Two-Dimensional Order in Mercury-Supported Langmuir Films of Fatty

Article pubs.acs.org/Langmuir

Two-Dimensional Order in Mercury-Supported Langmuir Films of Fatty Diacids Lilach Tamam,† Benjamin M. Ocko,‡ and Moshe Deutsch*,† †

Physics Department and Institute of Nanotechnology and Advanced Materials, Bar-Ilan University, Ramat-Gan 52900, Israel Condensed Matter Physics & Material Science Department, Brookhaven National Laboratory, Upton, New York 11973, United States

‡

ABSTRACT: The structure of mercury-supported Langmuir films of dicarboxylic acid molecules with 13 ≤ n ≤ 22 carbons is studied by X-ray methods and surface tensiometry. The molecules lie surface-parallel, forming mono-, bi-, or trilayers, depending on coverage. All films exhibit a full 2D order of the same single-molecule oblique unit cell. In particular, the distinct odd−even structure difference of 3D crystals of the same molecules is not observed. The unit cell’s width and angle show a small systematic decrease with n, while the length increases commensurately with the molecular length. These results show the films to consist of closely packed, extended, polymer-like chains of diacid molecules, bound by their carboxyl end groups. Evidence is presented for the inclusion of a single mercury atom in the carboxyl−carboxyl bond. The possible conformation of this bond and implications of the parity-independent structure are discussed.

■

molecular self-assembly into new supramolecular structures27 and novel bonding synthons like the hydroxyl−carboxyl tetramer found for hydroxy-functionalized fatty monoacids on a graphite surface, which does not exist in 3D crystals.28 On the other hand, a clear disadvantage of the confinement of the molecules to an ordered solid substrate is the epitaxial imposition, by the molecule−substrate interaction, of the substrate’s structure on the lateral structure of the monolayer.28,29 The use of a liquid surface, e.g. water, eliminates the epitaxy problem. However, for end-functionalized alkyl molecules, the system of choice for these studies, the water−alkyl repulsion imposes a surface-normal orientation on the molecules when studied on the surface of aqueous subphases.30 This reduces greatly the functional end groups’ contacts and thus the possibilities of hydrogen bonding between neighboring molecules. Moreover, the hydration shells of the waterimmersed headgroups strongly modify and weaken the HBs that would have formed between bare headgroups. In a recent study,31 we have demonstrated that liquid mercury has several important advantages as a subphase for HB studies in organic Langmuir films (LFs). As the alkyl chain’s hydrophobic water repulsion is replaced by a vdW attraction to the surface, mercury imparts a surface parallel orientation to linear chain molecules.32−34 This orientation provides an optimal headgroup orientation for forming HBs and, of course, eliminates the hydration shell problem. Moreover, for X-ray

INTRODUCTION Hydrogen bonds (HB) are ubiquitous in nature and life processes, dominating diverse phenomena like the doublestrand structure of DNA, protein folding and biomolecular recognition,1−4 the structure of charged geochemical interfaces,5 and surfaces and interfaces involving water.6 HBs are also a major synthon in supramolecular chemistry, crystal engineering, and materials science7−11 and are extensively employed in nanoscience and technology for nanopatterning and molecular electronics.12−14 Yet, in spite of a century of research,15−18 the forms, variety, and properties of these bonds, and the reasons underlying their central role in the fields mentioned above, are still not fully understood and are being intensively explored.7,19 Much of the progress made in understanding HBs, their underlying interactions, molecule type and architecture dependences, the bond’s strength-length relation, etc., was obtained from three-dimensional crystal structure determinations for a large variety of molecules exhibiting such bonds.19−23 In particular, the seminal work of Thalladi et al.24−26 resolved in detail the 3D crystal structures of diols, thiols, and diacids and the dependence of these structures on the delicate balance between the van der Waals (vdW) interaction of the alkyl moieties and the HBs formed by the functional end groups. Substrate-supported monolayers of end-functionalized alkyl molecules are a particularly attractive system for studying HBs. The molecular degrees of freedom in such monolayers are greatly reduced, and only a restricted set of crystallographic symmetry operations are possible, allowing to better isolate and study specific structural properties of HBs. Moreover, the molecular orientation imposed by such confinement promotes © 2012 American Chemical Society

Received: August 26, 2012 Revised: October 16, 2012 Published: October 16, 2012 15586

dx.doi.org/10.1021/la3034325 | Langmuir 2012, 28, 15586−15597

Langmuir

Article

intensity specularly reflected from the surface at an angle β equal to the incidence angle, α. This geometry yields a surface-normal wavevector transfer qz = (4π/λ) sin α. ρ(z) and R(qz) are related through36−38 R(qz)/RF(qz) = |ρb−1∫ (d⟨ρ(z)⟩/dz) exp(−iqzz) dz|2, where RF(qz) is R(qz) for an ideally smooth and abrupt surface and ρb = 3.25 e/Å3 is the bulk electron density of mercury. The angular brackets denote averaging over a surface-parallel plane. ρ(z) is extracted from the measured R(qz) by constructing a physically motivated model for ρ(z), calculating R(qz)/RF(qz) for this model from the equation above, and fitting the resultant expression to the measured R(qz)/RF(qz) to obtain the values of the model’s profiledefining parameters.34−38,44 The surface-parallel structure of the LF is determined by grazing incidence diffraction (GID). α is set here to α < αc ≈ 0.45° (where αc is the critical angle for total external reflection36−38). The detector is scanned out of the reflection plane by an angle 2θd. The scattering vector is then almost surface-parallel, q∥ = (2π/λ)(cos2 α + cos2 β − 2 cos α cos β cos θd)1/2 ≈ (4π/λ) sin(2θd/2). The GID resolution, Δres = 0.018 Å−1, is determined by a set of vertical-plate Soller slits placed in front of the detector. The scattered intensity I(q∥) is measured with a 10 cm long linear position-sensitive detector aligned in the surfacenormal z direction, providing a simultaneous measurement of the intensity distribution in the surface-normal direction at each 2θd position. At 2θd positions where GID peaks occur, this distribution, called a “Bragg rod” (BR), provides information on the molecular length, the molecular tilt, and the tilt’s azimuthal direction in the laterally ordered parts of the LF.30,34,35,37,44 The X-ray measurements spanned a temperature range of 5 °C ≤ T ≤ 25 °C, an XR qz range of 0 Å−1 ≤ qz ≤ 1.7 Å−1, and GID and BR qranges of 0.14 Å−1 ≤ q∥ ≤ 1.7 Å−1 and 0 Å−1 ≤ qz ≤ 1.25 Å−1. Materials and Procedures. Mercury (Merck Co., triple distilled, 99.999% pure, or Bethlehem Apparatus Co., quadruple distilled, 99.9995% pure) and highest-commercial-purity diacids (Aldrich and TCI, >98% pure) were used as received. Diacids of 13 ≤ n ≤ 22 carbons were studied and are denoted hereafter as dn. Examples are shown in Figure 1, along with a simple fatty monoacid, denote here as mn.

reflectivity and grazing incidence diffraction measurements, the method of choice for these studies, the high surface tension of mercury (∼500 mN/m) provides a low surface roughness (∼1.4 Å), allowing a 3-fold larger measurement range, and a commensurately higher resolution, compared to those obtainable on the lower surface tension (72 mN/m) surface of water. The higher resolution is particularly important for monolayers of lying-down molecules, like those studied here, which are only ≲5 Å thick. The study mentioned above31 demonstrated that at a full surface coverage by a monolayer of lying-down diacid molecules a 2D surface-parallel order is formed. The ordering is driven by the HBs at both ends of the diacid molecules. These bonds dominate the order along the molecular axis, while the vdW interaction dominate the order in the moleculenormal direction.31 That study addressed only a single diacid molecule of an even length, n = 16. The present study employs surface-specific synchrotron X-ray methods30,35−38 to explore the dependence of the 2D diacid crystal structure on the molecular length and the surface coverage, ranging respectively from 13 to 22 carbons and from a monolayer to a trilayer. Increasing the molecular length increases the vdW-to-HB strength ratio per molecule, allowing, in principle, to elucidate the relative importance of these interaction in the 2D crystal binding. For all coverages up to the film’s collapse and all molecular lengths, we find only phases of lying-down molecules, exhibiting a full 2D order with essentially the same 2D oblique unit cell. However, the lattice vector along the molecular width and the cell’s dihedral angle decrease slightly with n, while the other lattice vector, associated with the molecular length, increases commensurately with n. Evidence is found for the inclusion of a single mercury atom in each ringlike HB which bind the diacid molecules into long, multimolecular chains.

■

EXPERIMENTAL SECTION

The methods used here are the same as those employed successfully in our previous studies of LFs on mercury.33,34,39−42 A purpose-built Langmuir trough, allowing both X-ray and surface tension measurements, was employed. We now summarize briefly the relevant experimental details. Isotherms. In a surface pressure−area isotherm, π(A), the surface pressure, π = γ0 − γc, is the difference in surface tension (γ) between the bare (γ0 = 490 mN/m) and the film covered (γc) surface, and A is the nominal area per molecule calculated from the mercury’s surface area and the number of molecules deposited. The surface tension was measured by the Wilhelmy plate method,43 using a mercuryamalgamated platinum plate and a film balance based on a linear variable differential transformer (LVDT).34 The KelF trough was attached to a brass baseplate, the temperature of which was controlled to ±0.1 °C by a commercial water circulator. The trough and baseplate resided inside a hermetically sealed aluminum box, in which a slow gas flow of helium (for X-ray measurements) or nitrogen (for surface tension measurements) was maintained throughout the measurements, to minimize oxidation and possible beam damage, which however was not observed here. X-ray Measurements. The surface-specific X-ray measurements employed the Harvard/BNL liquid surface diffractometer at beamline X22B, NSLS, Brookhaven National Laboratory, with X-rays of wavelength λ = 1.517 25 Å. The trough enclosure was mounted on the active vibration isolation unit of the diffractometer’s sample stage. This arrangement eliminated effectively the pickup of vibrations from the environment. The surface-normal electron density profile ρ(z) of the LF-covered mercury surface is determined by X-ray reflectivity (XR) measurements.36−38 The XR curve, R(qz), is the fraction of the incident beam’s

Figure 1. Examples of odd and even diacid and monoacid molecules. The chemical formulas and notations used in this paper, “dn” for a ncarbon diacid and “mn” for a n-carbon monoacid, are also shown. Color code for atoms: carbon = dark gray, hydrogen = light gray, oxygen = red. Standard spreading solutions were prepared with molarities in the range of (3−8) × 10−4 using HPLC grade, 99.9% pure chloroform (Aldrich). Isotherms were measured for pure chloroform to ensure that it evaporates rapidly and neither leaves traces on the mercury subphase nor changes γ from γ0. Following procedures used in our previous studies of mercurysupported LFs,33,34,39−42 the enclosure and trough were cleaned with 15587

dx.doi.org/10.1021/la3034325 | Langmuir 2012, 28, 15586−15597

Langmuir

Article

isopropanol and chloroform at the start of each experiment, the enclosure sealed, and purged for 30 min by a gas flow to remove solvent vapor. The trough was then filled with mercury from a reservoir, mounted on top of the sealed enclosure, through a capillary, and the surface tension of the bare mercury measured to ensure surface purity. The LF was then formed by depositing the spreading solution on the mercury surface in steps of 4−8 μL through a small sealable opening in the enclosure’s top, waiting for the surface tension to relax after each step. This method was preferred over a standard moving barrier compression because of the difficulty to construct a leak-free barrier for mercury.45 The surface tension was monitored throughout the deposition process and stopped upon reaching the desired π or A, at which the X-ray measurements were carried out.

2b supports the former possibility. The very similarly sloping region found for m18 just below the bend was found to correspond to the progressive conversion of the lying-down monolayer phase to a lying-down bilayer phase. In contrast, the conversion of the lying-down phase to a standing-up phase produces the differently looking flat plateau observed for m18 in Figure 2b at 30 Å2/molecule ≤ A ≤ 60 Å2/molecule. Thus, the absence of such flat regions in our isotherms implies the absence of standing-up phases. Note also the short high-slope part of the m18 isotherm in Figure 2b at its low-A end, indicating a full surface coverage by a low compressibility monolayer of standing-up molecules.30,33 A similar feature is not observe at the low-A end of the diacid isotherms, further supporting the absence of a phase of standing-up molecules. These conjectures are supported by the X-ray reflectivity results discussed below. The fast-varying part of the isotherms to the right of the bend was reasonably well fitted in several studies of mercurysupported LFs31,33,34,40,46−48 by the Volmer equation of state of an ideal 2D gas of finite-area (A0) molecules,33,49 π(A − A0) = kBT. However, the fit of our diacid isotherms by the Volmer equation (dashed curves in Figure 2a) is poor, for reasons discussed below. Thus, the exclusion areas due to the molecular size, Aiso 0 , shown in Figure 2c, were approximated here by the intersection of short line sections fitted to π(A) above and below the bend, as shown in dashed lines for d22 in Figure 2a. The values thus obtained follow, within a conservative estimated error bar of ±8 Å2/molecule for each point, the linear n-dependence (red solid line, Figure 2c) expected from the molecules’ constant widths and linearly increasing lengths with n. The A0 values derived from the isotherms are further discussed below in comparison with the much more accurate molecular area values derived from the GID measurements. Surface-Normal Structure. Coverage Dependence. Representative measured R(qz)/RF(qz) curves are shown in Figure 3a for different nominal coverages, A, marked by arrows on the generic π−A isotherm in the inset. The measurement of each XR started 10 min after completion of the film deposition directly to the coverage indicated. The multislab model employed successfully in our previous mercury-supported LF studies33,39−42,46,47,50 to model the surface-normal electron density profile ρ(z) yields good fits (lines) to the measured curves (symbols). The multislab model mimics the depthdecaying layering at the surface of mercury51−53 by six fixedwidth slabs. The same Hg-defining parameters were used in the fit of all reflectivity curves, allowing only the roughness to vary between curves. This procedure allowed to better define the underlying mercury profile, which is of particular importance considering that the restricted q-range measured did not reach the mercury layering peak at q ≈ 2.2 Å−1. The overlying LF was modeled by the minimal number of slabs required to achieve a good fit to the measured R(qz)/ RF(qz) curves, each slab representing a monolayer of lyingdown molecules. The ρ(z) profiles obtained from the fits are shown in Figure 3b, with the LF residing at z < 0 and the mercury at z > 0. At the isotherm’s bend (curve A in Figure 3a) a single lyingdown monolayer of thickness 4.5 ± 0.25 Å yields a good fit to the measured R(qz)/RF(qz) for all n, although the existence of small bilayer patches (up to 10−15 area %) of lying-down molecules cannot be ruled out. The electron densities found for the monolayers, 0.31 ± 0.01 e/Å3, are in good agreement with the 0.30 e/Å3 obtained from fits of the XR curve of mercury-

■

RESULTS π−A Isotherms. The π−A isotherms measured for different-length diacids are shown in Figure 2a. All exhibit the

Figure 2. (a) Measured (symbols and eye-guiding lines) and Volmer equation fitted (dashed lines, d13 and d21) π−A isotherms of the molecules studied here. The black arrow marks the position of the bend, discussed in the text, for d15. The dashed lines on the d22 isotherm show the determination of A0. (b) Measured (symbols and eye-guiding line) π−A isotherm for an 18-carbon fatty acid, m18, with its fit by the Volmer equation (dashed line). (c) Exclusion areas due to molecular size, as derived from the isotherms, Aiso 0 (symbols), and their linear fit (solid line).

same generic shape: a flat, low pressure (π ≈ 0 mN/m) plateau at a (chain-dependent) low coverage (≥140 Å2/molecule for d13, ≥200 Å2/molecule for d22), a high pressure (π ∼ 30−40 mN/m), mildly sloping region at high coverage (25−60 Å2/ molecule), and a narrow A-range of fast-increasing π(A) with decreasing A, which connects these two regions. Our previous studies of mercury-supported LFs31,33,34,40,46−48 identify the bend where the fast-increasing part of the curve levels off, marked for d15 in Figure 2a by a black arrow, as the point where the surface-parallel-oriented molecules of the twodimensional gas phase (A ≥ 200 Å2/molecule) become packed against each other upon decreasing A. At this point the surface is covered by a dense monolayer of flat-lying molecules. As the area occupied by a lying-down molecule increases with the molecular length, the bend should shift to higher A’s with increasing n, as indeed observed in Figure 2a. For A smaller than the position of the bend the LF must consist of either a multilayer of lying-down molecules, or a monolayer of standing-up molecules, to accommodate the smaller average surface area per molecule. The comparison of our isotherms with the previously studied m18 isotherm33,46 shown in Figure 15588

dx.doi.org/10.1021/la3034325 | Langmuir 2012, 28, 15586−15597

Langmuir

Article

reproduce the main features of the measured curves well enough to demonstrate that the main structural features of the diacid film, such as the number of layers as well as their thicknesses and densities, have been faithfully captured. Time Dependence. The measurement of all XR curves discussed above started 10 min after the completion of the deposition of the film to the indicated molecular areas, to allow full solvent evaporation after the last deposition. However, similar to chiral fatty acids,41 the LFs of diacids exhibit also a time-dependent phase sequence, where an LF deposited at a nominal A undergoes a slow transition with time to a thicker LF. Such a time-dependent phase sequence is shown in Figure 4 for a d22 LF deposited with a nominal coverage of A = 126

Figure 3. (a) Measured, Fresnel normalized, X-ray reflectivity curves (symbols), R(qz)/RF(qz), and their slab-model fits (lines), for the indicated molecular lengths n and the coverages marked by arrows in the isotherm shown in the inset. These arrows correspond to a monolayer (A), a bilayer (B), and a trilayer (C) of surface-parallel molecules. (b) Electron density profiles ρ(z) derived from the fit (solid lines) and the contribution of the mercury to the density at z ≤ 0 (dashed lines). Without loss of generality, z = 0 is chosen as the dividing plane between the first Hg-representing slab and the first diacid-representing slab, both in their roughness-unsmeared state. The vertical dash-dotted lines mark schematically the boundaries of the individual layers within the LF. The curves in (a) and (b) are shifted from each other for clarity.

supported octadecanoic acid at its isotherm bend.33 The surface roughness values obtained from the fit, 1.4 ± 0.3 Å, agree with that of the bare mercury surface51−53 and those of similar mercury-supported LFs.33,34,41,46,48,50 The reflectivity curves measured at the higher-coverage B position (curves B in Figure 3a) require two slabs (in addition to those of the mercury) to fit the measured R(qz)/RF(qz) profiles, each as thick as the monolayer found in the A region. The LF here consists therefore of a bilayer of lying-down molecules, with the density of the bottom layer, 0.30 ± 0.01 e/ Å3, corresponding to a full coverage, and that of the top layer, ∼0.065 e/Å3, corresponding to a coverage of ∼24% of the surface area. The highest-coverage R(qz)/RF(qz) curves, at the C position (curves C in Figure 3a), require three slabs for achieving a good fit, indicating that the LF at this position consists of a trilayer of lying-down molecules. However, the layers here differ from those at the A and B positions. First, the two bottom layers here require a significantly higher density, 0.53 ± 0.3 e/Å3, for achieving a reasonable fit. This is mostly dictated by the high peak observed in R(qz)/RF(qz) at qz ≈ 1 Å−1. Such a high density cannot be achieved without a significant inclusion of mercury atoms in these layers. The fit also yields a somewhat reduced combined thickness, 8.2 ± 0.3 Å, for the two bottom layers, dictated mostly by the width of the qz ≈ 1 Å−1 peak. We discuss these features further below. We wish to stress that these fits employ the same Hgdefining parameters for all R/RF curves (roughness excepted). Only the diacid-film-defining parameters and Hg roughness were allowed to vary individually for each diacid film. This procedure allowed to minimize the overall number of fit parameters and concentrate on the coverage-dependent variations in the diacid film’s structure. These restrictions yield, admittedly, less than perfect fits. However, the fits

Figure 4. (a) Fresnel normalized, X-ray reflectivity curves (symbols), R(qz)/RF(qz), and their slab-model fits (lines), for a d22 LF, deposited at a nominal A = 126 Å2/molecule, and measured at the indicated start times (hours:minutes) after deposition. (b) Electron density profiles ρ(z) (solid lines) obtained from the model fits to the measured R(qz)/ RF(qz) curves in (a) and the contribution of the mercury to the density at z ≤ 0 (dashed lines).

Å2/molecule, slightly less than the bend in the isotherm in Figure 2. Time is measured from 10 min after completion of the deposition, and each XR curve takes 50 min to measure. As observed in Figure 4, the LF is found to show a small, ∼20% coverage by a bilayer after 1−2 h, with the full conversion to a bilayer coverage achieved after ∼5 h. A fully developed trilayer is found ∼10 h after deposition, and no further changes are observed in R/RF after that time. The high densities of the two bottom layer of the trilayer at ≤10 h coincide with those of the LFs deposited directly as trilayers, i.e., the type C curves in Figure 3. This implies, in turn, that the incorporation of mercury atoms into the LFs proceeds alongside the LF evolution from a monolayer to a trilayer. The rates of the two processes are roughly equal, as demonstrated by the profile measured 5:50 h after deposition. Here the top layer is only about 60% developed, as compared to the 10 h profile, and so is the density of the two bottom layers. The time evolution and phase sequence found here for the diacid LFs closely resemble those observed for chirally modified fatty monoacid LFs on mercury.41,48 The main difference between the two is the presence of phases of standing up molecules in the phase sequence of the chiral monoacids and their absence from the phase diagram of the diacids studied here. 15589

dx.doi.org/10.1021/la3034325 | Langmuir 2012, 28, 15586−15597

Langmuir

Article

Surface-Parallel Structure. For each of the diacid molecules studied here the observed GID peaks can be grouped into two: low-q∥ peaks, q∥ ≤ 1 Å−1, shown in Figure 5,

C in the inset to Figure 3a), since for a monolayer phase the GID intensities, particularly at high q∥, were extremely weak. However, except for the intensity, no changes with coverage were observed, within our resolution, in the GID patterns. An inspection of the GID patterns reveals that for each molecule all low-q∥ peak positions are integer multiples of the lowest-q∥ position, indicating that all are harmonics corresponding to a single lattice spacing. The derived values of the spacings, and their linear increase with n, show that the low-q∥ peaks are associated with the molecular length. The high-q∥ peaks follow a different pattern. Here the first peak appears at the same position for all molecular lengths: q∥ ≈ 1.3 Å−1, while the higher-order peaks within this group move closer to the first peak with increasing n. The fixed position of the first peak for all n identifies these peaks as corresponding to the common width of all molecules, while the bunching of the higher-q∥ peaks with n indicates that they involve both the width and length of the molecules. Indeed, all peaks observed can be indexed in an oblique single-molecule unit cell, the dimensions of which correspond to the length and width of the molecule. The unit cell is shown as an inset in Figure 5, and its dimensions, angle, and area for each n are listed in Table 1. The crystalline coherence length, ξ, of the LF’s 2D order can be calculated from the GID peaks’ width by the Debye− Scherrer formula.54 As we reported recently,55 the calculations yield for our diacid LFs ξ∥ = 410 ± 20 Å in the molecular length b direction of the unit cell and a smaller ξ∥ = 180 ± 15 Å normal to it, in the molecular width direction. These anisotropic coherence lengths are of the same order as the ξ ≈ 200 Å found for the mercury-supported LF of a poorly ordering ionic liquid.42 They are, however, much smaller than the isotropic ξ ≫ 1000 Å measured for purely vdW-interacting melt-supported surface-frozen alkane monolayers44 and mercury-supported LFs of surface-normal fatty acid molecules at molecular areas just before collapse.46 Possible reasons for this difference and for the anisotropy of ξ are suggested in the Discussion section. Additional details of the LF’s 2D order can be obtained from the equal-intensity (q∥, qz) contour plots. Figure 7b shows such plots for a bilayer of d21, at position B of the isotherm in the inset to Figure 3. The BRs of all low-q∥ peaks, represented by the (02) BR in Figure 7c, peak at qz = 0 Å−1, while those of the high-q∥ reflections, represented by the (10) BR in Figure 7c, are centered at nonzero qz values. BRs peak at qz = q∥ cos Ψi tan t, where q∥ is the GID line position, t is the molecular tilt angle from the surface normal, and Ψ is the azimuthal angle of the tilt direction relative to the q∥ component of the scattering vector.30,44,56 The contour plots and BR in Figure 7b,c show, therefore, a finite tilt in the azimuthal direction of the molecular width and none in the direction of the molecular length. Note also that for an oblique 2D unit cell like ours a fully split GID pattern should be observed with the three lowest-order reflections having different qz-peak positions.30,56 The qzpeaks of these BRs are related by30 qaZ = qbZ + qcZ. Since here = 0 Å−1, we should have q(10) = q(11) q(01) Z Z Z , as indeed observed in Figure 7b. An estimate for the thickness of the ordered part of the LF and for the tilt magnitude can be obtained from the widths and peak positions of the BRs,30,44,56 e.g., those shown in Figure 7c. The qz ≈ 0.7 Å−1 peak position of the (10) BR at q∥ = 1.3 Å−1 yields t = tan−1(0.7/1.3) = 28° and d = π/Δqz ≈ 9 Å, where Δqz ≈ 0.35 Å−1 is the half width at half-maximum of both the (02) and (10) BRs. A full profile fit by the theoretical BR44

Figure 5. Measured background-subtracted low-q∥ GID patterns (symbols) for monolayers of the listed compounds and their fits (lines) by a single Gaussian for each peak. All peaks of each compound are indexed, as shown for d16, in the oblique unit cell in the inset.

and high-q∥ peaks, q∥ ≥ 1.25 Å−1, shown in Figure 6. The diffraction patterns for the diacid layers are well described by an oblique unit cell (see inset to Figure 5), with a ≠ b and γ ≠ 90°. The peak positions and corresponding lattice parameters a, b, and γ are listed in Table 1. The GID patterns shown were measured for a complete bilayer phase (position between B and

Figure 6. Measured background-subtracted high-q∥ GID patterns (symbols) for monolayers of the listed compounds and their fits (lines) by a single Gaussian for each peak. All peaks of each compound are indexed, as shown for d22, in the oblique unit cell in the inset of Figure 5. 15590

dx.doi.org/10.1021/la3034325 | Langmuir 2012, 28, 15586−15597

Langmuir

Article

Table 1. GID-Derived q∥ Peak Positions, in Å−1, for the (h,k) Reflections Shown in Figures 5 and 6a q∥(h,k), d(h,k) (Å−1, Å) n

(01)

(02)

(03)

13

0.287 21.9 0.242 26.0 0.232 27.1 0.203 31.0 0.187 33.6

0.574 11.0 0.488 12.9 0.465 13.5 0.407 15.4 0.376 16.7

0.862 7.29 0.730 8.61 0.698 9.09 0.611 10.3 0.565 11.1

16 17 21 22

(04)

(10)

(11)

(12)

(13)

a (Å)

b (Å)

γ (deg)

Am (Å2)

1.388 4.53 1.370 4.59 1.356 4.63 1.338 4.70 1.335 4.71

1.535 4.09 1.480 4.25 1.455 4.32 1.417 4.43 1.402 4.48

1.620 3.88 1.630 3.85

0.815 7.71 0.754 8.33

1.293 4.86 1.298 4.84 1.293 4.86 1.291 4.87 1.293 4.86

5.009 0.006 4.98 0.02 4.950 0.004 4.936 0.004 4.924 0.005

22.52 0.03 26.5 0.1 27.53 0.02 31.26 0.03 33.80 0.04

103.9 0.2 103.2 0.5 101.0 0.1 99.5 0.1 99.2 0.1

109.5 0.2 128.5 0.9 133.8 0.2 152.2 0.2 164.3 0.3

1.516 4.14 1.492 4.21

The corresponding lattice spacings are listed below each peak position. a, b, γ, and Am are respectively the unit cell’s base vector lengths, dihedral angle, and area. The error bars are listed below each of these three values.

a

of the fit with the measured curve, although the fit itself was done on the raw, unsmoothed, data. BRs measured for a few other molecular lengths yielded d and t values that coincided with those above within their combined error bars. The apparent tilts found here and in previous studies31,33,46 are due to the form factor arising from the way that the surfaceparallel molecules of the bilayer are stacked: those of the top layer reside in the hollows of the bottom layer, as shown in Figure 8b. For molecules having a circular cross section, such

Figure 8. Schematic view of the diacid LF’s ordering. (a) Top view of the self-assembly of the diacid molecules into a 2D-ordered LF. The ring-motif HBs at both ends of the molecule form long polymer-like chains (highlighted in yellow), which assemble side by side to form the 2D ordered LF. The periods d1 and d2 correspond to the q-positions of the (01) and (10) GID peaks, respectively. The single-molecule unit cell of Figure 5 is shown in blue. (b) The surface normal stacking of a two-layer diacid LF, showing the origin of the observed tilt. Figure 7. (a) Measured background-subtracted GID pattern (circles) for a bilayer of d21 molecules at room temperature, with its fit by a single Gaussian per peak (line). The high-q∥ peaks (red) are multiplied by 30. (b) Equal intensity contour plot of the GID pattern shown in (a). The top end of the high-q∥ peaks is masked by the rising intensity of the mercury bulk scattering. (c) Measured background-subtracted Bragg rods of typical low-q∥, (02), and high-q∥, (10), GID peaks shown in (a,b). The finite detector length truncated the (10) rod at both its qz-ends. The heavy black line for (10) was generated from the measured values (open circles) by a 20-point “moving window” smoothing. The apparent short-period oscillations in the (10) rod are an artifact, which, however, does not adversely effect on the analysis’ results. The fits by the theoretical rod profiles are shown in red lines.

stacking yields an effective 30° tilt from the surface-normal for the line connecting the center of a molecule in the bottom layer with the nearest top-layer molecule. A deviation of the cross section from a circular shape modifies the apparent t value. These deviations are the likely causes for the few degrees lower t for our diacid LFs and for LFs of tetracosanoic monoacid on mercury.33,46

■

DISCUSSION The mercury surface confines the lying-down molecules to form a quasi-2D structure, without imposing epitaxial constraints on relative lateral positions and azimuthal orientation. This planar alignment of the molecules promotes molecular self-organization into structures dominated by the lateral interactions of their various moieties, in particular the HBs between the molecules’ craboxyl end groups. However, as for all 2D structures, the X-ray GID patterns presented above, although having considerably more peaks than those of most liquid-supported interfacial layers,30,31,42,57 have still much

yields a very good agreement (red lines in Figure 7c) with the measured BRs and corroborates these estimates, with d = 8.3 ± 0.9 Å and t = 28 ± 0.5° for the curves in Figure 7c. The very low intensity of the high-q∥ peaks required a 20-point movingwindow smoothing (heavy black line in Figure 7c) of the measured (10) curve (circles) to bring out the good agreement 15591

dx.doi.org/10.1021/la3034325 | Langmuir 2012, 28, 15586−15597

Langmuir

Article

fewer peaks than the diffraction patterns of 3D single crystals or powders of the same compounds, which may contain hundreds to thousands of lines.24,58−61 Thus, the elucidation of fine details in the headgroup and the chain ordering from our GID patterns is more prone to ambiguities and uncertainties. Using additional information from relevant 2D or 3D system can, however, enhance considerably the confidence and validity of the extracted structure. With these caveats in mind, we now proceed to offer a molecular-level view of the structure of the LFs studied here in terms of the orientations, relative positions, and bonding of the lying-down molecules. The elucidation of the LFs’ structure can be greatly assisted by considering the structure of 3D diacid and acid crystals.23,24,58−62 The dominant structural motif in the 3D crystals is the strong ring-type double HBs formed at both ends of the molecule. These yield long, linear chains of hydrogenbonded molecules,24 which mimic linear polymer chains. A schematic view of the formation of such chains, and their subsequent self-assembly into a 2D-ordered sheet, are shown in Figure 8. The same ordering was found to dominate the structure of the mercury-supported d16 LF,31 the only previously studied LF of a diacid. In that study it was shown that the vdW attraction of such long chains is strong enough to promote the emergence of order in the chain-normal direction, while the molecular units of the chain form the order in the chain-parallel direction. Thus, a full 2D order emerges.31 For monoacid LFs,46 the single carboxyl of each molecule forms only dimers. The dimers are methyl-terminated at both ends and thus cannot form long molecular chains. For monoacid molecules shorter than 24 carbons the large entropy associated with the dimers’ free, nonbonded, methyl end groups effectively prevents the emergence at room temperature of order in the direction normal to the extended dimer. For the diacids studied here, however, a full 2D order, including both chain-normal and chain-parallel directions, is observed already for the shortest diacid studied, d13, and very likely occurs for considerably shorter diacids as well. The structure shown schematically in Figure 8a renders the interactions dominating the LF’s structure along, and normal to, the molecular chains considerably different. As we discuss below, the carboxyl−carboxyl bonds are mercury-modified and mostly covalent in nature. This, and the carbon−carbon bonds, render the bonding along a chain almost completely covalent and thus strong and directional. In the chain-normal direction, however, a considerably weaker, nondirectional, vdW interaction prevails. This difference in the bonding nature and strength is responsible for the order’s coherence length ξ being considerably shorter in the softer-interaction chain-normal direction than in the stronger-interaction chain-parallel direction, as found above and in Tamam et al.55 The different bonding nature and strength in the two directions also causes the thermal expansion coefficient in these diacid LFs to be highly anisotropic.55 The thermal expansion coefficient along the chain, α∥ ≈ 1.8 × 10−5 K−1, dominated by covalent bonding, is found to be 17-fold smaller than that perpendicular to the chain, α⊥ ≈ 3 × 10−4 K−1, dominated by vdW interaction.55 The unit cell in Figure 8a shows that the length of the b lattice vector is determined by the molecular length, while that of a is determined by the molecular width, d2 = a sin γ. The variation with n of a, b, and other unit cell properties listed in Table 1 are plotted in Figure 9. The a values in Figure 9a yield an almost constant effective chain widths of d2 ≈ 4.8−4.9 Å,

Figure 9. Properties of the unit cell, shown in Figure 5. (a) GIDderived a lengths (circles) with linear fit (line). (b) Same as (a) but for b, corresponding to the effective molecular lengths. The calculated lengths (squares) and their linear fit (dashed line) are also shown. Note the almost constant downshift by 2.5 Å. (c) GID-derived cell angle. (d) Molecular area obtained as the GID-derived unit cell area (circle) and as the exclusion area (squares) derived from the isotherms in Figure 2, with their linear fits.

which exceed slightly the 4.7 Å interchain spacing measured for 3D alkane crystals63 but agree well with the 4.83 Å interchain spacing measured for 3D crystals of monocarboxylic hexadecanoic acid64 and for mercury-supported monolayers of lying-down molecules of monocarboxylic fatty acids.33,46 The small, 1.6%, linear decrease in a with increasing n, found here, seems reasonable in view of the equal widths of the alkyl moiety for all n and the increase with n of the interchain vdW attraction per molecule, which drives toward a closer packing. The XR measurements discussed above reveal a decrease in the effective layer thickness from 4.5 ± 0.25 Å for the monolayer and bilayer at positions A and B on the isotherm in the inset to Figure 3 to a combined 8.2 ± 0.3 Å for the Hg-containing bottom two layers of the trilayer at position C. This decrease in the effective layer thickness may well result from a compression effect similar to that of a upon increasing the vdW attraction when the number of neighboring molecules increases in the surfacenormal direction, and heavy Hg atoms are incorporated into the two bottom layers. The unit cell’s angle γ is also found to decrease roughly linearly with n with a slope of dγ/dn = −0.55 ± 0.09°/CH2, showing that the increasing vdW attraction between the molecules drives toward a less oblique unit cell. A linear fit to the measured b values in Figure 9b yields b = (6.6 ± 0.9) + (1.22 ± 0.07) × n Å. The slope of this line is somewhat smaller than, but still agrees within its error bar with, the 1.265 Å axis-projected C−C bond length,30,65−67 which is the expected increase in the molecular length upon increasing n by 1. Our b-lengths can be compared with the lengths of the corresponding diacid molecules, as calculated from the measured lengths of their monoacid counterparts46,65−67 by subtracting the projected length of the CH3 moiety67 and 15592

dx.doi.org/10.1021/la3034325 | Langmuir 2012, 28, 15586−15597

Langmuir

Article

orientations and nonhydrated carboxyls. Our linear bond geometry is closer to that of 3D crystalline metal−carboxylate salts, and Langmuir−Blodgett multilayers of metal−fatty acid complexes.71−74 Although no studies of such mercury-including systems are available in the literature, in general four types of metal−ligand complexes have been identified.75,76 For heavier elements like Pb and Cd a bidentate (chelate or bridging) bonding motif was observed, where a single metal ion is included in the carboxyl−carboxyl complex, forming a bond which is mostly or fully covalent.75−77 A similar motif may well occur for the similar diameter and atomic number mercury. Note also that the inclusion of a metal ion in a carboxyl− carboxyl bond often distorts the bond from the planar conformation of a simple carboxyl−carboxyl HB.73,74 Thus, great care must be taken when drawing structural conclusion for metal−organic complexes from the known structure of the corresponding metal-free complex. This point is important for understanding the obliquity of the unit cell, which we now discuss. To understand the difficulties in elucidating the molecular conformations leading to the oblique unit cell observed here, consider first the diacid odd (d15) and even (d16) molecules shown in Figure 1. Since the carboxyls are slightly rotated within the molecule’s backbone plane from the molecular axis, a bond between two molecules necessarily entails a lateral parallel shift of their axes by some distance.24 In odd-n diacids the two carboxyls point to the same side of the molecular axis, while in even-n diacids they point to opposite sides. Thus, molecules binding at the two ends of an odd molecule shift both to the same side of the molecular axis, while those binding to an even molecule shift to opposite sides. The consequent difference in the chain morphology of even and odd diacids is shown in Figures 10b and 10g, respectively. For both chain morphologies, close packing of chains to form 3D crystals or 2D sheets yields too-close contacts of the bond complexes of adjacent chains, as shown in the highlighted regions in Figures 10c and 10h. For even diacids a longitudinal relative shift of adjacent chains will move all neighboring bonding complexes away from each other (Figure 10d) and lead to an oblique unit cell, as shown in Figures 10d and 10e. However, as the yellowhighlighted regions in Figure 10i demonstrate, for odd diacids such a longitudinal shift will indeed relieve the contacts at one end of the molecule but bring the bonding complexes at the other end to an even closer contact. Thus, for 3D crystals of odd diacids the most common way of relieving the close contacts is by a rotation of the bond planes in adjacent chains away from the backbone plane,24 as shown in Figure 10j. This effect rotates the planes of the two carboxyls at an odd molecule’s ends out of coplanarity by ∼60° (Figure 10k), twisting the molecule around its long axis. The higher energy associated with the twisted molecular conformation of odd diacids, relative to that of the planar conformation of even diacids, leads to the well-known odd−even effect in the melting temperatures of diacids.24 Due to this relief-by-twist effect, a relative longitudinal chain shift does not occur in odd diacids, and their unit cells stay rectangular. Two different twisted conformations are found in 3D odd diacid crystals, designated α and β, which differ in whether only one or both carboxyls are twisted away from the alkyl backbone plane. Both conformations yield a rectangular unit cell which is larger than that of even diacids. For example, for the α form shown in Figure 10e only one carboxyl plane is rotated, and the alternating carboxyl rotations along the chain yield a large, four-molecule,

adding instead the projected length of the additional COOH moiety.46,65−67 The values thus calculated (Figure 9b, squares) are found to lie below the measured b values by a roughly nindependent ∼3 Å. This discrepancy can plausibly be assigned to the incorporation of a mercury atom into the carboxyl− carboxyl bond, as has already been suggested for monocarboxylic fatty acid LFs on mercury.33,46 The ∼3 Å length discrepancy per diacid molecule corresponds well to the atomic diameter of mercury, ∼3.2 Å.68 The reflectivity results discussed above, yielding a layer density higher than that of a pure diacid layer, also support this conclusion, although it is likely that additional mercury atoms are incorporated also into possible disordered parts of the LFs, in particular into boundaries between crystalline domains. Additional support for the inclusion of mercury in the carboxyl−carboxyl bond is obtained from the observation that for monoacids46 the single bond per dimer results in a ∼3 Å length discrepancy per dimer, while for the diacids the single bond per molecule yields the same ∼3 Å length discrepancy, however, per molecule. The molecular areas calculated for the unit cell are shown in = (36 ± 7) + (5.7 ± Figure 9c, along with its linear fit, AGID 0 0.4) × n Å2/molecule. This line has the same slope as, but is upshifted by a roughly constant ∼18 Å2 from, the molecular area calculated from the bend of the isotherm, Aiso 0 = (25 ± 7) + (5.4 ± 0.4) × n Å2/molecule. To understand this upshift, note that at the isotherm’s bend the monolayer still does not include Hg atoms, as demonstrated by its low ρ ≈ 0.3 e/Å3 obtained from XR fits and observed in the density profile of the type A curve in Figure 3b. By contrast, the GID unit cell includes Hg atoms in the carboxyl−carboxyl bond, making the lattice vector b ∼3 Å longer than an extended diacid molecule, as discussed above. Correspondingly, the area per molecule will be 3 × 4.85 ≈ 14.5 Å2 larger. The remaining ∼4 Å discrepancy may result from the presence of small bilayer patches already at the bend, thus reducing the effective area/molecule observed. This suggestion is supported by the strong rounding of the bends observed in all isotherms in Figure 2a and the large deviation of the isotherms in the bend region from the Volmer equation. These properties of the present isotherms are in contrast with the significantly sharper transition from a mono- to bilayer phase, and the much better agreement with the Volmer equation, observed in other mercury-supported LFs, e.g., alkanes34 and fatty acids,33,46 also exhibited by the m18 LF in Figure 2b. The formation of such patches may be promoted by the tendency of the diacid LFs to spontaneously self-aggregate into multilayers, as observed in the time-evolution measurements discussed above. The main influence of such patches on XR curves is to push up the intensity at ∼0.75 Å−1 and drive the shape of the XR curve from type A in Figure 3a to type B. For small patch coverages, the intensity increase will be, however, small. It will affect only the close vicinity of the dip at ∼0.75 Å−1, making it slightly shallower and thus difficult to distinguish in fits from the type A XR curve of a pure monolayer. The exact position and coordination of the mercury atom, proposed to be included in the carboxyl−carboxyl HB, cannot be refined with confidence from our sparse GID patterns, as discussed above. Nevertheless, some insight may be gained from studies of metal−ligand complexes appearing in the literature. Regrettably, the only study to date addressing specifically the mercury−carboxyl ligand in a quasi-2D context69,70 deals with water-supported monoacid LFs, where the surface-normal molecular orientation and hydrated carboxyl headgroups greatly differ from our surface-parallel molecular 15593

dx.doi.org/10.1021/la3034325 | Langmuir 2012, 28, 15586−15597

Langmuir

Article

however, that for our LFs the bond complex includes a mercury atom, which is likely to change the bond conformation from that of pure diacids, as indeed observed for several metal− carboxyl ligands.73,74 Moreover, the molecules in all 3D odd diacid crystal phases distort from the planar, all-trans, molecular conformation of 3D even diacid crystals.24,59,78,79 Although in most 3D crystalline phases the molecules are linear, phases denoted as B phases, where gauche conformations cause strong deviations from linearity, have also been reported for both diacid and monoacid crystals.78,80 Such conformations at one end of an odd diacid molecule can switch the orientation of one carboxyl group to point to the side of the molecular axis opposite that of the carboxyl at the other end of the molecule, as shown in Figure 10m. An additional twist of order ∼60° around the molecular axis would then bring the two carboxyls almost to coplanarity, while still pointing to opposite sides of the molecular axis. This carboxyl conformation is similar to that of an even diacid molecule and could lead to the same unit cell for this distorted odd molecule as that of an even extended molecule. Such a scenario is admittedly highly speculative and not without problems. For example, the gauche conformation will shorten the projected length of the molecule as compared to that of an extended one and increase the layer thickness by the twisting and bending of the molecules. These effects are not observed in our GID and XR data. Clearly, further experimental and theoretical input is needed to resolve this issue. It is perhaps appropriate to comment on the relation of our results to those of similar mono- or few-molecule-layers, weakly adsorbed (physisorbed) on solid surfaces. Regrettably, no studies of diacid monolayer ordering at solid interfaces are available to the best of our knowledge. However, the ring-motif hydrogen bonding,19 which plays an important structural role in the present study, also dominates the structure of monoacid monolayers, physisorbed on graphite.81 In contrast with the long polymer-like chains formed by diacids, as found here, only dimers can be formed by monoacids, as found for monolayers on graphite81 and on mercury.33,46 The packing of these dimers is dependent not only on the dimer−dimer interaction but also on the dimer−substrate interaction.82,83 Thus, the interaction with the hexagonally ordered graphite substrate induces an odd−even structure difference,81 where odd carbon number molecules order in a rectangular unit cell containing two dimers, while even carbon number molecules order in an oblique unit cell containing a single dimer. Both structures are largely commensurate with the underlying graphite’s structure. Such odd−even effects are not observed in the GID patterns of the present mercury supported diacid monolayers. The structural motifs of mercury-supported monoacid monolayer33,46 also differ from those on solid substrates. Here, a one-dimensional, smectic-like order of dimers is found for short molecules and a full 2D dimer order for long molecules. The latter exhibits a complex phase diagram of several different 2D unit cells, depending on the number of layers (single or double), molecular length, and temperature. A further, more remote, example is normal alkane adsorbed films. On mercury such films show no lateral order for any length, temperature, or surface coverage.34,50 On graphite,82,84 however, a rich lateral phase diagram is found, comprising a clear odd−even effect of different structure for different parities over different length ranges. These few examples demonstrate the important role played by the substrate’s order in the determination of an adsorbed film’s structure even when the adsorption is weak. Further work

Figure 10. Top view of possible molecular assembly and organization of (a−e) even (d16) and of (f−j) odd (d15) diacid molecules into a monolayer. (k) Terminal carboxyls conformation for twisted odd diacid molecules. The dashed line represents the long axis of the molecule. (l) An all-trans d15 molecule. (m) d15 molecule with one gauche conformation on the third carbon from top. Note the consequent opposite orientation of the terminal carboxyl as compared to the all-trans molecule in (l).

rectangular cell (blue in Figure 10j). This unit cell greatly differs from the smaller, single-molecule, oblique cell of even diacids, shown in blue in Figure 10e. Intriguingly, the odd−even structural dichotomy of 3D diacid crystals is not observed for our quasi-2D diacid LFs: all different-length, different-parity diacid LFs studied here exhibit the same oblique unit cell deduced above from the structure of 3D even diacid crystals. This is demonstrated unambiguously by the GID measurements shown in Figures 4 and 5 and by the continuous, monotonic variation with n of all the unit cell properties displayed in Figure 9. Moreover, the fact that the structure of even diacids’ LFs conforms to the conclusions drawn above from the detailed 3D crystal structures supports the validity of these conclusions. At the same time, it also indicates a need to explain why the structure of the odd diacids’ LFs does not conform to the same conclusions, which predict a larger and rectangular unit cell for odd diacid LFs. A definite answer to this question would clearly require an atomic-level determination of the bond-complex structure. This, and the conformations of the odd and even chains, cannot be crystallographically refined from our too-few GID peaks, as discussed in the opening paragraph of this section. We note, 15594

dx.doi.org/10.1021/la3034325 | Langmuir 2012, 28, 15586−15597

Langmuir

Article

on solid-supported films, outside the scope of this paper, is reviewed by Bruch et al.82 and by Arnold and Clarke.83 Our choice of a liquid mercury substrate, having no long-range lateral order, was designed to minimize the imposition of the substrate’s order on the diacid film. To gain a deeper insight into the role of the substrate specifically for diacid films, the present study must await the advent of corresponding studies of diacid films on solid substrates, currently unavailable.

(3) Seanger, W. Principles of Nuclic Acid Structure; Springer: Berlin, 1984. (4) Whitfield, T. W.; Martyna, G. J.; Allison, S.; Bates, S. P.; Crain, J. Liquid NMA: A surprisingly realistic model for hydrogen bonding motifs in proteins. Chem. Phys. Lett. 2005, 414, 210. (5) Toney, M. F.; Howard, J. N.; Richer, J.; Borges, G. L.; Gordon, J. G.; Melory, O. R.; Wiesler, D. G.; Yee, D.; Sorensen, L. B. Voltagedependent ordering of water molecules at an electrode-electrolyte interface. Nature 1994, 368, 444−446. (6) Chandler, D. Interfaces and the driving force of hydrophobic assembly. Nature 2005, 437, 640. (7) Desiraju, G. R. Hydrogen bridges in crystal engineering: interactions without borders. Acc. Chem. Res. 2002, 35, 565−573. (8) Desiraju, G. R. Crystal engineering: A holistic view. Angew. Chem., Int. Ed. 2007, 46, 8342−8356. (9) Moulton, B.; Zaworotko, M. J. From molecules to crystal engineering: Supramolecular isomerism and polymorphism in network solids. Chem. Rev. 2001, 101, 1629−1658. (10) Baptiste, A.; Gibaud, A.; F, B., J.; Wen, K.; Maoz, R.; Sagiv, J.; Ocko, B. M. X-ray, micro-Raman, and infrared spectroscopy structural characterization of self-assembled multilayer silane films with variable numbers of stacked layers. Langmuir 2002, 18, 3916−3922. (11) Ivasenko, O.; Perepichka, D. F. Common lessons from X-ray crystallography and scanning tunneling microscopy. Chem. Soc. Rev. 2011, 40, 191−206. (12) Srinivasan, C.; Mullen, T. J.; Hohman, J. N.; Anderson, M. E.; Dameron, A. A.; Andrews, A. M.; Dickey, E. C.; Horn, M. W.; Weiss, P. S. Scanning electron microscopy of nanoscale chemical patterns. ACS Nano 2007, 1, 191−201. (13) Mantooth, B. A.; Weiss, P. S. Fabrication, assembly, and characterization of molecular electronic components. Proc. IEEE 2003, 91, 1785−1802. (14) Kelley, A. T.; Ngunjiri, J. N.; Serem, W. S.; Lawrence, S. O.; Yu, J. J.; Crowe, W. E.; Garno, J. C. Applying AFM-based nanofabrication for measuring the thickness of nanopatterns: The role of head groups in the vertical self-assembly of omega-functionalized n-alkanethiols. Langmuir 2010, 26, 3040−3049. (15) Latimer, W. M.; Rodebush, W. H. Polarity and ionization from the standpoint of the Lewis theory of valence. J. Am. Chem. Soc. 1920, 42, 1419. (16) Lewis, G. N. Valence and the Structure of Atoms and Molecules; Chemical Catalog Co.: New York, 1923. (17) Pauling, L. The structure and entropy of ice and of other crystals with some randomness of atomic arrangement. J. Am. Chem. Soc. 1935, 57, 2680−2684. (18) Pauling, L. The Nature of the Chemical Bond; Cornell Univesity Press: Ithaca, NY, 1960. (19) Steiner, T. The hydrogen bond in the solid state. Angew. Chem., Int. Ed. 2002, 41, 48−76. (20) Bond, A. On the crystal structures and melting point alternation of the n-alkyl carboxylic acids. New J. Chem. 2004, 28, 104−114. (21) Nozaki, K.; Munekane, M.; Yamamoto, T.; Ogawa, Y. Crystalline phase of normal alkane. J. Mater. Sci. 2006, 41, 3935. (22) Uno, K.; Ogawa, Y.; Nakamura, N. Polymorphism of long-chain alkane-alpha,omega-diols with an even number of carbon. Cryst. Growth Des. 2008, 8, 592−599. (23) Moreno-Calvo, E.; Gbabode, G.; Cordobilla, R.; Calvet, T.; Cuevas-Diarte, M. B.; Negrier, P.; Mondieig, D. Competing intermolecular interactions in the high-temperature solid phases of even saturated carboxylic acids (C10H19O2H to C20H39O2H). Chem. Eur. J. 2009, 15, 13141−13149. (24) Thalladi, V. R.; Nuesse, M.; Boese, R. The melting point alternation in alpha,omega-alkanedicarboxylic acids. J. Am. Chem. Soc. 2000, 122, 9227−9236. (25) Thalladi, V. R.; Boese, R.; Weiss, H. C. The melting point alternation in alpha,omega-alkanediols and alpha,omega-alkanediamines: Interplay between hydrogen bonding and hydrophobic interactions. Angew. Chem., Int. Ed. 2000, 39, 918−922.

■

SUMMARY AND CONCLUSIONS We presented here a study of the structure of diacid LFs for a range of coverages and molecular length. With increasing coverage mono-, bi-, and trilayer LFs are observed. The LFs exhibit a full 2D crystalline order, with an oblique unit cell containing a single molecule. Structural motifs derived from the 3D crystal structure of diacids24,58,59,78,85,86 are found to account well for the structure of our 2D LFs. In both cases ringtype HBs19,87,88 connect the diacid molecules into long molecular chains. These chains self-assemble to form the 2D crystalline order of the LF, similar to that in 3D diacid crystals. Nevertheless, some important difference are found between our LFs and 3D diacid crystals. One important difference, clearly demonstrated by the unit cell’s extra length as compared to that of a diacid molecule, is the inclusion of a mercury atom into the carboxyl−carboxyl bond. This inclusion transforms the bond from a simple ring-motif carboxyl−carboxyl HB into a carboxyl−metal−carboxyl complex, which is highly covalent in nature. The ligand formed is likely to be of either a chelate, or a bridging, bidentate type, as found for Pb−carboxyl ligands in 3D structures. A more puzzling difference is the absence in our LFs of the odd−even difference in structure, found to be very prominent in 3D diacid crystals.24,89 Such absence is likely to result from conformational changes in the carboxyl−carboxyl bond complex upon inclusion of the mercury atom. Another possibility is the occurrence of molecular twists and near-end gauche conformations in odd diacid LFs. Such distortions are indeed found to occur in the 3D crystal phases of odd diacids.24 These two issues, namely the exact conformation of the carboxyl−mercury−carboxyl complex, and the reasons for the absence of the odd−even structural difference in diacid LFs, clearly merit further study, perhaps by computer simulations, and by IR and other experimental methods.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank Prof. Yitzhak Mastai for discussions, the US−Israel Binational Science Foundation, Jerusalem, for support, and NSLS, Brookhaven National Laboratory, for beamtime. Brookhaven National laboratory is supported by DOE Contract DE-AC02-76CH0016.

■

REFERENCES

(1) Jeffrey, G. A.; Saenger, W. Hydrogen Bonding in Biological Structures; Springer: Berlin, Germany, 1991. (2) Israelachvili, J.; Wennerstrom, H. The role of hydration and water structure in biological and colloisal interactions. Nature 1996, 379, 219−225. 15595

dx.doi.org/10.1021/la3034325 | Langmuir 2012, 28, 15586−15597

Langmuir

Article

(26) Thalladi, V. R.; Boese, R.; Weiss, H. C. The melting point alternation in alpha,omega-alkanedithiols. J. Am. Chem. Soc. 2000, 122, 1186−1190. (27) Majewski, J.; Edgar, R.; Popovitz-Biro, R.; Kjaer, K.; Bouwman, W.; Als-Nielsen, J.; Lahav, M.; Leiserowitz, L. Structure determination in the twilight region between monolayer and 3D crytsals - A Grazing incidence x-ray-diffraction study of nanocrystalline aggregates of alpha,omega-docosanediol at the air-water interface. Angew. Chem., Int. Ed. 1995, 34, 649−652. (28) Wintgens, D.; Yablon, D.; Flynn, G. Packing of HO(CH2)14COOH and HO(CH2)15COOH on graphite at the liquidsolid interface observed by scanning tunneling microscopy: Methylene unit direction of self-assembly structures. J. Phys. Chem. B 2003, 107, 173−179. (29) Xu, S.; Yin, S.; Liang, H.; Wang, C.; Wan, L.; Bai, C. Identification of the preferential-bonding effect of disubstituted alkane derivatives using scanning tunneling microscopy. J. Phys. Chem. B 2004, 108, 620−624. (30) Kaganer, V. M.; Möhwald, H.; Dutta, P. Structure and phase transitions in Langmuir monolayers. Rev. Mod. Phys. 1999, 71, 779− 819. (31) Tamam, L.; Kraack, H.; Sloutskin, E.; Deutsch, M. Hydrogenbonded order in mercury-supported monolayers of end-functionalized alkanes. J. Phys. Chem. C 2011, 115, 25451−25463. (32) Kraack, H.; Sirota, E. B.; Deutsch, M. Measurements of homogeneous nucleation in normal-alkanes. J. Chem. Phys. 2000, 112, 6873. (33) Kraack, H.; Ocko, B. M.; Pershan, P. S.; Deutsch, M. Structure of a Langmuir film on a liquid metal surface. Science 2002, 298, 1404. (34) Kraack, H.; Ocko, B. M.; Pershan, P. S.; Sloutskin, E.; Deutsch, M. Langmuir films of normal-alkanes on the surface of liquid mercury. J. Chem. Phys. 2003, 119, 10339−10349. (35) Als-Nielsen, J.; Jacquemain, D.; Kjaer, K.; Leveiller, F.; Lahav, M.; Leiserowitz, L. Principles and applications of grazing-incidence xray and neutron-scattering from ordered molecular monolayers at the air-water interface. Phys. Rep. 1994, 246, 252. (36) Deutsch, M.; Ocko, B. M. Encyclopedia of Applied Physics; VCH: New York, 1998; Vol. 23. (37) Als-Nielsen, J.; McMorrow, D. Elements of Modern X-ray Physics; Wiley: New York, 2001. (38) Pershan, P. S.; Schlossman, M. L. Liquid surfaces and Interfaces: Synchrotron X-Ray Methods; Cambridge University Press: Cambridge, UK, 2012. (39) Tamam, L.; Kraack, H.; Sloutskin, E.; Ocko, B. M.; Pershan, P. S.; Ulman, A.; Deutsch, M. The structure of mercapto-biphenyls on Mercury. J. Phys. Chem. B 2005, 109, 12534. (40) Kraack, H.; Tamam, L.; Sloutskin, E.; Pershan, P. S.; Deutsch, M.; Ocko, B. M. Alkyl-thiol Langmuir films on the surface of liquid mercury. Langmuir 2007, 23, 7571−7582. (41) Tamam, L.; Medina, D. D.; Menahem, T.; Mastai, Y.; Sloutskin, E.; Yefet, S.; Deutsch, M. The structure and phase diagram of chiral alkyl-serine monolayers on mercury. Soft Matter 2010, 6, 526. (42) Tamam, L.; Ocko, B. M.; Reichert, H.; Deutsch, M. Checkerboard self-patterning of an ionic liquid film on mercury. Phys. Rev. Lett. 2011, 106, 197801. (43) Gaines, G. L. Insoluble Monolayers at Liqid Gas Interface; Wiley: New York, 1966. (44) Ocko, B. M.; Wu, X. Z.; Sirota, E. B.; Sinha, S. K.; Gang, O.; Deutsch, M. Surface freezing in chain molecules: Normal alkanes. Phys. Rev. E 1997, 55, 3164. (45) Smith, T. Monomolecular films on mercury. Adv. Colloid Interface Sci. 1972, 3, 161−221. (46) Kraack, H.; Ocko, B. M.; Pershan, P. S.; Sloutskin, E.; Tamam, L.; Deutsch, M. Fatty acid Langmuir films on liquid mercury: x-ray and surface tension studies. Langmuir 2004, 20, 5375−5385. (47) Ocko, B. M.; Kraack, H.; Pershan, P. S.; Sloutskin, E.; Tamam, L.; Deutsch, M. Crystalline phases of alkyl-thiol monolayers on liquid mercury. Phys. Rev. Lett. 2005, 94, 017802.

(48) Tamam, L.; Menahem, T.; Mastai, Y.; Sloutskin, E.; Yefet, S.; Deutsch, M. Langmuir films of chiral molecules on mercury. Langmuir 2009, 25, 5111. (49) Volmer, M. Thermodynamic consequences from the equation of state for adsorbed substances. Z. Phys. Chem. (Leipzig) 1925, 115, 253−260. (50) Kraack, H.; Ocko, B. M.; Pershan, P. S.; Tamam, L.; Deutsch, M. Temperature dependence of the structure of Langmuir films of normal-alkanes on liquid mercury. J. Chem. Phys. 2004, 121, 8003− 8009. (51) Magnussen, O. M.; Ocko, B. M.; Regan, M. J.; Penanen, A.; Deutsch, M.; Pershan, P. S. X-ray reflectivity measurements of surface layering in liquid mercury. Phys. Rev. Lett. 1995, 74, 4444. (52) DiMasi, E.; Tostmann, H.; Ocko, B. M.; Pershan, P. S.; Deutsch, M. X-reflectivity study of temperature dependent surface layering in liquid Hg. Phys. Rev. B 1995, 58, R13419. (53) Elsen, A.; Murphy, B.; Ocko, B.; Tamam, L.; Deutsch, M.; Kuzmenko, I.; Magnussen, O. Surface layering at the mercuryelectrolyte interface. Phys. Rev. Lett. 2010, 104, 105501. (54) Guinier, A. X-Ray Diffraction in Crystals, Imperfect Crystals and Amorphous Bodies; Freeman: San Francisco, CA, 1963. (55) Tamam, L.; Kraack, H.; Sloutskin, E.; Ocko, B. M.; Deutsch, M. Highly anisotropic thermal expansion in molecular films of dicarboxylic fatty acids. Phys. Rev. B 2012, 85, 205449. (56) Wiegart, L.; Struth, B. Geometric boundary condition for the chain alignment in lipid monolayers. Physica B 2005, 357, 126−129. (57) Jeon, Y.; Vaknin, D.; Bu, W.; Sung, J.; Ouchi, Y.; Sung, W.; Kim, D. Surface nanocrystallization of an ionic liquid. Phys. Rev. Lett. 2012, 108, 055502. (58) Housty, J. Structural properties of saturated aliphatic diacids with an odd number of carbon atoms COOH(CH2)2N+1COOH. Acta Crystallogr. 1968, B 24, 486. (59) Vanier, M.; Brisse, F. Nouvel affinement de la structure de l’acide dodecanedioique. Acta Crystallogr., Sect. B 1982, 38, 643−645. (60) Bond, A. D.; Edwards, M. R.; Jones, W. Sebacic acid. Acta Crystallogr., Sect. E 2001, 57, o141−o142. (61) Ventola, L.; Metivaud, V.; Bayes, L.; Benages, R.; Cuevas-Diarte, M. A.; Calvet, T.; Mondieig, D. The binary system tetradecanedioic acid-hexadecanedioic acid: Polymorphism of the components and experimental phase diagram. Helv. Chim. Acta 2006, 89, 2027−2039. (62) Petit, L.; Lapalu, L.; Sautet, P. Self-assembly of diacid molecules: A theoretical approach of molecular interactions. J. Phys. Chem. C 2009, 113, 17566−17571. (63) Boese, R.; Weiss, H. C.; Blaeser, D. The melting point alternation in the short-chain n-alkanes: Single-crystal X-ray analyses of propane at 30 K and of n-butane to n-nonane at 90 K. Angew. Chem., Int. Ed. 1999, 38, 988−992. (64) Kaneko, F.; Kobayashi, M.; Kitagawa, Y.; Matsuura, Y. Structure of stearic acid E form. Acta. Crystallogr., Sect. C 1990, 46, 1490−1492. (65) Small, D. M. The Physical Chemistry of Lipids; Plenum: New York, 1986. (66) Kjaer, K.; Als-Nielsen, J.; Helm, C. A.; Tippman-Krayer, P.; Moehwald, H. Synchrotron x-ray diffraction and reflection studies of arachidic acid monolayers at the air-water interface. J. Phys. Chem. 1989, 93, 3200−3206. (67) Petrov, J. G.; Pfohl, T.; Mohwald, H. Ellipsometric chain length dependence of fatty acid Langmuir monolayers. A heads-and-tails model. J. Phys. Chem. B 1999, 103, 3417−3424. (68) Emsley, J. The Elements, 3rd ed.; Oxford University Press: Oxford, UK, 1998. (69) Broniatowski, M.; Dynarowicz-Latka, P. Search for the molecular mechanism of mercury toxicity. Study of the mercury(II)surfactant complex formation in langmuir monolayers. J. Phys. Chem. B 2009, 113, 4275−4283. (70) Broniatowski, M.; Flasinski, M.; Dynarowicz-Latka, P.; Majewski, J. Grazing incidence diffraction and x-ray reflectivity studies of the interactions of inorganic mercury salts with membrane lipids in Langmuir monolayers at the air/water interface. J. Phys. Chem. B 2010, 114, 9474−9484. 15596

dx.doi.org/10.1021/la3034325 | Langmuir 2012, 28, 15586−15597

Langmuir

Article

(71) Schwartz, D. K.; Viswanathan, R.; Garnaes, J.; Zasadzinski, J. A. Influence of cations, alkane chain-length, and substrate on molecular order of Langmuir-Blodgett films. J. Am. Chem. Soc. 1993, 115, 7374− 7380. (72) Vaknin, D.; Bu, W.; Satija, S. K.; Travesset, A. Ordering by collapse: Formation of bilayer and trilayer crystals by folding Langmuir monolayers. Langmuir 2007, 23, 1888−1897. (73) Mukherjee, S.; Datta, A.; Giglia, A.; Mahne, N.; Nannarone, S. Relating structure with morphology: A comparative study of perfect Langmuir-Blodgett multilayers. Chem. Phys. Lett. 2008, 451, 80−87. (74) Mukherjee, S.; Datta, A.; Giglia, A.; Mahne, N.; Nannarone, S. Chemistry at air/water interface versus reaction in a flask: Tuning molecular conformation in thin films. Langmuir 2009, 25, 3519−3528. (75) Dong, W.-F.; Wang, R.; Mao, G.; Moehwald, H. Self-assembled molecular patterns of fatty acid on graphite in the presence of metal ions. Soft Matter 2006, 2, 686−692. (76) Gericke, A.; Huhnerfuss, H. The effect of cations on the order of saturated fatty acid monolayers at the air-water interface as determined by infrared reflection-absorption spectrometry. Thin Solid Films 1994, 245, 74−82. (77) Deutsch, M.; Magnussen, O. M.; Ocko, B. M.; Regan, M. J.; Pershan, P. S. Thin Films: Self-Assembled Monolayers of Thiols; Academic: San Diego, CA, 1998. (78) Kaneko, F.; Ishikawa, E.; Kobayashi, M.; Suzuki, M. Structural study on polymorphism of long-chain dicarboxylic acids using oblique transmission method for micro FT-IR spectrometers. Spectrochim. Acta, Part A 2004, 60, 9−18. (79) Goto, M.; Asada, E. The crystal structure of the B-form of stearic acid. Bull. Chem. Soc. Jpn. 1978, 51, 2456−2459. (80) Moreno-Calvo, E.; Calvet, T.; Cuevas-Diarte, M. A.; Aquilano, D. Relationship between the crystal structure and morphology of carboxylic acid polymorphs. Predicted and experimental morphologies. Cryst. Growth Des. 2010, 10, 4262−4271. (81) Bickerstaffe, A. K.; Cheah, N. P.; Clarke, S. M.; Parker, J. E.; Perdigon, A.; Messe, L.; Inaba, A. The crystalline structures of carboxylic acid monolayers adsorbed on graphite. J. Phys. Chem. B 2006, 110, 5570−5575. (82) Bruch, L. W.; Diehl, R. D.; A, V. J. Progress in the measurement and modeling of physisorbed layers. Rev. Mod. Phys. 2007, 79, 1381− 1454. (83) Arnold, T.; Clarke, S. M. Diffraction from physisorbed layers. Curr. Opin. Colloid Interface Sci. 2012, 17, 23−32. (84) Arnold, T.; Thomas, R. K.; Castro, M. A.; Clarke, S. M.; Messe, L.; Inaba, A. The crystalline structures of the even alkanes hexane, octane, decane, dodecane and tetradecane monolayers adsorbed on graphite at submonolayer coverages and from the liquid. Phys. Chem. Chem. Phys. 2002, 4, 341−351. (85) Housty, J.; Hospital, M. Structure cristalline de l’acide dodecanedioique, COOH(CH2)10COOH. Caracteres structuraux des diacides aliphatiques satures e nombre pair de carbones. Acta Crystallogr. 1966, 21, 553−559. (86) Housty, J.; Hospital, M. Structure des deux formes cristallines de lacide azelaique COOH[CH2]7COOH. Acta Crystallogr. 1967, 22, 288. (87) Allen, F.; Motherwell, W.; Raithby, P.; Shields, G.; Taylor, R. Systematic analysis of the probabilities of formation of bimolecular hydrogen-bonded ring motifs in organic crystal structures. New J. Chem. 1999, 23, 25−34. (88) Lackinger, M.; Heckl, W. M. Carboxylic acids: Versatile building blocks and mediators for two-dimensional supramolecular selfassembly. Langmuir 2009, 25, 11307−11321. (89) Tao, F.; Bernasek, S. L. Understanding odd-even effects in organic self-assembled monolayers. Chem. Rev. 2007, 107, 1408−1453.

15597

dx.doi.org/10.1021/la3034325 | Langmuir 2012, 28, 15586−15597