In situ X-ray reflectivity studies of molecular and molecular-cluster intercalation within purple membrane films

It has been recently demonstrated that molecular and molecular cluster guest species can intercalate within lamellar stacks of purple membrane (PM), and be subsequently dried to produce functional bioinorganic nanocomposite films. However, the mechanism for the intercalation process remains to be fully understood. Here we employ surface X-ray scattering to study the intercalation of aminopropyl silicic acid (APS) or aminopropyl-functionalised magnesium phyllosilicate (AMP) molecular clusters into PM films. The composite films are prepared under aqueous conditions by guest infiltration into preformed PM films, or by co-assembly from an aqueous dispersion of PM sheets and guest molecules/clusters. Our results show that addition of an aqueous solution of guest molecules to a dried preformed PM film results in loss of the lamellar phase, and that subsequent air-drying induces re-stacking of the lipid/protein membrane sheets along with retention of a 2–3 nm hydration layer within the inter-lamellar spaces. We propose that this hydration layer is necessary for the intercalation of APS molecules or AMP oligomers into the PM film, and their subsequent condensation and retention as nano-thin inorganic lamellae within the composite mesostructure after drying. Our results indicate that the intercalated nanocomposites prepared from preformed PM films have a higher degree of ordering than those produced by co-assembly.


Introduction
Intact sheets of purple membrane (PM), consisting of hexagonally packed trimeric arrays of the protein bacteriorhodopsin (bR) embedded within a lipid bilayer membrane 4.7 nm in thickness, can be readily isolated from the organism Halobacterium salinarium. PM sheets are typically 500-1000 nm in size, and of technological importance due to the photochromic and light harvesting properties of bR, which can pump protons from the cytoplasmic to extracellular side of the membrane under illumination. 1 These properties have led to the development of novel PM-based materials for holography, photoswitching and memory storage applications. 2,3 The photocycle in PM is well understood, with established structural and temporal models describing the process of proton shuttling through the membrane with angstrom and femtosecond resolution. 4,5 In this process, water provides the main source of proton donation to bR, and so the behaviour and interaction of water molecules near the surface of the PM sheets is of particular mechanistic interest. [6][7][8][9] In many technical applications, PM sheets are stacked by controlled sedimentation into mesostructured thin lms that require some level of hydration to retain their proton-pumping functionality. 10,11 Hybrid lms formed by intercalating a guest material with the PM sheets are more structurally robust than pure PM lms, 12 but intercalation can impact on protein functionality due to inhibition of water transport through the lm. Therefore, understanding the mechanism of intercalation and the role of hydration in the intercalation process and resultant lm functionality are important aspects for the design and construction of hybrid PM lms.
Previous structural studies on PM thin lms using neutron reectivity (NR) and X-ray reectivity (XRR) have focused on the role of water that exists in the lms aer their formation. It has been shown that PM lms exhibit a lamellar structure with an interlayer spacing of 4.7 nm when prepared at less than 10% relative humidity (RH), which swells to 6.2-8.0 nm at RH > 95%. 13,14 The swelling in the lamellar spacing is attributed to interstitial water molecules, which were found to exhibit retarded rotational motion and reduced translational diffusion in the direction of stacking compared to the bulk free liquid. The role of water within the lms has been described as analogous to a lubricant, allowing for the necessary conformational changes in bR during illumination as well as providing a source of protons for the photocycle. 13 This has been conrmed by XRR studies on the thermal decomposition of fully dehydrated PM lms, which maintained protein functionality aer exposure to temperatures as high as 140 C. 15 In a water-free lamellar PM stack there was a reduction in allowable conformal motion available to bR, which inhibited denaturation. Under dehydration it was found that the bR assemblies rigidied and became locked in their 2D short range structure, resulting in a corrugation of the PM sheet, as observed by diffuse scattering. The corrugation disappeared when the lm was swollen with interdigitated water, indicating that exibility was restored to the PM sheet. 15 Hydration also plays an important role in the formation of intercalated PM lms hosting guest molecules and materials. It has been shown that hybrid lms consisting of PM sheets intercalated with a condensed organosilane can be formed by adding an aqueous precursor to a PM lamellar lm and then allowing the samples to dry. 12,16,17 These composite lms demonstrate enhanced mechanical tolerance and solvent resistance in addition to functional photocycle and photochromic responses. Importantly, the composite lms retain their lamellar structure as opposed to dispersal of the PM sheets randomly within an organic or inorganic supporting matrix. Critically, it was not known from these studies whether the hydrolysed organosilane precursors coated the PM sheets in suspension prior to reassembly and condensation, or the guest species permeated into the assembled lm as the concentration increased upon drying. Intercalation in PM lms of non-sol-gel precursors such as polysaccharides, inorganic polymers and polycationic organoclay oligomers has also been demonstrated, but the mechanism by which the guest materials permeated into the PM host structure was not fully elucidated. 16,18 In this paper we have used X-ray surface scattering to study the mechanism of formation of PM lms intercalated with aminopropyl silicic acid (APS) or cationic oligomers of aminopropyl-functionalised magnesium phyllosilicate (AMP), as two model guest species presented in molecular or molecularcluster forms, respectively. Both of these guests have been previously used to produce PM-based composite lms with an AB-type lamellar structure. 16 The composite lms are prepared under aqueous conditions by guest inltration into preformed PM lms, or by co-assembly from an aqueous dispersion of PM sheets and guest molecules/clusters. Our results show that addition of an aqueous solution of guest molecules to a dried preformed PM lm results in loss of the lamellar phase, and that subsequent air-drying induces re-stacking of the lipid/ protein membranes along with retention of a 2-3 nm hydration layer within the inter-lamellar spaces. We propose that this hydration layer is necessary for the intercalation of APS molecules or AMP oligomers into the PM lm, and their subsequent condensation and retention as nano-thin inorganic lamellae within the composite mesostructure aer drying. Our results indicate that the intercalated nanocomposites prepared from preformed PM lms have a higher degree of ordering than those produced by co-assembly.

Materials, sample preparation and experimental methods
In situ X-ray surface scattering studies of the PM lms were carried out at the XMaS UK CRG beamline (BM 28) at the European Synchrotron Radiation Facility (ESRF) in Grenoble, France. The monochromatic X-ray beam had an energy of 10 keV with a corresponding wavelength l ¼ 1.24Å. The beam size on the sample position was 300 mm Â 300 mm (height Â width). The incident angle q i was varied in the range 0.06-2 for each of the measurements, corresponding to a momentum transfer Q ¼ 4p sin q i /l range of 0.011-0.345Å À1 . The surface scattering data were collected by X-ray reectivity (XRR), [19][20][21]24,25 using an avalanche photodiode detector in the specular reection plane, but at reection angles q f with a small offset Dq ¼ 0.06 (q f ¼ q i AE Dq) to the incident angle (q i ) to minimise the contribution from the silicon substrate. PM samples on solid silicon substrates were mounted inside a small chamber enclosed with Kapton lms, which allowed a constant ow of helium (He) during the measurements to reduce background scattering and sample damage. Each sample was assessed for visible damage aer alignment. Beam damage was checked by taking measurements on several regions of the lm, moving the sample perpendicular to the incident beam.
Aminopropyl silicic acid (APS) solutions were prepared as detailed previously. 17 Briey, 5 mL of 3-aminopropyltriethoxysilane was added to 20 mL of 0.5 M HCl in a 50 mL round bottomed ask. The ask contents were rotary evaporated in a water bath at 45 C for 15-20 minutes until the generated ethanol had been removed. The APS concentration in the solution was determined by drying and condensing small aliquots in pre-weighed vials. Solutions were diluted with distilled water to the desired concentrations.
Aqueous sols of polycationic AMP organoclay oligomers were prepared at room temperature by drop-wise addition of 3-aminopropyltriethoxysilane (1.3 mL, 5.85 mmol) to an ethanolic solution of magnesium chloride (0.84 g, 3.62 mmol) in ethanol (20 g). 22,23 A white slurry was obtained aer 5 min which was stirred overnight and the precipitate isolated by centrifugation, washed with ethanol (50 mL) and dried at 40 C. Exfoliation of the clay was undertaken by dispersing 10 mg of the dried clay in distilled water (10 mL) followed by ultrasonication for 5 min. The resulting cloudy dispersion, with a measured zeta potential of +12 mV at pH 8.5, was passed through a Sephadex G-25/75 column (Aldrich), and the clear eluate of polycationic organoclay clusters collected.
The pristine PM lm (i.e. without guest materials) was prepared by placing a 60 mL droplet of an aqueous PM dispersion (5.5 mg mL À1 ) to the centre of a hydrophilic silicon wafer (1 Â 2 cm). The droplet was allowed to dry slowly at room temperature and 60-70% RH, resulting in a lm approximately circular in shape, 4-5 mm in diameter and 5-10 mm in thickness.
APS or AMP intercalation into preformed PM lms was undertaken by adding 60 mL of a 5.5 mg mL À1 3-aminopropyl silicic acid or organoclay solution to a preformed PM lm mounted on a silicon wafer using a volumetric pipette. For in situ XRR studies, the PM lm-containing silicon substrate was rst aligned in the X-ray beam, and then a droplet of APS or AMP solution added on top of the PM lm. XRR scans were then taken at different time intervals as the solution was allowed to dry at ambient RH (45-55%), and the measurements repeated at different positions on the lm. Alternatively, composite lms were prepared by a co-assembly procedure in which an aqueous solution of APS or AMP was mixed with an aqueous dispersion of PM sheets for $1 minute prior to deposition of the mixture on a bare silicon wafer that was pre-aligned in the X-ray beam. Co-assembly was undertaken by adding 60 mL of 5.5 mg mL À1 PM in water to 60 mL of 5.5 mg mL À1 aqueous APS or AMP solution in a small centrifuge tube. This mixture was drawn into a volumetric pipette and then deposited onto the silicon wafer.

Results and discussion
Formation of interstitial hydration layers due to wetting of PM lms The XRR prole of the pre-dried pristine PM lm control sample displayed two prominent Bragg peaks at Q 1 ¼ 0.134Å À1 and Q 2 ¼ 0.265Å À1 (Fig. 1a). The ratio of Q 1 /Q 2 z 1 : 2 indicated that these two peaks corresponded to the (001) (n ¼ 1) and (002) (n ¼ 2) diffractions respectively, as a result of a lamellar structure with an average d-spacing of d ¼ 2np/Q n ¼ 4.70 nm in the PM lm, with the lamellar plane approximately parallel to the Si substrate. This was consistent with the repeat distance for PM lms observed at minimum hydration, and the expected thickness for a single purple membrane bilayer without water present interstitially or absorbed from the atmosphere during the experiment. 6 From tting the Bragg peaks with a Gaussian function, the coherence length, L, of the ordered domains in the lm was calculated using the Scherrer equation, L ¼ 2pK/DQ, where DQ is the full width at half maximum (FWHM) of the peak and K is a shape factor of order unity. [24][25][26] Such an analysis gave a value of L of $71 nm for the dry PM control lm, which corresponded to an ordered domain of approximately 14 lipid bilayers (average number of bilayers, m ¼ L/d $ 14). The calculated values of d, L and m for all the PM lms studied are listed in Table 1.
To examine PM hydration, the XRR prole of a control PM lm was recorded, and then a 60 mL water droplet added on top of the lm, which subsequently spread and wetted the lm. The wetted PM lm was subject to a constant ow of He for 15 minutes to remove residual surface water, and then the He ow was stopped and XRR curves were collected as the lm gradually dehydrated at an RH of $50%. Aer 15 min of drying, the (001) and (002) Bragg peaks were shied to slightly smaller Q values (Q 1 ¼ 0.128Å À1 and Q 2 ¼ 0.256Å À1 respectively; average d-spacing ¼ 4.90 nm) (Fig. 1b), indicating a small swelling of $0.2 nm as compared to the dry control lm. Signicantly, a broad peak at Q ¼ 0.077Å À1 was also observed, corresponding to a d-spacing of 8.20 nm. This peak was attributed to an interstitial water layer of around 3.5 nm in thickness.
The coexistence of the three Bragg peaks suggested that the partially dried PM lm was not homogeneously hydrated but consisted of a mixture of extensively hydrated lamellar domains along with regions that remained essentially dehydrated. Aer further drying under He ow, the PM lm returned to a completely dehydrated state and the XRR curve showed that Bragg reections re-emerged at Q values (Q 1 ¼ 0.134Å À1 and Q 2 ¼ 0.268Å À1 ; average d-spacing ¼ 4.70 nm) (Fig. 1c), which were identical to those for the initial sample of the dried PM lm.  Table 1 Structural information for purple membrane (PM) films treated with solutions of different guest molecules. Composite films were prepared from preformed PM films or via host-guest coassembly. (* ¼ peak indicated by * in Fig. 1

Mechanism of aminopropyl silicic acid (APS) intercalation into PM lms
A droplet of APS was added to a dried PM lm aer obtaining the XRR prole of the preformed mesolamellar structure (Q 1 ¼ 0.134Å À1 and Q 2 ¼ 0.268Å À1 ; d-spacing $ 4.70 nm, L $ 60 nm, m $ 13; Fig. 2a), and the surface X-ray scattering curve of the wet sample re-recorded aer 12 minutes. The prole showed a complete absence of the characteristic Bragg peaks (Fig. 2b), suggesting that the PM stacks were delaminated or extensively disordered at a local level in the presence of aqueous APS. The wetted lm was then dried under a ow of He for 3 minutes and the XRR scan collected. Bragg peaks corresponding to (001) and (002) planes at Q 1 ¼ 0.116Å À1 and Q 2 ¼ 0.233Å À1 at an expanded d spacing of 5.40 nm reappeared (Fig. 2c). Re-emergence of the Bragg reections indicated that the wetted lm was reconstructed into a lamellar composite on loss of bulk water. However, the coherence length L was decreased to a value between 51.1 nm (10 layers) as compared to the native PM lm, suggesting a more disordered mesostructure. This was consistent with the intercalation of a 0.75 nm-thick sheet of organosilica between the membrane bilayers.
The complete loss of structural order upon wetting (Fig. 2b) suggested that the lamellar structure was largely dissociated prior to reassembly upon drying and intercalation. As a consequence, the possibility that intercalation of APS takes place via an exchange mechanism in which the guest molecules permeate a well-ordered matrix of PM sheets seems highly unlikely. In contrast, the presence of a 3.5 nm-thick hydration interstitial layer in the partially dried PM lms (cf. Fig. 1b), suggests that this transient phase could provide the mechanism for spontaneous sequestration of APS molecules. However, upon deposition of the APS containing droplet, the PM lm structure was much more disordered, evident from the absence of the absence of the Bragg peaks in Fig. 2b. APS may subsequently become entrapped in the form of a continuous organosilica matrix due to the onset of condensation reactions associated with further dehydration, leading to reformation of ordered lamellar structure and re-emergence of the Bragg peaks in Fig. 2c. The time for each XRR alignment and scan for such in situ measurements was $2 h. As such, it is also possible that we did not have the time resolution to fully capture the detail of the time evolution in the structural change during the intercalation process. However, it is clear that the aqueous environment and its lubricant-like role is essential to facilitating intercalation.
This mechanism was further tested by examining the structure of composite lms prepared via dehydration-induced coassembly from an aqueous mixture of PM sheets and APS. The XRR curve for the co-assembled lm showed a broad reection peak at Q ¼ 0.101Å À1 (d ¼ 6.24 nm) attributed to the (001) plane (Fig. 2d). The estimated coherence length from this was calculated to be L $ 23 nm (m $ 4). No (002) reection was observed, indicating that the composites produced by co-assembly were more disordered than those prepared by intercalation within a preformed PM lm. Given that the co-assembled lms also showed an increase in the interlayer spacing of approximately 1 nm compared with nanocomposites produced from preformed PM lms, the increased organosilica content and structural disorder suggested that mixing the components in suspension prior to lm formation increased the level of host-guest preorganization. A schematic representation of the proposed intercalation mechanism is shown in Fig. 3.

Intercalation of organoclay clusters into PM lms
As a comparison to the APS system, the intercalation of polycationic AMP organoclay clusters ca. 1.6 nm in size 16 with PM sheets to produce structured nanocomposites was investigated by XRR. While in situ condensation of entrapped APS resulted in the formation of an intercalated organosilica layer that conforms to the size of the interstitial water layer, the organoclay oligomers assemble as preformed units while conned between the PM layers to produce an ordered lamellar structure. In this regard, electrostatic interactions between the polycationic clusters and negatively charged surface of PM sheets could provide a feasible mechanism for sequestering high concentrations of the AMP clusters within the stacked mesolamellar structure.
An XRR scan of an AMP/PM composite lm prepared by addition of an aqueous droplet of AMP to a preformed air-dried PM lm displayed very broad (001) and (002) reections at Q 1 $ 0.115Å À1 (d $ 5.50 nm) and Q 2 $ 0.231Å À1 , respectively (Fig. 4a). The coherence length L calculated from the (001) peak was 34.3 nm (m ¼ 6), which was almost half that for the PM/ organosilane composite lms (m ¼ 10) produced under similar (c) the preformed PM film exposed to APS solution after being allowed to dry completely; and (d) a PM/organosilane film formed via co-assembly using identical amounts of PM and APS solution as for (b). Dot-dashed lines indicate the major reflections. All samples were formed on a silicon substrate.
conditions. The XRR prole of an AMP/PM composite lm prepared by co-assembly of the constituents also displayed one very broad (001) peak centred at Q $ 0.12Å À1 (d $ 5.23 nm, L ¼ 22.4 nm, m $ 4) (Fig. 4b), but no (002) reection was observed.
Although mixing the PM sheets and organoclay oligomers produces a uniform mixture prior to dehydration and coassembly, the increased level of interaction between the components when dispersed together in bulk solution, could result in complexation and formation of hybrid sheets that are less structurally homogeneous, and hence less likely to stack in an ordered conguration. Another possibility is that complexation between the components disrupts the formation of an interstitial water layer during the assembly process, which in turn results in poor ordering.
Overall, the results indicated that intercalation of the organoclay clusters produced a less ordered mesolamellar composite compared with samples produced in the presence of APS. The in situ reactivity of the APS molecules when dehydrated allowed the condensed phase to conform to the lamellar environment to a greater degree than the solid clay layers, resulting in longer range ordering observed. In both systems, coassembly of the constituents increased the level of disorder in the hybrid nanocomposites.

Summary and concluding remarks
Our surface scattering results show that hydration layers of $3.5 nm formed in the inter-lamellar space when PM lms were exposed to water. We suggest that this interstitial hydration layer plays a critical role in mediating intercalation of guest materials delivered via an aqueous medium. To elucidate the intercalation mechanism, we used in situ XRR measurements on lamellar-structured composite lms prepared from preformed PM lms or via host-guest co-assembly. The results showed that loss of the preformed PM lamellar structure occurs upon addition of a droplet of aqueous APS solution, and that a new lamellar phase with a larger interlayer spacing attributed to intercalation of APS or AMP is produced upon drying. We suggest that this assembly mechanism facilitates the inltration of the guest molecules or clusters, as diffusion into the highly hydrated disordered intermediate structure is facile. Moreover, formation of an ordered hydration layer during partial drying enables the guest molecules and clusters to be sequestered within the incipient lamellar phase. Subsequent solidication of the intercalated precursors via drying-induced sol-gel condensation reactions then provides a mechanism for retention of the guest species within the layered nanocomposite. In contrast, co-assembly of the host and guest components from a mixed aqueous dispersion produces intercalated composites but with a lower degree of structural order.
Understanding the intercalation mechanism should offer important insights for the synthesis of PM-based composite lamellar lms with a range of potential guest inorganic or polymeric materials. The presence of a guest material within the PM layers tends to reduce the extent of long range order (coherence length) in the composite lms, which could compromise functionality. Thus, mechanistic insights are required to inform how these materials should be optimised in terms of guest loading capacity and the concomitant effects on the structural organization of the PM composite lms. Such