiating an expansion and contraction solely along the fatty acid carbon chains. The fact that the lines move along the inclined direction rather than the direction perpendicular to the shadow edge also excludes other effects which could force the Bragg spots to move in the diffraction patterns, for example charging effects or direct beam movements. 3. Transient Structural Ordering. In Figures 12C, 14B, and 15D, the relative integrated intensity changes I/I0 of the (002) Bragg spots as a function of time for the 2-, 4-, and 8-layer samples are depicted. Even though we show only transient intensity behavior of the (002) Bragg spot, the intensities of

the other Bragg spots exhibit similar trends. All samples show comparable transient intensity change. For “conventional” heating, the intensity is expected to initially drop and subsequentially recover back to the initial value, similar to the trend observed for the Bragg spot center position ∆c0. This intensity drop follows the influence of the incoherent motions of the atoms, which are described by the Debye-Waller factors (see paper 1), and structural changes due to phase transitions (see Figure 10 for the static temperature behavior). For the fatty acids, the transient behavior is much different from that observed at steady state. We observe only a very short-lived intensity drop on the ultrafast time-scale, on the order of tens of ps. This initial drop is consistent with a


Seidel et al.

Figure 15. Static diffraction patterns at φ ) 0° and dynamics of the inclined 8-layer sample deposited at pH ∼ 7 and π ) 29 mN/m. (A) The static diffraction pattern before laser irradiation shows an inclination of ∼10° (B) the diffraction difference images show the movement of the smeared lines along the direction of the aliphatic carbon chains. (C) The axial change of c0 for the inclined fatty acids LB-films at T ) 300 K in comparison with the changes for the non-inclined 8-layer sample deposited at pH ) 6.4 and 29 mN/m at φ ) 90° (see Figure 7). (D) The corresponding relative intensity changes I/I0.

disorder-induced heating, as it represents a decrease in intensity. However, after this initial decrease, the intensity recovers also fast and even gains more intensity from that of the initial value. The fact that the transient intensity is higher than the initial one indicates the formation of a transient structure with a higher degree of order. We termed this behavior transient structural ordering in our previous study of phospholipids. Interestingly, the intensity change mirrors the line width behavior as shown, e.g., in Figure 12, panels D and C. The line width of the Bragg spots reflects the order of the crystalline structure and is narrower for a more ordered state. This trend is again consistent with transient structural ordering, which can be understood using the following picture, schematically depicted in Figure 16. Initially, the dynamics follows the expected loss of order due to rapid thermal motions similar to that observed in the early stage of the static temperature dependence; note that the lower part of Figure 16 is similar to that shown in Figure 11. A static temperature rise results in an expansion of the fatty acid lattice along a0 and b0 (see Figure 10B), whereas upon laser pulse heating, the energy is mainly transferred along the aliphatic chains, which results in a large increase in c0. By overcoming some energy barrier and through alignment of the aliphatic chains and other motions (rotation, bending, etc.), and possibly through transient annealing of the film, the fatty acid molecules find an energetically more favorable structure in the film. As a

result, the formerly imperfect structure transforms into a more ordered system. Because of energy dissipation on the nanosecond time scale, the film eventually restructures to its initial state. The Bragg spot center position, intensity and the line width recover toward their equilibrium values with ns recovery component, and on the ms (our laser repetition time) scale, the signal is certainly back to the initial values. This behavior of structural ordering is also observed for phospholipids, as discussed in the next section. The comparison of transient and static structural changes was directly made by studying the transient behavior at different substrate temperatures. In Figure 17, the transient axial change ∆c0 and the integrated intensity change for different static temperatures are plotted for the bilayer deposited at pH ∼ 7 and π ) 29 mN/m. The dynamics are shown for the (002) Bragg spot observed at φ ) 0° and T ) 100, 295 and 333 K. Indeed, the temporal behavior for the expansion and restructuring is similar for the three different temperatures, but the initial maximum amplitudes of expansion are different, increasing with increasing temperature. This trend is again consistent with the picture discussed above, namely that elevated temperatures facilitate a more pronounced expansion. The results shown in Figure 17B further support the picture, as the intensity changes show more pronounced transient structural ordering extending over a longer time span at elevated temperatures. For T ) 295 and 333 K, the intensity recovers


Figure 16. Schematic view of the transient structural ordering. Immediately following the ultrafast laser heating pulse, the carbon chains expand, and with the acquired energy they “anneal” to form a more ordered structure. This transient structure changes upon cooling by energy redistribution and diffusion toward the state of equilibration. This behavior is different from that of steady-state heating; see Figure 11 and text.

back to its initial value on a nanosecond time-scale or larger, whereas for T ) 100 K, the transient structural ordering persists only for ∼350 ps after laser pulse heating, after which the intensity drops below the initial value again and, therefore, resembles the conventional Debye-Waller behavior. This strengthens the view that expansion along c0 and preparation of the ordered state are not two unrelated processes, but are facets of the same underlying molecular motions. In summary, all investigated fatty acid adsorbates show expansion and restructuring along the fatty acid carbon chains. From the transient and static diffraction behavior with time and temperature, it is shown that transient structural ordering on the ultrafast time-scale, initiated by laser pulse heating, is a common feature of these layered structures. Such transient structures, their anisotropy of expansion, and inhomogenity would remain unrevealed without the resolutions of UEC. E. Phospholipids. The above results of structural dynamics of fatty acid adsorbates on substrates encouraged us to extend our studies to even more complex structures, such as phospho-


Figure 17. Dependence of the transient dynamics on initial substrate static temperature: T ) 100, 295, and 333 K are shown. (A) The transient molecular axial change ∆c0 of the CH2 subunit cell as a function of time for φ ) 0°. (B) The corresponding relative intensity change I/I0 as a function of time. We note the increase in amplitude of ∆c0 as temperature increases, and that transient structural ordering (Figure 16) becomes more pronounced at higher substrate (initial) temperatures.

lipids, which are one of the actual building blocks of biomembranes. The chemical structure of phospholipids has in common with the fatty acids the two aliphatic chains instead of one, but with a different headgroup (see Figure 1A). Therefore, similarities in dynamics of chains are possible. In what follows we present studies of dimyristoyl phosphatidic acid (DMPA). 1. Structure. In Figure 18, panels A and B, the static diffraction patterns for DMPA monolayers and bilayers are depicted, respectively. The two patterns are very similar, and their strongest feature is a horizontal diffraction (curved) line, which is labeled by the Miller indices (hk2). Since the line is basically the composite of Bragg diffractions in the second diffraction order (compare to the diffraction pattern of fatty acids in Figure 4), we are able to determine the subunit cell parameter c0 to be 2.54 Å from the s value in the vertical direction. This again is expected because of the covalent -CH2-CH2-CH2distances involved in the aliphatic chains of DMPA, which are essentially the same as in arachidic acid (see Figure 1A). The patterns indicate that the aliphatic chains are packed parallel to each other and are aligned nearly perpendicular to the surface.


Seidel et al.

Figure 18. Observed structural dynamics of the phospholipid DMPA. Shown are the static diffraction patterns and the diffraction difference patterns at t ) 128 ps for DMPA (A) monolayers and (B) bilayers. The axial change ∆c0 and the normalized intensity of the (hk2) “diffraction line” as a function of time are given for DMPA (C) monolayers and (D) bilayers.

However, long-range in-plane order was not observed, as we could not resolve separate Bragg spots, as we did for the fatty acids. This limited resolution may be due to the fact that the more complex chemical structure of DMPA does not allow the molecules to pack as easily into orthorhombic (or other symmetries) crystals over an extended area. The crystallinity of the film is more restricted; “bulk” crystals of DMPA show inclination (β ) 114.2°) of the carbon chains,76 whereas in the adsorbate DMPA on the substrate the chains are forced to be nearly perpendicular to the surface. Diffraction patterns show diffuse scattering in the lower s range, and the independence of the diffraction patterns on electron incidence angle θi and the azimuthal angle φ is consistent with the existence of polycrystalline domains. 2. Dynamics. The diffraction frames were monitored as a function of time after the temperature jump in the underlying substrate. The patterns show mainly a change in the (hk2) line position, becoming dark at the higher s value and bright at the lower s value, indicating a downward movement of the diffraction line (see Figure 18, panels A and B, for a delay time of t ) +128 ps). This translates to an increase of the -CH2CH2-CH2- distances along the aliphatic chains. As before, we were able to fit the diffraction line with a Voigt function in the vertical s direction, providing us with accurate center position and intensity changes with time, as shown in Figure 18, panels C and D, for the monolayer and bilayer, respectively. The error range for the width did not permit accurate values. The dynamics of the center position change is similar for both samples, monolayer and bilayer, and is similar to the behavior observed for fatty acids. An increase in c0 within ∼50 ps was observed with a recovery on the ∼500 ps and longer time scale. This similarity in trend is not surprising since we

are probing the structural dynamics of the aliphatic chains, and in this respect DMPA resembles the fatty acid samples. The amplitude of the expansion for the bilayer is nearly three times that of the monolayer. A similar trend was observed for samples of 2, 4, and 8 layers of fatty acids (see Figure 14B) but not with as large of a change. This larger change of amplitude may be because the first layer exhibits different forces than the subsequent layers as it is directly bound to the silicon substrate. If the interaction forces are stronger than in the subsequent layers, we expect a tighter binding to the substrate and an increased order in the first layer, resulting in less mobility for expansion. It is interesting that for bilayers of fatty acids and phospholipids the expansion of the former is nearly half that of the latter, suggesting a more flexibility for expansion in phospholipids, which is again consistent with the above-mentioned picture. The monolayer was studied on a hydrophilic surface, unlike the bilayer, and this is entirely in line with the proposed strong interaction with the substrate. The intensity changes with time are shown in Figure 18, panels C and D. As in the case of the fatty acids, initially an intensity drop is recorded, consistent with the disorder being induced within 15 ps by the heating pulse. After this change, the intensity increases to a value above the static one within ∼130 ps, before final recovery toward the initial configuration at ∼400 ps and longer times. The transient state of increased intensity is indicative of transient structural ordering, as discussed above for fatty acid structures. In conclusion, the phospholipid monolayer and bilayer samples of DMPA show a transient behavior similar to that of fatty acids, both being adsorbates on a substrate. We observed elongation along the aliphatic chains following the ultrafast

Ultrafast Electron Crystallography. 2 temperature jump, accompanied by transient structural ordering. The similarity is not surprising because structural changes involve the -CH2-CH2-CH2- subunit cell. However, the amplitude of expansion, the influence of substrate forces, and the mobility of the structure are found to have differences. V. Structural Dynamics Picture and Conclusion In the preceding paper 1, it was demonstrated that ultrafast dynamics and heating of materials with a band gap are directly evident in the change with time of Bragg diffraction spot positions and their intensities and widths. The mechanism for the generation of the temperature jump is basic: the femtosecond infrared pulse excites carriers which then by electron-phonon coupling heat up the lattice and the material sustains the lattice temperature until diffusion of heat takes place on a much larger time scale. Because of the change in potential by carrier excitation and the fact that the excited region with the laser pulse is much larger than the absorption length, the gradient change is perpendicular to the surface of the material. This gradient results in a large expansion of surface atoms which is a force on the ultrashort time scale. In this paper 2, this T jump of the material (substrate), or the force of surface atoms, is exploited to heat up the adsorbate in direct contact with either a hydrophobic or hydrophilic substrate; the femtosecond infrared pulse has no resonance for absorption to the adsorbate. Such studies made here for monolayers, bilayers, and multilayers of fatty acids and phospholipids provide an opportunity to study structural dynamics at these interfaces on the nanometer scale and to examine changes due to the transition from 2-D to 3-D dimensionality. To achieve crystallinity, we use the methodology of Langmuir-Blodgett film, providing control over pH, thickness, and pressure. Elsewhere,4 we studied adsorbates formed only by self-assembly. Four types of UEC measurements make possible the study of structural dynamics: the change of position with time of Bragg spots; the temporal evolution of the diffraction intensity change, the increase/decrease in diffraction width, and the change of diffraction with angle of incidence θi (rocking curve) and azimuthal angle φ for the position of the electron pulse relative to the zone axis of the substrate (see Figure 3). As shown in Figures 13, 12, and 5, the measurements can be made with the sensitivity achieved in UEC. Prior to these measurements, we establish the static (time-averaged) structures of the adsorbates (fatty acids and phospholipids) by determining the orientation of the chains relative to surface plane and the -CH2-CH2-CH2- chain distances which are the subunit cell dimensions a0, b0, and c0. For arachidic fatty acid, depending on pH and deposition conditions, a0 ranges from 4.7 to 4.9 Å, b0 ranges from 8.0 to 8.9 Å, and c0 ranges from 2.54 to 2.59 Å, and for dimyristoyl phosphatidic acid (DMPA) c0 ) 2.54 Å. The rocking curve in Figure 5 indicates two important points to be made regarding differences in 2-D and 3-D crystal behavior. First, at large θi values, the electron packets penetrate the bulk, and we observe the higher orders of Bragg diffraction of the substrate. In fact, it is from these high order diffractions that we can determine the lattice spacing of the used substrate. At lower values of θi, the spots are stretched diagonally, making the substrate spacing less accurate to determine. At the lowest θi, the pattern is dominated by the adsorbate diffraction features. The behavior with θi of an adsorbate spot shows insensitivity in the region of θi studied. This brings to focus the second point regarding measurements of rocking curves. Unlike 3-D crystals, 2-D systems exhibit “rods” in the diffraction reciprocal space. As the layers build

J. Phys. Chem. C, Vol. 111, No. 13, 2007 4935 up in thickness, the rods are modulated by the inverse chain distance of c0. Thus, for 2-D systems the intensity should not change drastically with θi, whereas for 3-D they will. From the horizontal width of our diffraction (∆s ∼0.06 Å-1), we obtain a nanometer scale coherence length. Moreover, the vertical width of ∆s gives a maximum inhomogenity in -CH2-CH2-CH2distances of ∼20%, relative to the 2.54 Å distance of c0. The transient anisotropic change in c0 of fatty acid and phopholipid layers is vastly different from that observed in the steady-state equilibrium state. At equilibrium, we observe changes in a0 and b0 (not c0), and the diffraction intensity only decreases to reflect thermal, incoherent motions (Debye-Waller effect) and phase transitions; it is known, e.g., that for phospholipids there exist different phases (“gel” and “liquid”), and it is interesting that neutron scattering studies give a lifetime of 17 ps for the acoustic phonon excitation in the gel phase.77 On the ultrafast time-scale, the expansion is along c0, unlike the thermal case, and the amplitude of change is far larger than predicted by incoherent thermal expansion. The expansion amplitude depends on layer thickness and the nature of bonding to the substrate (hydrophilic vs hydrophobic). The intensity and width changes are very different from those observed by equilibrium heating. Following the initial ultrafast heating, the structure first expands (atomic displacements) along the c direction (compare Figures 11 and 16). These motions with the acquired energy in the layers lead into transient structural ordering through “annealing” and/or chain motions, as evident from increased diffraction intensity and narrowing of diffraction width beyond the initial values. On the nanosecond and longer time scale, the structure reaches quasiequilibrium or equilibrium (incoherent movement of atoms), and by dissipation of heat (diffusion), the structure acquires the original configuration, certainly on the millisecond time scale between pulses. This behavior is in contrast with that observed at steady state as mentioned above (see also Figures 11 and 16). The net change in displacement is determined by the impulsive force (“temperature”) of the substrate (including coupling), and the maximum value of the extension depends on elasticity and heat capacity. If heating occurs for an equilibrated system, the change in the value of c0 with temperature ∆c0 (by anharmonicity) should be independent of the number of -CH2-CH2-CH2- subunits in the chain, and ∆c0/c0 becomes simply R (the thermal expansion coefficient), which is typically very small (∆c0/c0 ∼ 10-5 T-1);78 for a 10degree rise, this expansion would be on the order of 10-4 Å, whereas the observed transient change is as large as 0.01 Å. For nonequilibrium dynamics in the chains, this large amplitude is understood even for harmonic chains. The impulsive force at short times transmits a large change in the value of ∆c0 as the disturbance (wave-type) accumulates to give the net effect that is dependent on the number of C atoms.79 In other words, as the disturbance passes through the different bonds, the diffraction amplitude builds up and exhibits a delay, ultimately giving a rise and large total amplitude for the change. This picture also explains the dependence on the total length of the chains, the increase in initial maximum amplitude as the temperature of the substrate increases, and the effect of substrate strong (hydrophilic) vs weak (hydrophobic) binding (see Figures 14, 17, and 18). Quantification of the total change must take into account the nature of the substrate force and variation in the density of the LB films upon going from single to multiple layers. It is known that “defects” or “holes” can be formed when multiple layers are deposited and that


Seidel et al.

Figure 19. Structures and models used for obtaining the MD simulations (see text).

disorder is expected, as the influence of the substrate subsides. The fact that the initial change in intensity (and elongation) occurs on the 10 ps time scale and that the distance traveled is approximately 20 Å (for a monolayer), the speed of propagation should be sub kilometers per second, which is close to the propagation of sound waves. Because the substrate is heated through optical and acoustic phonons (paper 1), the rise is convoluted with the process of phonon generation which is on the time scale of 10 ps (see paper 1); we note that the resolution imposed by the geometry of electron and light propagation on the surface is ∼1-2 ps (see section III.A). Accordingly, the speed could be of higher value reaching the actual speed of sound in the layers. Future experiments will further resolve this region to elucidate the maximum extension possible and the expected coherent features. The coherent coupling among bonds in the underdamped regime of harmonic motions is vastly different from the diffusive behavior in the overdamped regime. The force and vibrational frequencies of -CH2-CH2-CH2motions in the diffusive (overdamped) regime have to be unrealistically large, as will be shown elsewhere. In the underdamped regime, although the force is significant to cause

the large change in ∆c0, and coherent propagation is within the chain, the change in the value of ∆c0 relative to that of c0 is relatively small to preserve the robustness of the Bragg spots throughout the temporal change. It is important to point out that for large complex systems the diffraction has features of both “crystalline” and “diffusive” scatterings. The latter gives structural information on the collective motions, which can be examined through cross terms of the Debye-Waller factors.80 In this regard, the ring(s) apparent at s ∼ 0.3 Å-1 in the diffraction difference patterns should be analyzed for such correlations in the phospholipid, and in future work we shall consider this analysis. Besides the analytical theoretical work discussed above, which will be published as part of this series, in collaboration with Prof. T. Shoji and colleagues in Japan, we have studied MD simulations of a prototype system. The model used (see Figure 19) is that of a silicon substrate with the adsorbates made of C20H42 chains, covering a total combined length of 95 Å. The potentials for the chains, substrate, and the interaction at the interface were from ab initio calculations. The time step was 0.5 fs, and the total number of steps was 200 000. The heat pulse was modeled based on the kinetic energy of the substrate

Ultrafast Electron Crystallography. 2 atoms. The radial distribution function and the actual vibration motions of the atoms were obtained at different times, and we considered both the self-assembly and the confined geometry of the chains. These calculations provided the structural cell dimensions observed experimentally, and elucidated the coherent motion in the chain bonds and their time scales. Preliminary results show the increase in -CH2-CH2-CH2- distance near the silicon surface by 0.08 Å in about 5 ps. More details will be given in a separate contribution. In conclusion, ultrafast electron crystallography provides the opportunity to study nonequilibrium complex structures of adsorbates on substrates, with unprecedented atomic-scale spatial and ultrafast temporal resolutions, with the sensitivity of a monolayer. The reported studies here for fatty acids and phospholipids demonstrate very different behaviors in the nonequilibrium state when compared with those studied, also here, at equilibrium. Perhaps unexpected are the collective and coherent motions of atoms in such complex systems and the transient structural ordering that exist premelting and phase transitions. As such, UEC is unique in uncovering these phenomena on the time-scale they occur. Future work will consider the role of hydration, and the extension to biological membranes. Acknowledgment. This work was supported by the National Science Foundation and the Gordon and Betty Moore Foundation. We thank Prof. J. Heath for the use of his facilities and E. DeIonno, J.-W. Choi, P. Cao, Dr. Y. Luo, and Dr. H. Yu for advice in preparing and understanding the LB films. Helpful discussions with D.-S. Yang and Drs. N. Gedik, S. Habershon, and J. Tang are gratefully acknowledged; the simulations in Figure 6 were made by Dr. S. Habershon while at Caltech. References and Notes (1) Yang, D.-S.; Gedik, N.; Zewail, A. H. J. Phys. Chem. C 2007, 111, 4889-4919. (2) Ruan, C.-Y.; Vigliotti, F.; Lobastov, V. A.; Chen, S.; Zewail, A. H. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 1123-1128. (3) Vigliotti, F.; Chen, S.; Ruan, C.-Y.; Lobastov, V. A.; Zewail, A. H. Angew. Chem., Int. Ed. 2004, 43, 2705-2709. (4) Ruan, C.-Y.; Yang, D.-S.; Zewail, A. H. J. Am. Chem. Soc. 2004, 126, 12797-12799. (5) Ruan, C.-Y.; Lobastov, V. A.; Vigliotti, F.; Chen, S.; Zewail, A. H. Science 2004, 304, 80-84. (6) Chen, S.; Seidel, M. T.; Zewail, A. H. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 8854-8859. (7) Chen, S.; Seidel, M. T.; Zewail, A. H. Angew. Chem., Int. Ed. 2006, 45, 5154-5158. (8) Roberts, G. Langmuir-Blodgett Films; Plenum: New York, 1990. (9) Petty, M. C. Langmuir-Blodgett Films; Cambridge University Press: Cambridge, U.K., 1996. (10) Blodgett, K. B. J. Am. Chem. Soc. 1935, 57, 1007-1022. (11) Blodgett, K. B.; Langmuir, I. Phys. ReV. 1937, 51, 964-982. (12) Blodgett, K. B. J. Phys. Chem. 1937, 41, 975-984. (13) Nørgaard, K.; Bjørnholm, T. Chem. Commun. 2005, 14, 18121823. (14) Girard-Egrot, A. P.; Godoy, S.; Blum, L. J. AdV. Colloid Interface Sci. 2005, 116, 205-225. (15) Khanarian, G. Thin Solid Films 1987, 152, 265-274. (16) Zasadzinski, J. A.; Viswanathan, R.; Madsen, L.; Garnaes, J.; Schwartz, D. K. Science 1994, 263, 1726-1733. (17) Takamoto, D. Y.; Aydil, E.; Zasadzinski, J. A.; Ivanova, A. T.; Schwartz, D. K.; Yang, T.; Cremer, P. S. Science 2001, 293, 1292-1295. (18) Karle, J.; Brockway, L. O. J. Chem. Phys. 1947, 15, 213-225. (19) Kinzler, M.; Schertel, A.; Ha¨hner, G.; Wo¨ll, Ch.; Grunze, M.; Albrecht, H.; Holzhu¨ter, G.; Gerber, Th. J. Chem. Phys. 1994, 100, 77227735. (20) Russell, G. J.; Petty, M. C.; Peterson, I. R.; Roberts, G. G.; Lloyd, J. P.; Kan, K. K. J. Mater. Sci. 1984, 3, 25-28. (21) Robinson, I.; Jarvis, D. J.; Sambles, J. R. J. Phys. D: Appl. Phys. 1991, 24, 347-359. (22) Klechkovskaya, V. V.; Feigin, L. A. Crystallogr. Rep. 1998, 43, 917-924.

J. Phys. Chem. C, Vol. 111, No. 13, 2007 4937 (23) Schwartz, D. K.; Viswanathan, R.; Garnaes, J.; Zasadzinski, J. A. J. Am. Chem. Soc. 1993, 115, 7374-7380. (24) Mukhopadhyay, M. K.; Sanyal, M. K.; Datta, A.; Webster, J.; Penfold, J. Chem. Phys. Lett. 2005, 407, 276-282. (25) Karageorgiev, P. P.; Karageorgieva, V. V.; Luchinin, V. V.; KazakKazakevich, A. Z. Crystallogr. Rep. 1998, 43, 968-977. (26) Shih, M. C.; Bohanon, T. M.; Mikrut, J. M.; Zschack, P.; Dutta, P. J. Chem Phys. 1992, 96, 1556-1559. (27) Kundu, S.; Datta, A.; Hazra, S. Chem. Phys. Lett. 2005, 405, 282287. (28) Evenson, S. A.; Badyal, J. P.; Pearson, C.; Petty, M. C. J. Phys. Chem. 1996, 100, 11672-11674. (29) Peterson, R.; Russell, G. J.; Earls, J. D.; Girling, I. R. Thin Solid Films 1988, 161, 325-331. (30) Brzezinski, V.; Peterson, I. R. J. Phys. Chem. 1995, 99, 1254512552. (31) Robinson, I.; Sambles, J. R.; Peterson, I. R. Thin Solid Films 1989, 172, 149-158. (32) Gaines, G. L. JR. Thin Solid Films 1980, 68, 1-5. (33) Bonnerot, A.; Chollet, P. A.; Frisby, H.; Hoclet, M. Chem. Phys. 1985, 97, 365-377. (34) Peterson, I. R.; Russell, G. J. Thin Solid Films 1985, 134, 143152. (35) Leuthe, A.; Riegler, H. J. Phys. D: Appl. Phys. 1992, 25, 17861797. (36) Leuthe, A.; Chi, L. F.; Riegler, H. Thin Solid Films 1994, 243, 351-357. (37) Morelis, R. M.; Girardegrot, A. P.; Coulet, P. R. Langmuir 1993, 9, 3101-3106. (38) Schwartz, D. K. Surf. Sci. Rep. 1997, 27, 241-334. (39) Peng, J. B.; Barnes, G. T.; Gentle, I. R. AdV. Colloid Interface Sci. 2001, 91, 163-219. (40) Dyck, M.; Kru¨ger, P.; Lo¨sche, M. Phys. Chem. Chem. Phys. 2005, 7, 150-156. (41) Helm, C. A.; Tippmann-Krayer, P.; Mo¨hwald, H.; Alsnielsen, J.; Kjaer, K. Biophys. J. 1991, 60, 1457-1476. (42) Fujiwara, I.; Ohnishi, M.; Seto, J. Langmuir 1992, 8, 2219-2222. (43) Subirade, M.; Salesse, C.; Marion, D.; Pezolet, M. Biophys. J. 1995, 69, 974-988. (44) Zhang, J.; Unwin, P. R. J. Am. Chem. Soc. 2002, 124, 2379-2383. (45) Hui, S. W.; Parsons, D. F.; Cowden, M. Proc. Natl. Acad. Sci. U.S.A. 1974, 71, 5068-5072. (46) Orekhova, L. L.; Mogilevich, A.; Orechov, S. V.; Klechkovskaya, V. V.; Feigin, L. A. Physica B 1994, 198, 144-145. (47) Stevens, M.; Longo, M.; Dorset, D. L.; Spence, J. Ultramicroscopy 2002, 90, 265-272. (48) Seul, M.; Eisenberger, P.; McConnell, H. M. Proc. Natl. Acad. Sci. U.S.A. 1983, 80, 5795-5797. (49) Gregory, B. W.; Dluhy, R. A.; Bottomley, L. A. J. Phys. Chem. 1994, 98, 1010-1021. (50) Lozano, P.; Ferna´ndez, A. J.; Ruiz, J. J.; Camacho, L.; Martı´n, M. T.; Munoz, E. J. Phys. Chem. B 2002, 106, 6507-6514. (51) Ho¨rber, J. K. H.; Lang, C. A.; Ha¨nsch, T. W.; Heckl, W. M.; Mo¨hwald, H. Chem. Phys. Lett. 1988, 145, 151-158. (52) Solletti, J. M.; Botreau, M.; Sommer, F.; Brunat, W. L.; Kasas, S.; Duc, T. M.; Celio, M. R. Langmuir 1996, 12, 5379-5386. (53) Naselli, C.; Rabolt, J. F.; Swalen, J. D. J. Chem. Phys. 1985, 82, 2136-2140. (54) Naselli, C.; Rabe, J. P.; Rabolt, J. F.; Swalen, J. D. Thin Solid Films 1985, 134, 173-178. (55) Richardson, W.; Blasie, J. K. Phys. ReV. B 1989, 39, 12165-12181. (56) Bo¨hm, C.; Steitz, R.; Riegler, H. Thin Solid Films 1989, 178, 511517. (57) Riegler, J. E. J. Phys. Chem. 1989, 93, 6475-6480. (58) Tippmann-Krayer, P.; Kenn, R. M.; Mo¨hwald, H. Thin Solid Films 1992, 210/211, 577-582. (59) Chou, C. F.; Jin, A. J.; Hui, S. W.; Huang, C. C.; Ho, J. T. Science 1998, 280, 1424-1426. (60) Viswanathan, R.; Madsen, L. L.; Zasadzinski, J. A.; Schwartz, D. K. Science 1995, 269, 51-54. (61) Sikes, H. D.; Schwartz, D. K. Science 1997, 278, 1604-1607. (62) Peng, J. B.; Foran, G. J.; Barnes, G. T.; Gentle, I. R. Langmuir 2003, 19, 4701-4706. (63) Fontana, M. P.; Facci, P. J. Chem. Phys. 1999, 111, 5562-5567. (64) Basu, J. K.; Sanyal, M. K.; Mukherjee, M.; Banerjee, S. Phys. ReV. B 2000, 62, 11109-11117. (65) Mukhopadhyay, M. K.; Sanyal, M. K.; Datta, A.; Mukherjee, M.; The, Geue; Grenzer, J.; Pietsch, U. Phys. ReV. B 2004, 70, 245408. (66) Harris, A. L.; Levinos, N. J. J. Chem. Phys. 1989, 90, 3878-3879. (67) Ma, S.; Aono, M.; Suzuki, T. Colloids Surf. A 2005, 257-258, 357-362.

4938 J. Phys. Chem. C, Vol. 111, No. 13, 2007 (68) Rousse, A.; Rischel, C.; Uschmann, I.; Forster, E.; Albouy, P. A.; Geindre, J. P.; Audebert, P.; Gauthier, J. C.; Antonetti, A. J. Appl. Crystallogr. 1999, 32, 977-981. (69) Kobayashi, K.; Takaoka, K.; Ochiai, S.; Kundu, S.; Datta, A.; Hazra, S. Thin Solid Films 1988, 159, 267-273. (70) Kitaigorodskii, A. I. Organic Chemical Crystallography; Consultants Bureau: New York, 1961. (71) Buhaenko, M. R.; Grundy, M. J.; Richardson, R. M.; Roser, S. J. Thin Solid Films 1988, 159, 253-265. (72) Kaganer, V. M.; Mo¨hwald, H.; Dutta, P. ReV. Mod. Phys. 1999, 71, 779-819. (73) Dierker, S. B.; Murray, C. A.; Legrange, J. D.; Schlotter, N. E. Chem. Phys. Lett. 1987, 137, 453-457.

Seidel et al. (74) Seidel, M.; Jethwa, J.; Vo¨hringer, P. Russ. Chem. Bull., Int. Ed. 2004, 53, 1471-1476. (75) Levinger, N. E. Curr. Opin. Colloid Interface Sci. 2000, 5, 118124. (76) Harlos, K.; Eibl, H.; Pascher, I.; Sundell, S. Chem. Phys. Lipids 1984, 34, 115-126. (77) Rheinsta¨dter, M. C.; Ollinger, C.; Fragneto, G.; Demmel, F.; Salditt, T. Phys. ReV. Lett. 2004, 93, 108107. (78) Fukui, T.; Sugi, M.; Iizma, S. Phys. ReV. B 1980, 22, 48984899. (79) Tang, J.; Zewail, A. H. in preparation. (80) Meinhold, L.; Smith, J. C. Phys. ReV. Lett. 2005, 95, 218103.

Ultrafast Electron Crystallography. 2. Surface Adsorbates of Crystalline

Recommend Documents