Encapsulation of Cadmium Selenide Nanocrystals in Biocompatible Nanotubes: DFT Calculations, X‐ray Diffraction Investigations, and Confocal Fluorescence Imaging

Abstract The encapsulation of CdSe nanocrystals within single‐walled carbon nanotube (SWNT) cavities of varying dimensions at elevated temperatures under strictly air‐tight conditions is described for the first time. The structures of CdSe nanocrystals under confinement inside SWNTs was established in a comprehensive study, combining both experimental and DFT theoretical investigations. The calculated binding energies show that all considered polymorphs [(3:3), (4:4), and (4:2)] may be obtained experimentally. The most thermodynamically stable structure (3:3) is directly compared to the experimentally observed CdSe structures inside carbon nanotubes. The gas‐phase DFT‐calculated energy difference between “free” 3:3 and 4:2 structures (whereby 3:3 models a novel tubular structure in which both Cd and Se form three coordination, as observed experimentally for HgTe inside SWNT, and 4:2 is a motif derived from the hexagonal CuI bulk structure in which both Cd and Se form 4 or 2 coordination) is surprisingly small, only 0.06 eV per formula unit. X‐ray powder diffraction, Raman spectroscopy, high‐resolution transmission electron microscopy, and energy‐dispersive X‐ray analyses led to the full characterization of the SWNTs filled with the CdSe nanocrystals, shedding light on the composition, structure, and electronic interactions of the new nanohybrid materials on an atomic level. A new emerging hybrid nanomaterial, simultaneously filled and beta‐d‐glucan coated, was obtained by using pristine nanotubes and bulk CdSe powder as starting materials. This displayed fluorescence in water dispersions and unexpected biocompatibility was found to be mediated by beta‐d‐glucan (a biopolymer extracted from barley) with respect to that of the individual inorganic material components. For the first time, such supramolecular nanostructures are investigated by life‐science techniques applied to functional nanomaterial characterization, opening the door for future nano‐biotechnological applications.

The encapsulation of CdSe nanocrystals within single-walled carbon nanotube (SWNT)c avities of varying dimensions at elevated temperatures under strictlya ir-tight conditions is described for the first time. The structures of CdSe nanocrystals under confinement inside SWNTsw as established in ac omprehensives tudy,c ombining both experimental and DFT theoretical investigations. The calculated binding energies show that all considered polymorphs [(3:3), (4:4), and (4:2)] mayb eo btained experimentally.T he most thermodynamically stable structure( 3:3) is directly compared to the experimentally observed CdSe structures inside carbon nanotubes. The gasphase DFT-calculated energy differenceb etween "free" 3:3a nd 4:2s tructures (whereby 3:3models an ovel tubular structure in which both Cd and Se form three coordination, as observed experimentally for HgTeinside SWNT,and 4:2isamotif derived from the hexagonal CuI bulk structure in which both Cd and Se form 4o r2coordination) is surprisingly small, only 0.06 eV per formula unit. X-ray powder diffraction,R amans pectroscopy,h igh-resolutiont ransmission electron microscopy,a nd energy-dispersive X-ray analyses led to the full characterization of the SWNTsf illed with the CdSe nanocrystals, shedding light on the composition, structure, and electronic interactions of the new nanohybrid materials on an atomic level.Anew emerging hybrid nanomaterial, simultaneouslyf illed andb etad-glucan coated, waso btained by using pristine nanotubes and bulk CdSe powder as startingm aterials. This displayed fluorescencei nw ater dispersions andu nexpected biocompatibility was found to be mediated by beta-d-glucan (a biopolymer extracted from barley) with respect to that of the individual inorganic material components. For the first time, such supramolecular nanostructures are investigated by life-science techniques applied to functional nanomaterial characterization, openingt he door for future nano-biotechnological applications. tals have not yet been encapsulatedw ithin these strands, and the potentiala pplications of CdSe@SWNTsa sf unctional nanomaterials have not yet been explored. In this context, the synthesis of inorganic nanocrystals encapsulatedi nS WNTsh as been considered asapossible route for studying the properties and applicationso fl ow-dimensional materials. SWNTsh ave been filled with av ariety of materials including transitionmetal halidesa nd chalcogenides,a nd the changes in the local chemistry of nanotube-incorporated crystalsh ave been observed directly by high-resolution transmission electron microscopy (HRTEM). It is believed that encapsulationo ft hese salts introduces ac hange in the structureo ft he included material relative to its bulk form, owing to the reduced space and interactions with the walls of the SWNTs. For example, an earlier structural analysis of HgTe@SWNT showed that coordinationo f Hg and Te was alteredsignificantlyf rom the tetrahedral coordination found in the bulk HgTez inc blende structure to trigonal planar and trigonal pyramidal geometries,r espectively, in a SWNT composite. [1g, j] In some cases, the encapsulationo f1 D crystals( e.g. KI) in SWNTsy ields as tructure without an overall change, but with as ystematic reduction of coordination mode. [1e] In HRTEM, heavy atoms can be observed at higher resolution;h owever,t he positions of light atoms are ac hallenge, so theoretical model structures must be proposed to generatei mages for comparison with experimental data. In this sense, first principles calculations have been supportive in predictingt he structures of nanocrystals foundi nside the nanotube and have the bonuso fe lucidating physicala nd chemicalp ropertieso ft he composites. Althought here is a substantial computational cost associated with studying such large systems, few attempts have been made to study the structures of 1D crystals encapsulated within SWNTs. In this regard, Kuganathan and Green have shownt heoretically that HgTes tructures inside SWNTs, proposed as ar esult of the experiment, are in excellent agreement with the detailed theoretical calculations. [2] Furthermore, gas-phase density functional theory (DFT) calculations reproducing the experimental results on KI@SWNT are availableinthe literature. [3] In this work, we have filled nanotubes of two different SWNT batches (arc-madea nd CVD-made, forc omparison purposes) with CdSe, for the first time, and completely characterized the CdSe@SWNTsh ybrids emerging from the filling by Xray powder diffraction, Ramans pectroscopy,H RTEM, and energy-dispersive X-ray spectroscopy (EDX) analyses. We have also isolated and characterized (in bulk) an intermediate hybrid material, denoted as CdSe[CdSe@SWNT],w hich comprises simultaneously filled and coated nanotubes, and shows dispersibility in water and luminescencei nt he blue-to-green region of the light spectrum.T hese resultsp ave the way for the formation of new luminescentnanohybrids that retain their functionality in aqueous dispersions. Theoretical calculations based on DFT have been employed to elucidate the nature of the thermodynamically stable 1D CdSe nanocrystals that can be formed inside the SWNTs, and an evaluation of the usefulness of this materialf or sustainable chemistry applicationsw as carried out by using techniques at the interface betweenl ife sciences andn anomaterialsresearch.
We hereby report ac omprehensive structurals tudy of new functional nanocrystallinem aterials formed as ar esult of selfand directed-assembly processes under confinement from bulk materialsi nt he presence of ar elatively under-studied biopolymer,the polysaccharide b-d-glucan. This aims at understanding the nature of the CdSe crystals when confined within the SWNT cavities, the impact on their likely optical properties, and considerations of the biocompatibility mediated by this glucan. [3] 2. Results and Discussion

HybridNanoassembly Methodology
Twod ifferent batches of SWNTs[ electric arc growth (Arc-SWNT)a nd chemical vapor deposition (CVD-SWNT)] were used for comparison purposes. Both SWNT samples were filled with CdSe nanocrystals and purified following the method described by Salzmann and co-workers [4] as well as Green and coworkers [5] (specific details are given in the Experimental Section). All resultsw ere consistent with those previously reported. [5,6] In the synthetic approach ( Figure 1), an annealing process involving temperatures approximately 50 8Ca bove the meltingp oint of the CdSe startingm aterialw as used, with the cautiouse xclusion of air and moisture, to fill the SWNTs. In both cases, hybrid material 1 was isolated and showeds imilar features in SEM investigations.
Steam-purified CVD (Elicarb,T homas Swann)p ristine SWNTsw ith open tips weret horoughly characterized in bulk by using Ramana nd solid-state 13 CNMR spectroscopy to extrapolate the diameter distribution of the startingm aterial ( Figure 2). To ensure the advanced purity of the material used, and its comparable features to previous batches usedi nsimilar high-temperature-based solid-state filling methods, the pristine CVD and Arc-made SWNTsw ere firstly fully characterized by solid-state 13 CNMR, Raman spectroscopy,a nd HRTEM. The diametera nd number of concentric walls of carbon nanotubes (CNTs) can be extrapolatedf rom solid-state 13 CNMR by analyzing the chemical shifts of carbon NMR resonances. [7] In this context,t he solid-state 13 CNMR spectrum of pristine CVDmade and steam-purified SWNTsw as de-convoluted into three overlapping Gaussian peaks (d = 121.0, 136.3, and 164.7 ppm), following an approach analogoust ot hat proposed by Engtrakul, Blackburn and collaborators. [7b] The peak centered at 121 ppm, which represents 91.5 %o ft he whole sample, shows ac hemical shift consistent with those (between1 18.8 and 123.8 ppm) reported for SWNTs. [7] The average diameter of the SWNTsw as estimated to be 0.99 AE 0.20 nm. The resonances at 136.3 and 164.7 ppm may indicatet he presence of impurities, structural defects, or the presence of CÀOb onds, and are equivalent to 4.5 and 4.0 %o ft he whole sample, respectively.
The filling of SWNTsw ith metal ions through as olid-state molten-salt approachu sing CdSe requires meticulous removal of air and moisture overnight on aS chlenkl ine followed by a rather elevated-temperature insertion,a nd was performed following the established method by Sloan et al. [1e] The Elicarb SWNTss amples of advanced purity were then used for the larger scale (5 mg) crystallization of CdSe nanocrystals inside SWNTs. For comparison, the same thermalt reatment was applied to the SWNT-free bulk CdSe material. During the thermal treatment (T a > 1268 8C, under argon, with strictly controlled ambient environment during filling,u sing standard Schenk line techniques), the bulk cadmium selenide was convertedi nto a nanocrystalline material encapsulatedi nt he inner channelso f SWNTs, presumably owing to capillary wetting in the molten phase that then allowedc rystallization upon controlled overnight cooling in the constrainingp resence of SWNTso fv arying diameters. Arc-made SWNTs( of Carbolex,R ice University vari-ety,a nd furthers team-purified) were treated following the same procedure in order to fillthem with CdSe.
To clean the surface of the obtained samples and remove the CdSe particles adsorbed to the external surfaceo ft he tubes, the hybrids were cleanedb ya pplyinga na cid treatment with HCl under reflux, and then they were washed with a0 .1 m EDTA/H 2 Osolution and finally annealed at 180 8C. The obtained products were dispersed in absolute ethanol to give the sample denoted hybrid material 1 CdSe[CdSe@SWNT] or in water,i nt he presence of b-d-glucan, for advanced characterization ande valuation of their functionality,h ybrid material 3. Alternatively,i na na dvanced purificationp rocess, hybrid material 1 was treated with reflux in 0.1 m aqueous HClf ollowedb y washing with Milli-Q water and annealing, leadingt ot he sample CdSe@SWNT,d enoted hybrid material 2.T he HRTEM estimated that the filling yield was in the range of 70 %( as tatistical treatment of several TEM and HRTEM images was carried out to have an analysis as real as possible of the sample as the bulk).

Nanohybrid Characterization in the Solid State and Dispersed Phase
As described, CdSe-filled SWNTsw ere obtained by mixingt he SWNTsw ith the excessf illing material in as ilica ampoule sealed under vacuum in two-stagep rocesses. This resulted in  the formation of an anohybridd enoted as hybrid material 1 (CdSe[CdSe@SWNT],F igure 1), whichw as surprisingly stable with respect to loss of inorganic materials coatinge ven after a furtherm ild washing with diluted HCl (0.1 m)a nd H 2 Of ollowed by filtration. This hybridm aterial 1 wasc haracterizedb yT EM, SEM, ands olid-state Raman spectroscopy and compared with pristine SWNTs (Figure 3). Furthermore, beta-1,3-1,4,g lucan from barley,a lso known as b-d-glucan (i.e. aw ater-soluble derivative of aw ell-known glucan-based dispersants for SWNTs, [9] which is also known to form as ingle-stranded polymericc hain in DMSO and self-assemble as at riple helix in water) was used as the adjuvant of choice to aid the formation of an aqueous dispersion of hybrid material 1,i mproving its solubility in water,l eadingt oh ybrid material 3 (Figure 1). This was fully characterizeda nd an additional functionality was imparted by glucan coating (Figure 4). The obtained glucan-wrapped materialw as re-dispersed in H 2 Oo rH 2 O/DMSOm ixtures,g iving composites that are stable for weeks in aqueous media. Aggregateso fb arley beta-dglucan-coated SWNTss eem to emerge from dispersions after centrifugation, and these were visualized by AFM, TEM, and fluorescences pectroscopy and compared to those of CdSefree, SWNTs@glucan hybrids( vide infra). An advanced purification process of hybrid material 1 was carriedo ut, leading to CdSe@SWNT,d enoted hybrid material 2.F or hybrid material 2, only am inimum inorganic extraneousm aterial( CdSe or decomposition products)w as found by HRTEM/EDX and XRD in both types of SWNTs(CVD-SWNTsand Arc-SWNTssamples).
The presence of ad iversen anotube size distribution in the sample was also observedi nd ispersions of the hybrid materials in natural biopolymers such as b-d-glucan (from barley,A ldrich) diffusion-oriented spectroscopy (DOSY)e xperiments on the 1 HNMR spectrumo fb-d-glucan-wrapped CdSe- [CdSe@SWNT],h ybrid material 3.F igure S20 (see the Supporting Information) shows the DOSY 1 HNMR spectrum of hybrid material 3 in a1:1 D 2 O: [D 6 ]DMSO solution, within the region of the spectrum associated with the signals of glucan;s ix NMR spin systems can be observed, with values of diffusion coefficients between 1.61 10 À10 AE 1.06 10 À11 and 7.90 10 À12 AE 1.04 10 À11 m 2 s À1 .T herefore, it can be concluded that, when in solution, the fibers of b-d-glucan wrap the SWNTsa nd generate nanohybrid species of different sizes, which are soluble in D 2 Oa nd [D 6 ]DMSO. Ramans pectroscopy (830 nm laser) measurements of CVD-SWNTsa nd dispersed CVD-SWNT (in presence of b-d-glucan) were carried out to estimate the density of structural defects in CNTs. The intensity ratio of the Raman D band to G band (I D /I G )p rovides ar elative measure of the density of structural defects and, therefore, the structural qualityo fasample. [8] If both of these bands are similari ni ntensity,t he density of structural defectsi sa ssumed to be high. Figure5 shows the Ramans pectra for CVD-SWNT andg lucan-CVD-SWNT samples. It can be observed that the startingm aterial presents aDb and at 1288 cm À1 and aGband at 1584 cm À1 .
The I D /I G ratio of the CVD-SWNT material, reportedi nT able 1, was found to be 0.107, indicating al ow level of surface defects and disorder. [8] The sharp Da nd Gb ands also confirm that the provided tubes are single-walledr ather than multi-walledi n  structure. The CVD-SWNTa lso exhibits three sharp peaks below 300 cm À1 (146, 160, and 233 cm À1 ), which correspondt o the radial breathing mode (RBM). The calculated diameters from these RBM bands are 1.62, 1.50, and 0.85 nm, respectively. The other peaks observed between 300 and 1300 cm À1 are probably due to residual metal nanoparticle catalyst impurities (majorly Fe residual). It is also expected that they are wrapped in carbonaceous material. On the other hand, the Raman spectrum for the glucan-CVD-SWNTss hows as hift in the Gb and to higher wavenumber (1590 cm À1 ), owing to the presence of glucan on the surface. The Db and appears at the same position (1288 cm À1 ). The I D /I G ratio increased to 0.342,w hich can be attributed to the presence of glucan adsorbed on the surface and/or an increased on the level of surface defects and disorder. However, the value is still low,i ndicating ag ood structural quality.T he peaks corresponding to the RBM are also shifted with respect to the CVD-SWNTs( 152, 189, 237 cm À1 ), owing to wrapping with glucan.
The morphology and structure of CVD-SWNTsw ere investigated by HRTEM. Figure 6s hows the microscopy images of the ultrapure CVD-SWNTs, which confirm good structuralq uality with clean surfaces. The SWNT diameters are in agreement with those estimated from Ramanand 13 CNMR data.
The grade of crystallinity of pristineS WNTsa nd purified CdSe@SWNTs( hybrid material 2)w as further investigated by Xray diffraction ( Figure 7). As inferredf rom the pattern depicted in Figure 7a,t he CVD-SWNT sample is notc rystalline, whereas the SWNTsp repared by electric arc growth present some crystallinity, as confirmed by the Bragg peaks in the spectrum (Figure 7b). XRD analysiso ft he CVD-SWNT sample after the thermal treatment with CdSe[CdSe@CVD-SWNT] shows Bragg peaks that can be attributed to carbonaceous subspecies and/ or surface defects, thereby indicating the partial degradation of the SWNTsu nder thermalt reatment. The Arc-SWNT sample after the thermal treatment with CdSe (CdSe@Arc-SWNT) also shows peaks corresponding to carbonaceous subspecies or surfaced efects, but the pattern also shows the existence of a hexagonal CdSe phase (Cadmoselite, ICDD 08-0459) together with ahexagonal seleniumphase (ICDD 06-0362).
TEM images of the CdSe@SWNTs( hybridm aterial 2)s how structuralf ragments of intercalated materiali nb oth samples, but to aw ider extent in the CdSe@Arc-SWNT sample (Figures 8a and 8b). So, it is confirmed that, during the thermal treatment, the CdSe melt was encapsulated in the inner channels of SWNTs, owing to capillaryw etting. The EDX analyses confirm the presence of both Cd and Se in the CdSe@CVD-SWNT sample.
After careful scrutiny of the HRTEM images at severald ifferent magnifications (see the Supporting Information), there is no evidenceo fC dSe or Se aggregates outside the NTs. Close examination of the encapsulated crystal by electron diffraction completed with HRTEM in the CdSe@Arc-SWNT (Figure8c) shows that dark spots can form au nit with the shape of ah exagon, and this unit is regularly repeated along the tube axis. Interestingly,t his hexagonal pattern can be associated with the crystal phase of the CdSe (Figure 8d)f illing inside the tubes, which is in agreement with the hexagonal CdSe phase observed by X-ray powder diffraction. The spots observed across the SWNT capillary are spaceda ta veragei ntervals of   )i ndicatest hat CdSe inside the nanotube is confined and the structure is compressedi no rder to adapt to the available internal space, as determined by the nature of the nanotube host employed. The diameter of someo ft he observed SWNT samples is found to range between 1.4 and 1.7 nm, which is closetothe diameter of atypical (9,9) SWNT. Raman spectroscopy measurements allowed the identification of the inner structure of the CdSe@Arc-SWNT sample (hybrid material 2), and it shows the optimum achievable filling of the nanotubes with CdSe nanowires under the solidstate conditions employed (Figure 9). The starting material presentsaDb and at 1289 cm À1 and aGband at 1578 cm À1 . The I D /I G ratio of this startingm aterial sample of SWNTsw as found to be 0.36. Such an I D /I G ratio indicates al ow level of surfaced efects and disorder. [8] The sharp Da nd Gb ands also confirmed that the provided tubes are single-walledr ather than multi-walled in structure. The startingA rc-SWNTsa lso exhibit three sharp peaks below 300 cm À1 (at 141, 156, and    cm À1 ), whichc orrespondt ot he RBM. The calculated diameters from the RBM bandsa re 1.59, 1.43, and 1.01 nm, respectively.T he Raman spectrum of intact SWNTsa lso exhibited several peaks from 300 to 1300 cm À1 ,w hich were probably caused by the residual metal nanoparticle catalysti mpurities (mainly Fe and Cr traces from the catalytic process). It is also expectedt hat they are wrapped in carbonaceous materiala s suggested by HRTEM.
The I D /I G ratio of SWNTsa fter thermal treatment was slightly reduced to 0.33, and this decrease can be assignedt ot he reductiono fs urface defects (Figure 9). Only the bands assigned to SWNTsw ere found in the Raman spectrum of SWNTsf illed with CdSe ( Figure 9). However, Ramanb ands have undergone ashift to higher frequency.Itisknown [9b-d] that ahigh-frequency shift of Ramanb ands of filled SWNTst akes place if the encapsulated materiali sa na cceptor. The observation of this shift in the spectrum of CdSe@Arc-SWNT is probably due to the charge transfer from SWNTst oC dSe. The Raman spectrumo f CdSe@Arc-SWNTr emained almost unchanged with respect to the Arc-SWNTsa nd exhibited three RMB peaks with maxima at 232, 156, and 143 cm À1 .This suggeststhe formationo fp urified tubes with calculated diameters of mainly 1.01, 1.43, and 1.62 nm, respectively,and rather narrow distribution ranges.

Luminescence Studies in the Dispersed Phase for FunctionalNanohybrids
There is an increasing demandf or water-soluble in vitro and in vivo molecular probes for imaging ande arly detection of cancers,a nd we demonstrate that water-solubilized b-d-glucan carbon nanotubes have an acceptable level of biocompatibility to enable the evaluation of their cellular uptake. The newly synthesized nanohybrids showed biocompatibility comparable with that of the free glucan, both in healthy and in cancerous cells. Luminescent nanowires obtained under extreme conditions display,f or the first time, CdSe-decorated (inside as well as outside, depending on the conditions employed) carbon nanotube strands. These can be "recognized" and individually modifiedb yt he polysaccharide fibers of b-d-glucan, whicha ct as adjuvants, allowing the complete full characterization of the encapsulated materials in water media.
To evaluate the potential applications of the obtained hybrid materials, the photoluminescence properties of the hybrid functional materials (hybrids 1, 2,a nd 3)i na queous media were investigated and compared to those of CdSe quantum dots and commercially available core-shell CdSe/ZnSnanocrystals (Lumidot TM 480). All of these materials were imaged by laser scanning confocal microscopy on thin films, thereby adaptingt hese techniques available at the life science interfaces to address characterization and functionality probingc hallenges in nanomaterials.
Confocal fluorescence microscopy of ad ried thin film of water soluble hybrid material 3 was performed, and the corresponding images are showni nF igure 10. The figure shows bright-field images (Figures 10 a, 10 f, and 10 Figures 10 k-o). The glucan-wrapped water dispersible hybrid material 3,f orming at hin film, was found to emit in the blue (417-477 nm) and red (570-750nm) range when excited with a4 05 nm laser light.Adirect comparison of such emission performances with those of hybridm aterial 2 ( Figures  S5 a-e) seemst os uggestt hat the coating of the CdSe@SWNT with glucan directly affects the absorption/emission properties of the composite material, enhancing its capability to emit at higher frequencies. Interestingly,i fw ei rradiated at hin film of CdSe with a4 88 nm excitation laser light, the microscopy images (Figures 11 a-d) reveal as trong fluorescence intensity in both green (l em = 417-477nm) and red (l em = 570-750nm) channels, and only in green channel for at hin film of the commerciallya vailableC dSe Lumindot TM 480 nm. Such emission properties are significantly quenched when CdSe structures are confined within nanotubes or glucan-nanotubes nanostructures, as seen in Figures 11 i-l and1 1m-p,r espectively,s uggesting the formation of ap hotoexcited energy-transfer complex. [10] Single-photonl aser-scanning confocal microscopy of CdSe, in the form of the commerciallya vailableL umindot TM 480 nm and CdSe@SWNTsu sing 405, 488 and 561 nm laser excitation lines, is reported in the Supporting Information for comparison.

Computational Structural Investigations
As previously mentioned, the diameter of someo ft he observed SWNTsw as found to range between 1.4 and 1.7 nm, which is close to the diameter of at ypical (9,9) SWNT.T hus far, structuralm odelsa re not established for CdSe@SWNTsp roducts and X-ray diffraction studies coupled with DFT modelling were necessary to shed light on the structure of encapsulated nanocrystals. In ap revioust heoretical study on CuI@SWNT,w e have predicted novel structures for CuI crystals that mayb e found in SWNTs. [22] The startings tructures of CuI crystals were based on the HgTe structures derived from bulk and observed within SWNTsa st he both Cu + and Hg 2 + have both d 10 ions. As Cd 2 + is also ad 10 ion, it is possible to proposeastarting structure for CdSe nanocrystals based on the HgTeo rC uI structures availableinthe literature. [1g, 4, 10] Theoretical calculations have been employed in this work for the aid of understanding the structure and stability of CdSe@SWNT hybrid material. Computational parameters used in this study using CASTEPc ode [11] (see the Experimental Section) are consistent with those reportedi no ur earlier studies, [12] where parameters (ultrasoft pseudopotentials generated using the "on the fly" formalism, the cut-off energy of 320 eV and Monkhorst grid k-points) were checked for convergence, and the resultso btainedr eproduced the experimental observations.T he specimens studied hereby showedt hat av ariety of diameters are available for CNT-templated filling with CdSe, ranging from 1.1 to 3.0 nm for the internal diameters of CNTs,  Figure 8c showsaselected HRTEM-imaged CdSe-filled CVD MWNT strand as an example (whilst additional images of SWNTsa nd DWNTsa re shown in the SupportingI nformation), whereby ab road CdSe nanowire is encapsulated in am ulti-walled CNT with an inner diameter of at least 3.0 nm. However,t he vast majority of the tubes observed showeda na verage inner diameter of maximum diameter of approximately 1.4 nm. As such, we focusedo ur attention on the theoretical modelling of the tubes with average diameters of 1.4 nm and chose as eries of simplified models to investigate furtheratacomputational level.
Three different startings tructures are proposed for the CdSe nanocrystal. The first structurei sanovel structure relatedt o the HgTen anocrystal observed within SWNTs( 3:3) and the second one is al ow-dimensional structure derived from the NaCl rock salt bulk structure with 4:4c oordination. [2] The third structurei samodel structure related to that predicted for CuI nanocrystalsf ound inside SWNTs( 4:2). [13] The lowest energy structureo ft hese model structures are encapsulated in a( 9,9) nanotube based on the experiment and electronic structure and properties are calculatedi nt he following section. First principles DFT calculations were performed on 1D CdSe nanocrystalse ncapsulated within SWNTsi no rdert od etermine the electronic structure and the nature of the interaction between the nanocrystals and SWNTs. The CASTEP code, [11] which solves the standardK ohn-Sham (KS) equations using plane wave basis sets, was employed in all calculations.
For calculations on infinite CdSe crystals and CdSe@SWNT composites, periodic boundaryc onditions were appliedt oe nforce am inimum lateral separation of 25 between structures in adjacent unit cells. At this separation, the interaction between these structures and their periodic images are negligible.
For the DFT modelling of 1D crystalso fC dSe, the geometries of three different types of CdSe structuresw ere optimized in the absence of ac onfining nanotube and using periodic boundaryc onditions. The initial structures are shown in Figure 12 A, and further optimized modeled structures in the Sets of Figure 12 Band 12 C.
The first modelled structure is directly linked to the experimentallyo bserved novel structure for CdSe within SWNT with 3:3c oordination.I nt his structure, aC da tom is in at rigonal planar coordination and Se atom has apyramidal environment. The second structure is a2 2l ow-dimensional structure derived from the NaCl rock salt bulk structure with 4:4c oordination of the atoms. The third structure is based on the model structure predicted for CuI structure (derived from the hexagonal CuI bulk structure), which has 4:2c oordination.A ll three structures were considered as infinite 1D crystalsa nd calculations were performed with periodic boundary conditions. The geometry optimization of these three structuresi ndicates that CdSe (3:3) structure is the lowest in energy. The second most stable structure is 4:2( higheri ne nergy by 0.06 eV). The relaxed structure of 4:2f orms another three coordinated structure (see Figure 12 Ch,w hich is closer to the 3:3s tructure in energy.T he optimized structuresa re shown in Figure 13 and  Table 2. The optimized free 3:3s tructure within periodic boundary conditions gives as tructure that resembles the startings tructure. However, CdÀSe bond lengths are slightly shorter compared to the startingb ond lengths proposed for HgÀTe bonds (2.80 ). Calculated bond angles are close to the values predicted for HgTec rystals found inside SWNTs. The structural parameters (bond lengths and angles) of the optimized structure of the CdSe crystal are given in Figure 12 Cg.O ptimization of the 4:2c rystals gave as lightly distorted structure with threefold coordination compared to the initial structure with the CdÀSe bond lengths 2.63 ,S e ÀCdÀSe angles close to 130.08, and CdÀSeÀCd angle 83.68,a ss hown in Figure 12 Ch,a nd did not lead to the lowest energy 3:3s tructure. The optimized structure of a free 4:4CdSe crystal with periodic boundary conditions gave ar hombohedral structure (Figure 12 Ci). The distances of the CdÀSe bonds (2.73-2.76 )w ere longert han the distances found in the 3:3s tructure (2.28-2.72 ). Calculated bond anglesS e ÀCdÀSe and CdÀSeÀCd are found to be 107.6 and 72.38,respectively.
Model studies focusedo nasimplified selection based on three different types of polymorphs of CdSe that were encapsulatedw ithin a( 9,9) nanotube and calculations in the gas phase werep erformed. The calculated CdÀSe bond lengths, binding energy,a nd charge transfer are given in Ta ble 3.The calculated binding energies indicate that all three composites can form aC dSe molten state in the experimental approach. The main contribution for binding energy was found to be the van der Waals interactions in such tubes;h owever,t he reasons for the formation of nanowires in aC NT-templated environment are likely to be kinetic, as well as thermodynamic, in nature.O ur calculations of the density of states (DOS) for a typical (9,9)-SWNTs how that filling CdSe nanowires into the nanotube sidewall only very subtly perturbst he electronic structure of the SWNT,l ikely owing to non-covalenti nteractions between the p-electronic system of the host nanotube and the hexagonal network of the CdSe chains. Such DOS data   furthers uggest that the formation of nanocrystals inside the SWNTsc annot be explained by the consideration of the van der Waals interaction only.T he calculated binding energies and DOS in this study show that the interaction between nanotubes and the CdSe crystals is non-covalent in nature. This is further confirmed by the experimental observations (from HRTEM and Ramans pectroscopy) that the tube wall was not significantly damaged. These calculations show the formation energy of CdSe nanocrystals inside the nanotube, assuming that the CdSe crystals are already available: these also show that all three crystalsc onsidered are thermodynamically stable inside the tube. In ab ulk covalent materials uch as CdSe, changes in the bonding arising directly from low dimensionality are the dominantd rivingf orce towards structural change, and interactions with the tube wall (predominantly van der Waals) are of secondary importance.T he computational calculations interprett he experimentally observed 3:3C dSe tubes in terms of energy,compared with the other polymorphs considered. However, the magnitude of binding is dependent on the exchange correlation functional and may well change once the dispersion corrections are included;h owever,t his aspect has not been explored in this study.

Biocompatibility Evaluationsfor Glucan-Wrapped Nanohybrids
To evaluate the potentiala pplication of the obtained water soluble for glucan-wrapped nanohybrids (hybrid material 3)i n biosensing devices, biocompatibility evaluations were carried out.

The inhibitory concentration (IC 50 )o fC dSe[CdSe@SWNTs]
composites in PC3 under normoxia was evaluated by MTTs, giving an estimated IC 50 of (9.69 AE 1.81) 10 À6 mg mL À1 (from six measurements).F or pure raw SWNTsi ni solated and independente xperiments,t he IC 50 was in the range of 1.17 10 À7 to 6.3 10 À7 mg mL À1 under the same conditions, showing the expected high toxicityo fu n-functionalized SWNTs. Thisc ontrasts with the decreased toxicityo fh ybrid material 3,w hich ranges from 1.08 10 À5 to 7.40 10 À5 mg mL À1 ,d epending on cell line:H eLa or PC3. In cytotoxicity evaluations in HeLa cells,   under similar conditions using covalently functionalized SWNTs, the IC 50 was reported to be in the range of IC 50 @ 5 10 À7 m. [23] However,u nexpected IC 50 values emerged for pure CdSe crystalsinindependent estimations. [24] The IC 50 values indicated high toxicity,b ut we did not find them reproducible;t his is likely to be due to av arying cellular uptake of samples caused by the low solubilityo ft his inorganic material. Upon addition of the b-d-glucan and the formation of encapsulatings trands aroundC dSe[CdSe@SWNTs] fibers, an increased solubility of hybrid material 3 significantly facilitated its cellular uptake. Figures 14 and 15 show the IC 50 of CdSe@SWNTs@glucan and b-dglucan required to reduce the cell growth by half. In general,a lower IC 50 value indicates am ore toxic sample. Figure 14 shows that the polysaccharide b-d-glucan has comparable values of IC 50 in either FEK4 or PC-3 cell lines,o f( 6.64 AE 0.35) 10 À5 or (6.72 AE 1.08) 10 À5 mg mL À1 ,r espectively.T he inhibitory concentrationso fC dSe@SWNTs@glucan in PC-3 andF EK4 differ by an order of magnitude,a nd their values of IC 50 in FEK4 and PC-3 are (1.01 AE 0.29) 10 À4 and (4.58 AE 0.76) 10 À5 mg mL À1 ,r espectively. These resultsi ndicate thath ybrid material 3 (denoted glucan@CdSe[CdSe@SWNTs]) display a higher toxicityinc ancerous cellsthan in healthy cells, in agreement with previouss tudies regardingt he preferential uptake of polysaccharides by cancer cells. The cytotoxicity estimations for PC-3 cells treated with hybrid material 3 (glucan@CdSe-[CdSe@SWNTs]) as well as b-d-glucan alone, over 48 ha t3 78C under hypoxia-induced condition, are presented in Figure 15. The viability of cancerousc ells was determined by using crystal violet assays. The IC 50 values estimated for b-d-glucan and(glu-can@CdSe[CdSe@SWNTs]) were found to be (5.30 AE 2.00) 10 À4 and (9.54 AE 6.36) 10 À5 mg mL À1 ,r espectively.T herefore, these estimations seem to suggestt hat under the conditions where PC-3 cells were grown in conditions deprived of oxygen, mimicking the acute hypoxicc onditions present within certain tumors,h ybrid material 3 (glucan@CdSe[CdSe@SWNTs]) is likely to show approximately ten times more toxicityt han b-dglucan alone. The significantly reduced toxicity for quantum nanomaterials as well as for carbon nanotubes in the presence of this biopolymer is remarkable,a nd consistent with our very recent observations in new radiolabeled and simultaneously filled SWNT nanohybrids, which showed promisingb ehavior in living systems in terms of kinetic stabilitya nd probe integrity with respect to loss of internal/external functional moieties. [3c]

Conclusions
For the first time, pristine SWNTsw ere filled with CdSe nanowires on al aboratory scale, giving riset ot hus-far elusive luminescent nanohybrids suitable for functional biomaterial investigations. Their emergence was madep ossible by operating at elevated temperature and in the advanced exclusion of air and moisture during the nanosynthetic protocols. The intimate structure of CdSe nanocrystals following encapsulation inside pristine SWNTse mergedb yc ombining experimental and theoretical investigations. One batch of SWNTsw as as made by electric arc growth (Arc-SWNT) and the other by chemical vapor deposition (CVD-SWNT). The filling yield, estimated by HRTEM is high (above6 0%), and is comparable with that normally expected for such solid-state encapsulation techniques. The new obtained CdSe@SWNTsh ybridsw ere characterized by X-ray powder diffraction, high-resolution transmissione lectron microscopy,a nd Ramans pectroscopy.X -ray diffraction together with HRTEM and EDX analyses confirm the filling of the SWNTsw ith ah exagonal CdSe phase. The lower interplanar crystal phase of the CdSe filling the SWNT,i nc omparison with the spacing for af ree hexagonal CdSe, indicates that CdSe inside the CNTsi sc onfined and the structure is compressed in order to adapt to the internal space of the CNT.W eh ave also found ah igh-frequency shift of the stretching modes in Raman.T hiss hift of the Ramanb ands might imply charge transfer from SWNTst oC dSe, in which CdSe nanocrystals are the acceptor. DFT calculations validated the predictions on possible structures forC dSe that can be found inside the SWNT under the experimental conditions applied. The geometries of three different 1D CdSe nanowire standsw ere optimizedf rom differents tartings tructures. The lowest energy structure was found to be the 3:3f orm, whereas the next thermodynamically stable (with lowest minimum energy) structure was 4:2. The calculated binding energies show that, although all three polymorphsm ay be found experimentally,t he most thermodynamically stable structure (3:3) is directly comparable to the CdSe structures widely observede xperimentally inside these carbon nanotubes. DFT calculations in the gas phase indicate that there is no significant modification or damage found in the SWNT walls under the influence of CdSe, which is consistent with experimentalo bservations from Raman spectroscopy and X-ray diffraction. This seems to suggestt hat interactions with the tube wall, predominantly van der Waals interactions, are of secondary importance in stabilizingt he resulting geometry under confinement. The reasonsf or the formation of nanowires in aC NT template environment are likely to be kinetic, as well as thermodynamic,i nn ature. The binding energy and charge transfer between CdSe crystalsa nd the SWNTsw ere modeled by using DFT,a nd this information may assist experimentalists in their furthers tructural interpretations of semiconductorsn anoparticles inside SWNTs. Finally,t he inhibitory concentration (IC 50 )o fanew luminescent, functional materiald enoted glucan@CdSe[CdSe@SWNTs] (hybridm aterial 3)i n healthya nd cancerous cells was evaluated. These studies showedt hat, under normoxiaa nd hypoxiac onditions, hybrid material 3 shows only marginally increased toxicityf or cancer cells with respectt ot he pristine polysaccharide fibers of b-dglucan, highlighting the fact that the use of such adjuvants can mediate the high cellular toxicity of toxic materials such as CdSe@SWNTsa nd also allows the complete characterizationo f the encapsulatedm aterials in water media.
This work sheds light on the nature of encapsulated nanocrystalsi nh ierarchical and biocompatible CdSe-SWNT-glucan hybridsb yu sing techniques situated at the interface between materials chemistry research andl ife sciences, such as confocal fluorescencei maging and biocompatibility assays.S tructural predictionso ft he luminescent nanowires formed under confined conditions in the solid state are validated by DFT calcula-

Experimental Section
Arc-and CVD-synthesized SWNTsw ere available from Carbolex (with average diameters of 1-2 nm) or from Thomas Swann (with average diameters of 1.4-5 nm). Steam refinement of raw Arc-and CVD-synthesized SWNTsw as used here as the method of choice to minimize the presence of external defects and functional groups into the scaffold whilst simultaneously opening the tubes.

Advanced Purification and CdSe Filling of SWNTs
The purification of as-made CVD SWNTs( Thomas Swann, Elicarb) was achieved by introducing steam carried by an argon flow through the reactor.S WNTs( 500 mg) were introduced in aq uartz reactor (9 mm diameter) placed inside of the quartz tube (5 cm diameter) and within the alumina lining of af urnace tube, which was flushed with argon for at least 1h prior to the experiment to minimize the presence of air.A rgon-degassed steam was introduced by bubbling argon through boiling water.T he furnace temperature was gradually increased to 900 8Co ver approximately 30 min and the reaction was maintained for at least 20 min and no longer than 2h,d epending on the sample size. The resulting sample was refluxed in 16 %HCl overnight, which removes most of the metal catalyst, and washed with NaOH and double-distilled water on a2mmm embrane filter until the filtrate was neutral pH. Af urther toluene washing/filtration step was carried out in order to obtain the highest purity SWNTsa fter filtration. Prior to filling, the tubes were standardized and dispersed by sonication at room temperature for 10 min. We found that, when arc-tubes were used, only av ery small purification yield was obtained. To gether with CdSe, the nanotubes were then kept in aS chlenk tube and heated at 50 8Ca bove the melting point of CdSe (CdSe, Sigma-Aldrich, mp. 1268 8C) for 12 h. The annealing process was carried out under an inert atmosphere by using standard Schlenk techniques and glassware in order to exclude air and moisture from the reaction vessel. The resulting CdSe[CdSe@SWNTs] materials were refluxed in HCl for 1h,w ashed with a0 .1 m EDTA/H 2 Os olution, and finally heated at 180 8C, giving ap urified CdSe[CdSe@SWNTs] hybrid material 1.T oo btain hybrid material 2( CdSe@SWNTs), 1 was refluxed in 0.1 m aqueous HCl for 1h,w ashed with Milli-Q water,a nd heated at 180 8C. Hybrid material 3 was obtained by dispersing CdSe [CdSe@SWNTs] (1)i nw ater (1 mg mL À1 ). b-d-Glucan from barley (1 mg mL À1 )w as heated and sonicated (15 min) in DMSO. The two suspensions were mixed, sonicated for 15 min, and stirred for 2h.T he resulting composites were dried under reduced pressure for 12 ha nd re-dispersed in distilled water (2 mL). The SWNTcontaining samples (ca. 20 %b yw eight) were isolated from H 2 O after approximately 3h of centrifugation. The aqueous layer was removed by af reeze-drying procedure and the samples denoted (glucan@CdSe[CdSe@SWNTs]) were re-dispersed in EtOH for TEM/ HRTEMo rA FM analysis.

Powder X-ray Diffraction
Analyses of the crystalline structures and phase identification were performed by X-ray diffraction (XRD Bruker D8 ADVANCE) with a monochromatized source of Cu K a 1r adiation (l = 1.5406 nm) at 1.6 kW (40 KV,4 0mA). The samples were prepared by placing a drop of ac oncentrated ethanol dispersion of particles onto a single-crystal silicon plate.

TransmissionElectronMicroscopy (TEM)
TEM images were obtained with aG atan Dual-vision digital camera on aJ EOL 1200EXII transmission electron microscope coupled with EDX spectroscopy (point resolution, 0.16 nm). The operating voltage was 120 kV.H RTEMi mages were obtained from the sample (deposited on aL acey carbon film-copper grid, purchased from Agar Scientific) on at ransmission electron microscope (JEM-2100 LaB6 or JEOL 2100 FEGTEM) operated at 100 kV.T he differing imaging approaches and repeated experiments showed av ery high batch-to-batch consistency of the nanohybrids synthesized. For the glucan-coated nanocomposite (hybrid material 3)a nd free CdSe, as well as for the uncoated hybrid materials 1 and 2,H RTEM images were also recorded on aJ EOL 3000F field-emission gun instrument. This instrument is equipped with an Oxford Instruments EDX spectrometer with as uper atmospheric thin window (SATW) detector that allows chemical analysis of elements down to boron under suitable conditions. In addition, aG atan imaging filter (GIF) equipped with a2 k7 94IF/20 MegaScan CCD camera allows chemical analysis using electron energy loss spectroscopy (EELS).
Generally,f or glucan-wrapped SWNTs( filled or unfilled), all examined samples showed an excess of organic materials forming a film, so TEM imaging of the glucan-wrapped CdSe filled SWNTs was rather challenging. Nevertheless, AFM imaging confirmed the presence of the glucan-wrapped material (hybrid material 3).

Raman Spectroscopy
Raman spectroscopy was carried out on aR enishaw inVia Raman spectroscope. The measured samples used in this project normally contained carbon nanotubes, graphene oxide, or reduced graphene oxide. The specimens were either in the solid state or dispersed in pure water (Milli-Q) or aw ater/ethanol (1:1) mixture. During the measurement, carbon nanomaterials samples were deposited on an aluminum plate substrate. In some cases, silicon wafers or ag lass lens were also used as the substrate. The input wavelength was set at 830 nm for SWNTsand their composite samples. For the hybrid 1 nanocomposite and its glucan derivative, spectra at 514 nm were also recorded. More than ten repeats were applied in the Raman spectroscopy measurements to achieve sufficient signal-to-noise ratios, and the beam was focused on at least three different positions across the specimen;t hese spectra were averaged to obtain representative peaks of the sample.

Atomic Force Microscopy (AFM)
AFM images were recorded on aD igital Instruments Multimode SPM instrument with aNanoscope IIIa controller.This was operated in tapping mode with a" J" scanner having al ateral range of approximately 100 mma nd av ertical range of 6 mm. Silicon probes (Nascatec GmbH model NST NCHFR) were used, with resonant frequencies of approximately 320 kHz. Calibration of the AFM was accomplished by scanning a1 0mmp itch with 200 nm 3D reference from Digital instruments. Contact-mode AFM was found to be unsuitable (despite its better lateral resolution with respect to tapping-mode AFM), as it physically removed any deposits from the area scanned.

Solid-State NMR
The solid-state 13 CNMR spectrum of pristine steam-purified SWNTs was recorded at the Inorganic Chemistry Laboratory,U niversity of Oxford, using am ixture of SWNTss piked onto ab ulk, anhydrous KBR sample that was homogenized by grinding in ag love box under N 2 .S pectra were recorded on aB RUKER AVANCE III HD 400 spectrometer equipped with a4mm magic angle spinning (MAS) probe at 101 MHz at room temperature (298 K). All NMR spectra were acquired by using MAS with as pinning speed of 10 kHz and were fully interpreted for this type of SWNTsf or the first time.

Confocal Microscopy
Confocal fluorescence microscopy was performed by using aN ikon eclipse Ti-E inverted microscope with 60 oil objective lens, equipped with an LU-N laser unit and three continuous visible lasers (405.0, 488.0, 561.0 nm). All images were processed by using functions within the NIS elements software package. Thin films of relevant samples were prepared by drying out their suspensions onto ahorizontal surface.

Computational Details
First principles DFT,a si mplemented in the CASTEP code, [11] was employed to optimize structures and plot densities of states of SWNTsc ontaining different 1D CdSe polymorphs. The exchange and correlation interactions are described by using generalized gradient approximation (GGA) parametrized by Perdew,B urke, and Ernzerhof (PBE). [14] Ultrasoft pseudopotentials were generated by using the "on-the-fly" formalism in CASTEP.Ap lane-wave basis set with the energy cutoffo f3 20 eV was used to expand the wave function. Structure optimizations were performed using the BFGS algorithm and the forces on the atoms were obtained from the Hellman-Feynman theorem including Pulay corrections. In all optimized structures, forces on the atoms were smaller than 0.05 eV À1 .Asingle k-point (G)w as used for all calculations on molecular CdSe crystals and for the CdSe@SWNT composites. In calculations on CdSe with periodic boundary conditions, reciprocal space was sampled at between 1-5 k-points by using the method of Monkhorst and Pack. [25]