Directed Assembly of Multi‐Walled Nanotubes and Nanoribbons of Amino Acid Amphiphiles Using a Layer‐by‐Layer Approach

Abstract Monodisperse unilamellar nanotubes (NTs) and nanoribbons (NRs) were transformed to multilamellar NRs and NTs in a well‐defined fashion. This was done by using a step‐wise approach in which self‐assembled cationic amino acid amphiphile (AAA) formed the initial NTs or NRs, and added polyanion produced an intermediate coating. Successive addition of cationic AAA formed a covering AAA layer, and by repeating this layer‐by‐layer (LBL) procedure, multi‐walled nanotubes (mwNTs) and nanoribbons were formed. This process was structurally investigated by combining small‐angle neutron scattering (SANS) and cryogenic‐transmission electron microscopy (cryo‐TEM), confirming the multilamellar structure and the precise layer spacing. In this way the controlled formation of multi‐walled suprastructures was demonstrated in a simple and reproducible fashion, which allowed to control the charge on the surface of these 1D aggregates. This pathway to 1D colloidal materials is interesting for applications in life science and creating well‐defined building blocks in nanotechnology.

Abstract: Monodisperse unilamellar nanotubes (NTs )a nd nanoribbons (NRs) were transformed to multilamellar NRs and NTsi nawell-defined fashion. This was done by using as tep-wise approachi nw hich self-assembled cationic amino acida mphiphile (AAA) formed the initial NTso r NRs, and added polyanion produced an intermediate coating. Successive addition of cationic AAA formed ac overing AAA layer,a nd by repeating this layer-by-layer (LBL) procedure, multi-walled nanotubes (mwNTs) and nanoribbons were formed. This process wass tructurally investigated by combining small-angle neutron scattering (SANS) and cryogenic-transmission electron microscopy (cryo-TEM), confirming the multilamellar structurea nd the precise layer spacing. In this way the controlled formation of multi-walled suprastructures was demonstrated in a simple and reproducible fashion,w hich allowed to control the chargeo nt he surface of these 1D aggregates. This pathway to 1D colloidal materials is interesting for applications in life science and creating well-defined building blocks in nanotechnology.
Self-assembled nanotubes (NTs) and nanoribbons (NRs) have become one of the most interesting topics of recent nano-science research, as they allow simple access to well-defined one-dimensional (1D) colloidal objects. In 1984 Yager and Schoen, [1] Nakashima et al. [2] as well as Yamada et al. [3] independently observed the formationo fh ollow cylindrical microstructures self-assembled from different amphiphilic molecules; these structures were termed "tubules". Frequently encountered buildingb locks are peptidea mphiphiles, [4] where nanotube formation is driven by hydrophobic interactions, chirality and hydrogen-bonds, which lead to very well-defined cylindrical structures.T hey may be formed by the diphenylalanine motif as the essential nanotube forming peptide [5] but in principle, there is quite high flexibility with respect to the selection of the molecular buildingb locks. [6] The NTsr adius depends mainly on the particularm olecular architectureo ft he peptide amphiphile. The radius is typically veryu niform and in the range of 5t o5 0nm. [7] Peptide nanotubes are commonly very long and single-layered. It can be noted that self-assembled nanotubes can also be formed spontaneously by other systems, such as catanionic amphiphile mixtures containing bile acid, [8] mixtures of sodiumd odecyl sulfate( SDS) and cyclodextrine, [9] or from phospholipids by applyinge lectric field. [10] The state of the art of self-assembly of organic molecules into nanotubes has been reviewed comprehensively by Shimizu. [11] Startingf rom ah eated molecular or micellar (or vesicular) amphiphile solution that is cooledd own (often simply to room [ temperature), nanoribbons and nanotubes form spontaneously by self-assembly and concurrent chain crystallization due to the intrinsic nature of the building blocks.T his simple and spontaneous bottom-up process then allows access to well-defined 1D structures that can be used in many applications,i ncluding drug delivery [12] andn anotechnology. [13] Of course, of key importance is the ability to control and modify the mesoscopic architectureo ft hese structures as that determines their functional properties. In particular,c ontrolling their wall thickness, or equivalently the number of shells,i ss cientifically promising as it affects their mechanical properties as well as releasep roperties when structures are loadedw ith active agents. One simplew ay of controlling the effective wall thickness is via tuning the chemistry (typically via the length of the hydrophobic part) of the molecules employed,b ut that is rather limited in scope. For too short chains crystallization does not occur,a nd for too long chainsm olecules crystallize always and structures precipitate, without forming 1D structures. An attractive andv ersatile approachw ould be to modify those assemblies by going from unilamellart om ultilamellar systems. There have been few reports of multilamellar nanotubes, but in these cases the multilamellar structures are formed spontaneously in an uncontrolled manner,t hereby leadingt os tructurally rather ill-defined systems, [14] where the outer layer often shows helical features. [15] Multilayer nanotubes recentlyh ave also been employed as drug carriers, where the drug was covalently bound to the tube-formingc atanionic amphiphile mixture. [16] In this work we describe the designo fal ayer-by-layer (LBL) technique in which consecutive addition of the amino acid amphiphile (AAA) to already prepared AAA NTsa nd NRs correspondingly increases the number of layers, allowing to control the wall thickness while retaining the monodisperse nature in terms of radial extension. Of course,b yt hemselves the AAA would have little tendency to produce as econd layer around an already existing nanotube of equal charge, and they remain single-layered ( Figure 1A). Accordingly,w ee mployed the concept of charger eversal of the surfaceb yd epositing al ayer of oppositely charged polyelectrolyte. This then facilitates subsequent deposition of anotherp ositively charged AAA layer with the polyelectrolytee ffectively working as glue between the equallyc harged AAA layers. By repeating this step-by-step approach multi-walled AAA NTsa nd NRs can be formed in aw ellcontrolled manner.R elated multilayer systems have been prepared and studied quite extensivelyf or oppositely charged polyelectrolytes, in different shapes (e.g.,f lat surfaces or microcapsules) and functions. [17] Here we extend this concept and preparerigid composite 1D materials of AA and polyelectrolyte buildingb locks.
For the purpose of spontaneously forming1 Ds tructures we employed the AAA C 12 KC 12 K-NH 2 (previously named C 12 b 12 ). [19] By bringingaheated solution back to room temperature, long well-defined NTsa re formed spontaneously( Figure 1). Nanotube formation proceeds by chiral self-assembly from long, thin fibers that immediatelys elf-assemble after mixing to thin twisted tapes that form within hours by widening of the fiber widths. Ta pes further widen to helically coiled ribbonsi na matter of days, thereby changing their surface curvature from twisted to helical. This transformation is the result of ad elicate interplay between variouse lastic forces and the chirality of the system. [20] The coiled ribbons then mature over weeks into well-defined closed nanotubes, with an average inner radiuso fa pproximately 50 nm and an average wall thickness of 2.4nm, which corresponds well to the length of ab ack-folded molecule 18 (see Figures 1A and S1a).
This thickness was deduced from small-angle Xray and neutron scattering (SAXS, SANS) experiments,a nd modeling with the suitable form factor for long hollow cylinders as well as by aK ratky-Porod analysisf or locally flat structures (see Supporting Information 4.2 and Figure S3 for details; note that in SANS one sees primarily the hydrogenated core of the molecule and less the hydrophilic moieties exposed to the aqueous solution), as previously done for other peptiden anotubes. [21] Both approaches are in good agreement and yield an average wall thickness of unmodified NTso fa pproximately 2.4 nm. Cryo-TEM results support these findings ands how that over time intermediate fibers and ribbons disappeara nd eventually very stiff, several microns long NTsa re formed, exclusively of a single layer ( Figure 1B,C ).
Fully developed NTsr emained unchanged for more than two years as confirmed by Cryo-TEM.T he very highm onodispersity of the radius (52.5 nm) is best seen in the SAXS experiment presentedi nF igure 1D in aK ratky-Porod plot to enhance the visibility of the form factor oscillationsd one on 2m onths old samples. Here, up to 14 form factor oscillations are visible,w hich indicates the extraordinarily high degree of monodispersity of the radius of the NTs. Modeling the experimental scattering curve with a model of ac ylindrical shell (for details see Supporting Information 3.1 and 4.1) indicates ap olydispersity index (PDI) of less than 3% [see FigureS2a nd 4.1 for furtherd etails including fits with am odel of hollow cylinders described by eq. (2)].I tm ay be noted that this simple shell model does not capturet he evolution of the relative amplitude of the oscillations well. This has to be attributed to the factt hat in reality the contrast profile is not simply step-wise, but will have marked features, especially due to the fact that the amphiphilic layer is organized in ac rystalline manner.H owever, relevant for determining the extent of polydispersity is mostlyt he high number of form factor oscillations and how they are decaying.
These well-defined unilamellar NTsa nd NRs then served as templates for depositing subsequent layers of C 12 KC 12 K-NH 2 (schematically depicted in Figure2). In the first step of this preparation proceduret he cationic NTs/NRs were treated with sodium poly(methacrylate) (NaPMA) to reverse their charge. After applying a2 -fold excess of polyanion charges, waiting for 2days, and careful washing, polyanionm odifiedN Rs andN Ts were obtained. Next, an ew C 12 KC 12 K-NH 2 solution was added to these polyelectrolyte-coated structures, employed again at 2-fold charge excess. This led within 2days to the formation of as econd C 12 KC 12 K-NH 2 layer aroundt he initial structures. Successive steps were done accordingly,a lternating between NaPMA and C 12 KC 12 K-NH 2 ,e ach modification step being followed by aw ashingc ycle. Continuous washing steps proved necessary to minimizet he formation of insoluble complexes when addingm aterialf or the next coat. Af raction of the structures will precipitate due to net charge compensation with each coating step, thereby decreasing its efficiency.Bye mploying this procedurer epeatedly,i tw as possible to produce NTs and NRs with aw ell-defined number of layers, in af ashion not yet done before. In this work, multi-walled NTsa nd NRs with a maximum of 7A AA shells wereformed.
The characterization of these multi-shell 1D assemblies was done using complementary SANS and cryo-TEM. [22] Figure 3a presentsS ANS curves directly confirming the attachment of subsequentl ayers as by the increase of intensity in the lowqrange. At the same time the slope of the curvesi ncreases substantially,w hich indicates an effective growth of the thickness of the tube walls. In addition, ac orrelation peak appears at q m = 1.6-1.7 nm À1 (Figure 3b)w hich indicates as pacing of the subsequentA AA shells by approximately 3.4-3.8 nm (= 2p/q m ). This peak becomes more pronounced with increasingn umber of deposited shells. It might be noted that in this case Bragg's law (d = 2p/q m )i sj ust an approximation as the form factor of the structure walls will also contributet ot he scattering pat- tern, thereby shifting q m to as omewhat lower q value compared to the pure structure factor peak.
Cryo-TEM corroborates this picture of forming well-defined layers around the initial unilamellar NRs (Figure 4) and NTs (Figure 4d). Thus, the cryo-TEMd atar eveals multilayered NRs, and confirmst hat our procedure enables to form aw ell-defined number of AAAl ayers aroundt he initial AAA assemblies.
The SANS data shown in Figure 3w ere quantitatively analyzed with respect to the number of shells,w hich manifested mainly in the increasing slope of the SANS curves (thicker effective shell) and the correlation peak at q % 1.65 nm À1 that becomes more pronounced with increasing number of correlated layers due to Bragg scattering. Taking into account the appro-priate difference in scattering length densities between the assemblies and the solventD 2 O( for details see SupportingI nfor-mation6 .) the overall scatteringi ntensity I(q)w as described by the product of the form factor P cyl (q)o favery long hollow cylinder with a para-crystalline structure factor S PLT (q)t hat accounts for the correlated layers( see Supporting Information 3.2 for details). [23] The form factor for such ag eometry is easily obtained by factorizing it into ac ross section form factor P cs (q)f or the radial direction and the shape factor P'(q)f or the length direction of the tube, as those two dimensions are virtually uncoupled due to the extreme length of the nanotubes (it might be noted that the samples show anisotropic scattering patterns, see FigureS5) [24a] [Eqs. (1) and (2)   R core is the nanotubes inner (core) radius, DR its wall thickness, and L itsl ength. J 1 is the cylindrical Besself unction of the first kind.
The paracrystalline lamellae theory( PLT) was applied to account for some disordera nd lattice defects such as stacking disorderc aused by small variations D in the average layer separation d. [25a, b] The resulting multilamellar arrays are treated as purely one-dimensional systems along the lattice plane k. [25c, d] The theory has long been established with 1D systemsa nd thus is applicable for our case. [24] For furtherd etails on these modelssee Supporting Information 3.2 [Eq. (3)]: The corresponding fits when applying this model are included in Figure 3a nd show very good agreement with the experi-mentalS ANS data. It might be noted that already as imple multi-slitm odel gives generically similarr esults( see Figure S4), but the paracrystalline modele mployed here gives more reliable resultsr egarding the average number of layers seen. The deduced parameters are the core radius R core and the outer radius R out ,t he interlayer spacing d,a nd the number of layers N max ,w hich are summarized in Ta ble 1f or different numberso f modification steps.
One finds as ystematic increaseo ft he average thickness of the NT walls, which goes hand in hand with the increasing number of correlated layers (see Ta ble 1a nd Ta ble S1 for more details). This confirms nicely the good control of the multilayer structure by our layer-by-layer preparation approach. The average outer radius R out and inner radius R core are roughly changing by the same amount;t his indicatest hat the layering process occurs simultaneously on both the outside and the inside of the NTs, whichi sc onfirmed by the direct cryo-TEM analysis (Figure 4). However,t he resulting layers are not always perfectly enveloping the whole structure and can be incomplete, whichi sa lso evidentf rom the cryo-TEMi mages (box in Figure 4). This results in as lightly varying number of AAA shells along the length axis, seen as step-like substructures on the mwNTss urface. This also explains why N max increasesm ore than the number of steps involved but by slightly less than by af actor of two as would be the case for complete formation of inner and outer AAA layers.T he good agreement between the mass per length derived from the SANS intensity data and calculated theoretically (see Supporting Information 5.) is confirming the self-consistency of our analysis.
Circular dichroism (CD) measurements of various samples were done in order to learn on the internal order of the AAA in the assembled state and the influence of additional polyelectrolyte layers on the structure. Our results demonstrate that the AAA alone ( Figure S6) has no classical signal of alpha or beta sheet,b ut instead a" knee" structure around 220 nm. The filtration did not change the signal ( Figure S6B). Adding a PMA layer also did not change the signal shape, but with further addition of alternate AAA and PMA layers ( Figure S6C to F) the "knee" disappeareda nd at the same time an increase is seen at low wavelength.
Finally,i no rder to gain insight into the arrangement of the AAA molecules in the NRs we also performed X-ray diffraction (XRD) experiments on as ample before and after addition of NaPMA.T he spectra are shown in Figure S7 and they indicate the absence of ah ighly ordered, crystalline packing.T he observed main diffraction peak corresponds to as pacing of 0.295 nm and one observes as hiftt oas maller spacingb y 1.6 %u pon addition of the NaPMA. This somewhatd enser packing can be explained by the fact that the binding of the oppositely charged PMA on the surface of the nanotubes compensates the charges of the C 12 KC 12 K-NH 2 AAA, thereby reducing the electrostatic repulsion of the amino acid head groups. In general, XRD confirms that the crystalline ordering of the C 12 KC 12 K-NH 2 AAA is not affected by the presence of the polyelectrolyte.
In summary,w ed escribe an ew procedure for the formation of well-definedm ulti-walled AAA nanotubes and nanoribbons by as tep-by-step, layer-by-layer templating approach. For this purpose, first alayer of negatively charged polyelectrolyte (polymethacrylate, NaPMA)w as depositedo nto the surfaceo fv ery monodisperse positively charged single-walledA AA assemblies. Then, these negatively charged NTso rN Rs functiona sa scaffold fort he deposition of the next layer of cationic C 12 KC 12 K-NH 2 AAA, where the polyanion functions effectively as ag lue between the different AAA layers. This process of polyelectrolyte adsorption and AAA deposition can then be repeated severalt imes (done up to an umber of 7s hell in total) and the number of repetitions determines the number of AAA layers contained in these mixed composite multi-walled NRs/ NTs. This preparation Scheme is described in Figures 2a nd S3. It should be noted that in principle the ability for forming mul- Table 1. Parameters obtained from SANSd ata analysis for as tep-wise modificationo fs elf-assembled 1D structures to obtain ordered mwNTs composed from alternatings hells of C 12 KC 12 K-NH 2 and NaPMAl ayers (application of one polyelectrolyte and one AAAl ayer are considered as one modifications tep) [a] :M aximum number of AAA shells N max ,i nner radius R core ,o uter radius R out ,l ayer spacing d,m olecularw eight per unit length determined by SANS( M w,SANS L À1 )a nd theoretical calculated values (M w,th L À1 )( for details see Supporting Information 5.), and stable coated fraction cf. tilayered nanotubes with C 12 KC 12 K-NH 2 is quite surprising, as the amphiphile itself forms very monodisperse nanotubes of R = 55.0 nm and PDI < 0.03 ( Figures 1D and S2) but is then able to form tubes with radii of 42.5 and 66 nm for inner or outer shells,r espectively,w hich indicates rather high adaptivity of the peptides hell to allow for other curvatures. Accordingly, an increasing tendency of defects is expected with increasing layer number,a si ndicated by the cryo-TEM images. The layer spacingi sv ery well-defined at ac onstant value of approximately 3.6 nm, as evidencedb yS ANS (Figure 3). Apparently,t he process leads to the formationo ff ully controlled multi-walled NRs and NTsa sd epicted in Figure 2. Densely packed layers of AAA are separated by at hin layer ( % 1.1nm) of the polyanionic NaPMA. Interestingly,t his shell contains by mass only approximately 15 wt %o fh ydration waterr elative to its polyelectrolyte content. It should be noted that depending on the outer layer (AAA or polyanion) the multi-layer structures are positively or negatively charged. This constitutes an additional degree of flexibility of the formed aggregates,w hich could be relevantf or biomedical applications (especially as one can freely choose the outer polyelectrolyte) but also for other potential applications in material science.

Mod
The formed hybrid nanotubes are interesting materials as buildingb locks in nanoscience due to their being composed of bio-friendly materials. Their effective thickness (i.e.,n umber of AAA shells)a llows to control their robustness, while at the same time retaining aw ell-definedp orosityv ia their inner tube opening. Another advantage is that one can have the final mwNT structures with an anionic or cationic cover.T herefore, this work allows for the formation of ac olloidal material, as it can be interesting for applicationsi nl ife science, but also as aw ell-controlled buildingb lock in nanotechnology.E xamples of possible applications would be incorporation of active agents between the amino acid walls and their controlled release (for instance by enzymatic uncoating of the AAA layers), the use of the mwNTsa st emplates (for instance for metallization), or as multi-walled channels in nanotechnological devices (where the transport along the cylinder axis could be controlled by the choice of polyelectrolyte).