Selective Enhancement of Inner Tube Photoluminescence in Filled Double‐Walled Carbon Nanotubes

A highly selective enhancement of the optical response of the inner tubes of double‐walled carbon nanotubes has been identified upon transformation of the residual C atoms inside the hollow core to linear carbon chains (LCC). By varying the growth conditions and using standardized suspensions, it has been observed that this optical response depends sensitively on the tube diameter and LCC growth yield. It is reported how the formation of LCC by postsynthesis annealing at 1400 °C leads to an increase of the photoluminescence (PL) signal of the inner tubes up to a factor of 6 for tubes with (8,3) chirality. This behavior can be attributed to a local charge transfer from the inner tubes to the carbon chains, counterbalancing quenching mechanisms induced by the outer tubes. These findings provide a viable pathway to enhance the low PL quantum yield of double‐walled carbon nanotubes and proof the capability of inner tubes to exhibit photoluminescence.


Introduction
Double-walled carbon nanotubes (DWCNT) are a class of carbon nanostructures formed from two graphene sheets conceptually rolled into two concentric tubes. As such, they combine the outstanding properties of single-walled (SW) and multiwalled carbon nanotubes (MWCNT): The outer tube acts as a shield for the inner tube, granting a very high stability even when confronted with highly aggressive mechanical, [ 1 ] thermal, [ 2 ] and chemical treatments. [ 3 ] The latter is of special importance since the optical properties of SWCNT are easily altered or even suppressed due to environmental effects. [ 4 ] Iakoubovskii et al. could reveal in optical absorption and photoluminescence (PL) measurements that this shielding process in DWCNT is effi cient even when ozone oxidation is applied, [ 5 ] and Hayashi et al. had similar observations of PL in fl uorinated DWCNT. [ 6 ] Despite this fascinating effect, the major but also that the LCC invoke a charge transfer from inner tube to the LCC which depends on their length, and that the length distribution can be tailored by using appropriate growth conditions (especially the synthesis temperature and inner tube diameter). [ 20 ] This charge transfer from inner tube to LCC in return has the same order of magnitude as the one reported by Zólyomi et al. [ 8 ] for the charge transfer from outer tube to inner tube, therefore potentially neutralizing the outer tube's infl uence in order to restore the inner tube's intrinsic optical properties. At last, an essential aspect for the discussion of the earlier mentioned capability of inner tubes to exhibit PL is that the LCC could only be grown inside the inner tubes of DWCNT and not within SWCNT of the same diameter. [ 19 ] With this contribution we show for the fi rst time that functionalization with linear carbon chains represents an ideal pathway to selectively enhance the optical response of inner tubes from DWCNT. By using standardized suspensions of DWCNT either pristine or fi lled with LCC grown at different temperatures, we demonstrate that the optical absorption and PL of the inner tubes does not only depend on their chirality but also on the chain length distribution within the sample. PL line scans clearly show a diameter dependent optimum for this enhancement and three different regions regarding the inner tube diameter. Concomitant to previous results, these regions can be assigned to nanotube species being ideal for the growth of the longest linear carbon chains as well as those tubes with an optimum balance between diameter and chain length for PL enhancement or tubes being too large.

Characterization of LCCs@DWCNTs Grown at Different Temperatures by Raman Spectroscopy
The DWCNT samples used in this study were synthesized with a procedure recently reported elsewhere. [ 14,19 ] The same also accounts for the growth of linear carbon chains inside the inner tubes of the DWCNT. [ 19 ] Both procedures and the details of their growth inside DWCNT with thin inner tubes are summarized in the Experimental Section provided after the main text. We have optimized the diameter distribution of the DWCNT for inner tube diameters between 0.6 and 1.1 nm suitable for near-infrared PL by a multifrequency Raman analysis of the radial breathing modes. [ 21 ] Regarding the LCC fi lling, our previous study revealed that only DWCNT with inner tubes between about 0.6 and 0.9 nm are suitable for a high yield growth of LCC and that more than 90% of those tubes are fi lled by LCC. Details about the determination of this bulk fi lling with LCC are given by Shi et al. [ 19 ] and in the Experimental section. Before deepening into the effect of LCC encapsulation inside DWCNT on the optical properties of their inner tubes, we have to revisit the Raman response of DWCNT samples fi lled with LCC grown at different temperatures. In bulk measurements, LCC can be detected by Raman spectroscopy, as they exhibit a Raman active mode in the range between 1790 and 2200 cm −1 . The frequency of the observed Raman shift of the mode assigned to the carbon chains (LCC-Band) is strongly correlated to the chain length, [ 19,20,[22][23][24][25][26] with longer chains giving smaller shifts. Consequently, when investigating on a bulk sample with different chain lengths, the transformation of the convoluted LCC signal line shape can be explained by different chain length distributions between the samples, meaning that a higher intensity on the low frequency side indicates a higher abundance of longer carbon chains and vice versa.
For our chains inside DWCNT a quasilinear relation between the inverse chain length and the red shift of the Raman frequency of the LCC-Band is observed. This is in strong contrast to chains in vacuum [ 19,26 ] where a clear saturation of the Raman frequency emerges when approaching the infi nite chain limit. This shows that the environment of the DWCNT and especially the charge transfer between the inner tube and the linear carbon chains plays an important role. Recent ab initio calculations by Wanko et al. confi rmed that, in agreement with the experimental observation, this charge transfer has linear relation on the red shift of the Raman frequency and becomes important for confi ned LCCs. [ 20 ] It is consistent with the one between inner and outer tubes in a pristine DWCNT. [ 8 ] This points out that the charge transfer to the inner tubes, which could be the main reason for the low PL response of DWCNT inner tubes, might be balanced by a charge transfer to the carbon chain in this LCC@DWCNT hybrid system.
In Figure 1 a the Raman response of the G-Band and the LCC-Band after excitation at 2.18 eV (568.2 nm) is plotted with all signals being normalized to the intensity of the G-Band. Interestingly, the shape of the latter is only altered considerably when the annealing temperature of 1460 °C is reached, where two extra peaks appear as shoulders of the main peak that slightly diminish at even higher growth temperatures. The appearance of the G-Line resembles the one found by Andrade et al. for MWCNT with encapsulated carbon chains after a high pressure treatment. [ 27 ] In general, the frequency of graphitelike modes of pristine (not fi lled) nanotubes is pressure dependent as the main peak can either shift up (compression) [ 28 ] or down (strain). [ 29 ] By performing temperature dependent Raman measurements, we found that the LCC and the inner tube are coupled by van der Waals interactions. [ 19 ] In the case for the G-Band, the LCC may locally induce a strain in the inner tube, which then causes a compression in other parts of the same nanotube, enabling the observation of both effects. Also, the height of the two extra shoulders diminishes with increasing temperature, concomitant with the decomposition of the carbon chains, which we discuss later. For the LCC mode itself, several factors have to be taken into account: The utilized excitation energy of 2.18 eV (568.2 nm) allows to detect LCC of different lengths very effi ciently as it is close to the resonance energy of long (>60 carbon atoms) [ 19,30 ] and short (down to 10 carbon atoms) LCC. [ 31 ] By assuming that the amount of material available for carbon chain synthesis compared to the nanotube abundance is similar for all samples, the intensity of the different chain length contributions to the overall Raman response solely depends on the number of LCC being most in resonance with the incident laser. The highest intensity was always found for a frequency of 1852 cm −1 , indicating that LCC containing ≈40 carbon atoms are most in resonance with the laser.
The total LCC-Band is composed of signals from many differently long chains, since the varying diameters and chiralities of inner tubes provide different growth conditions for the LCC. Although it is not possible within the given spectrometer resolution to directly assign a certain chain length (i.e., Raman frequency) to only one specifi c (n,m) value, measurements on the radial breathing modes of the inner tubes revealed that only within small diameter tubes the longest chains can be found, with the optimum diameter ≈0.71 nm. [ 19 ] Therefore, we can safely say that the closer the inner tube's diameter is to the evaluated optimum for LCC encapsulation, the longer the LCC can become. Eventually, for ease of comparison, the LCC-Band was simply divided into two parts at the position of the main peak at 1852 cm −1 . In Figure 1 b the integrated area coverage of the low frequency contribution (representing long chains) to the overall signal as a function of the LCC growth temperature is plotted.
With this background, we propose the following model: Up to a growth temperature of about 1400 °C, only relatively short carbon chains with a length of up to 40 carbon atoms are grown. When the growth temperature is raised, smaller chains coalesce to form longer and longer chains, with the highest yield of long LCC found at a temperature of 1460 °C. In Figure 1 a we can see clearly that the LCC-Band is composed of several contributions, as two extra peaks at 1845 and 1830 cm −1 appear. When increasing the growth temperature, the 1830 cm −1 peak is still visible but the 1845 cm −1 peak disappears, while the main peak reaches a maximum in the intensity. This is a sign that a decomposition of this type of carbon chains starts at this temperature, ending up as short chains whose Raman response is found at higher frequencies. Again comparing to the measurements on the radial breathing mode and the ideal nanotube diameter for LCC growth, [ 19 ] this decomposition can be traced to inner tubes owing a diameter deviating too much from the ideal value. Therefore, they are not providing enough stability for the chains, in contrast to their thinner counterparts that provide more stability to the chains and are capable of containing and protecting the longest LCC. Eventually, by further raising the growth temperature, the aforementioned trend continues.

Optical Absorption of the Differently Treated DWCNT Samples
After the Raman measurements all samples were solubilized to elucidate on their optical properties. The parameters of the procedure on how homogeneous DWCNT suspensions are received are of uttermost importance for the absorption and the photoluminescence of DWCNT. The applied technique follows a method which has proven that inner tube extraction can be avoided; [ 14 ] the exact details can be found in the Experimental Section. Figure 2 a shows the absorption spectra of a pristine, not-fi lled DWCNT sample in comparison with HiPco SWCNT before background subtraction as reported earlier, [ 14 ] (n,m) assignment was performed based on the work of Weisman and Bachilo. [ 32 ] Features of the excitonic transitions between corresponding van Hove singularities of DWCNT are strongly attenuated in comparison to SWCNT. [ 12,14,33,34 ] On the one hand, a decrease in absorption features is often a result of carrier/hole doping to the carbon nanotube as observed for SWCNT. [35][36][37][38] As mentioned earlier, Zólyomi et al. found that a charge transfer from outer to inner tube is taking place. [ 8 ] On the other hand, the very nature of absorption measurements has to be considered, as in ensemble measurements contributions from different inner and outer tubes can be observed. By using an ozone etching technique, Iakoubovskii et al. could deconvolute the signals and showed for their samples that the signals of E 11 transitions of inner tubes and E 22 transitions of outer tubes were overlapping, "blurring" the inner tube signal. [ 5 ] A similar observation of overlapping inner and outer tube contributions was also reported by Miyata et al. who extracted the inner tubes. [ 33 ] Since the diameter distribution for their samples was similar as it is in this work, such overlaps can also be expected. Figure 2 b shows the normalized (normalization procedure described in the Experimental Section) absorption spectra for E 11 transitions of the pristine, not-fi lled DWCNT sample and two differently fi lled DWCNT samples: the DWCNT annealed at 1460 °C that inhibit the highest number of the longest chains and the DWCNT annealed at 1400 °C, as these showed the greatest change in absorption features compared to the pristine sample. For all fi lled DWCNT, the size and shape of the inner tube's E 11 peaks are altered compared to the pristine sample. For example, the signal of the (7,5) species is much more prominent for all samples, also the (8,3) and (6,5). Another interesting feature is the peak around 1120 nm that represents a superposition of E 11 peaks for the (8,4) (d t = 0.84 nm), (7,6) (d t = 0.89 nm), and (9,4) (d t = 0.91 nm) species. The shape of this peak changes for the annealed samples compared to the pristine sample. This is a strong hint that the excitonic transitions for these tubes are affected differently by the encapsulated carbon chains which can be related to the growth capabilities for LCC given by the nanotube diameter. [ 19 ] Interestingly, as already mentioned, the strongest enhancement of absorption features is observed for the sample annealed at 1400 °C. On the one hand, we found the charge transfer from the inner tubes to the carbon chains to be strongest for the longest chains, causing the Raman frequency to decline linearly with the inverse number of carbon atoms instead of saturating at a certain value. [ 20 ] On the other hand, the charge transfer already starts causing this deviation in the Raman frequency from the vacuum case for chains containing around 20 carbon atoms. Therefore, the stronger enhancement in absorption features caused by shorter chains rather than the longest ones bears a striking resemblance to the enhancement of absorption features (and photoluminescence intensity) found for SWCNT fi lled with ferrocene: [ 18 ] There, the best enhancement was attributed to an optimum value for the n-type doping of the SWCNT by the ferrocene molecule, cancelling out the p-type doping from the opening process and the surfactant molecules. In our case, an optimum value for the charge transfer from the inner tube to the carbon chain given by the chain length has a similar effect, nullifying the outer tubes charge transfer to the inner tube. To investigate the infl uence of the LCCs grown at different temperatures on each of the different species without the overshadowing outer tube contribution, photoluminescence spectroscopy was applied subsequently to the absorption measurements.

PL of Inner Tubes Filled with LCC
When investigating on the PL of inner tubes fi lled with LCC, several factors have to be considered. Summarizing from our recent reports, [ 19,20 ] the interaction of a LCC with the inner tube depends on the following parameters: • Inner tube diameter : For LCCs with similar length, the interaction with the nanotube depends on the distance between chain and nanotube wall, i.e., the nanotube's diameter. • The length of the carbon chain : The longer the LCC, the stronger the charge transfer from inner tube to LCC. • Growth conditions : The eventual length that the LCC can reach is given by the inner tube diameter (with d t ≈ 0.71 nm being ideal for the longest chains) and the growth temperature. 1460 °C gives the longest chains, surpassing this temperature destabilizes the chains, as shown earlier.
For an evaluation of the luminescence behavior of DWCNT fi lled with linear carbon chains several different species for each sample were measured with an excitation wavelength corresponding to their E 22 resonance. Figure 3 shows three examples of the numerous PL spectra taken for pristine DWCNT and again the sample annealed at 1400 °C with the biggest change in absorption features and 1460 °C representing DWCNT fi lled with the longest LCC. The fi rst line scans in Figure 3 a show the change in the PL signal of (6,5) tubes with a diameter d t = 0.76 nm. The intensity of this chirality is amplifi ed greatly for the chain fi lled samples compared to pristine inner tubes, although, in contrast to the absorption spectra, the chains grown at 1460 °C even give a slightly higher intensity. From the intensity of the side peaks, i.e., tubes that are not in resonance with the incident laser wavelength, this seems to also hold true for the even smaller (6,4) tubes (d t = 0.69 nm) but not for the bigger (7,5) or (8,4). This deviating behavior becomes apparent when other species are measured at their E 22 resonances. For example, for the (8,3) inner tubes with d t = 0.78 nm in Figure 3 b the sample annealed at 1400 °C containing more relatively short chains gives the highest intensity enhancement, outperforming their long chain counterpart greatly. Although not in resonance, the same statement can already be made here for the (7,5) and (7,6) tubes Adv. Funct. Mater. 2016, 26, 4874-4881 www.afm-journal.de www.MaterialsViews.com Figure 2. a) Absorption spectra of the pristine DWCNT sample as compared to a HiPco SWCNT reference sample. [ 14 ] Assignment based on the work of Weisman and Bachilo. [ 32 ] b) Comparison of the normalized E 11 absorption spectra from pristine DWCNT and the samples annealed at 1400 and 1460 °C. Similar as for the ferrocene example mentioned in the text, the E 11 of several species is enhanced by the presence of LCC.
(diameters d t = 0.83 nm and d t = 0.89 nm, respectively). Furthermore, as depicted in Figure 3 c, the (10,2), d t = 0.88 nm, behave similarly. In order to quantitatively analyze the PL response, a line shape analysis for all spectra was performed to evaluate the correct PL signal of the species being in resonance. By doing so, we are able to receive the right correlation between PL enhancement, diameter dependence, and annealing temperature, i.e., growth yield of the LCC.
Measuring in total nine different species for all samples at their respective E 22 resonance wavelength and comparing their intensity to the pristine sample gives two types of correlations: First, the PL amplifi cation as function of the inner tube diameter is plotted in Figure 4 for three representative samples. The amplifi cation reaches a maximum for the (8,3) species with a diameter of 0.78 nm for all samples annealed at temperatures ≤1480 °C (here only 1400 °C, red curve, and 1460 °C, green curve, are shown for clarity), for the 1500 °C (blue) and 1530 °C sample (not shown) the (6,5) species exhibits a maximum in the amplifi cation. The appearance of a diameter-dependent optimum in the PL amplifi cation is again similar to the PL enhancement of SWCNT fi lled with ferrocene as observed by Liu et al. (black curve), [ 18 ] although, since the ferrocene molecule demands larger tubes to be encapsulated in, this trend can only be observed for tubes with diameters ≥0.9 nm. In the case of ferrocene, the size of the molecule is fi xed and the charge transfer to the nanotube depends solely on the tube's diameter. Considering the aforementioned parameters determining the inner tube-LCC interaction [ 19,20 ] and the DWCNT's intrinsic charge transfer from outer tube to inner tube, [ 7,8 ] an optimum between chain length, inner tube diameter and inner/outer tube combination must exist so that the infl uence of the outer tube to the inner tube is ideally compensated. For our DWCNT distribution the optimum is apparently reached for LCC grown at 1400 °C. The overall decrease in PL amplifi cation when increasing the growth temperature from 1400 to 1460 °C (i.e., receiving longer chains) is then caused by an overcompensation of the n-type doping of the inner tube by the outer tube. Consequently, the PL of the inner tubes becomes stronger Adv. Funct. Mater. 2016, 26, 4874-4881 www.afm-journal.de www.MaterialsViews.com Figure 3. Photoluminescence line scans of different inner tube species. Spectra were normalized according to their optical density and to the selected species being in resonance for pristine DWCNT so that I Pristine 1, (n,m) assignment based on the work of Weisman and Bachilo. [ 32 ] PL line scan at a) 569 nm, the E 22 resonance of (6,5) tubes, b) 670 nm, the E 22 resonance of (8,3) tubes and, c) 742 nm, the E 22 resonance of (10,2) tubes. The appearance of a diameter-dependent maximum is similar to the observation of PL amplifi cation found for SWCNT fi lled with ferrocene, [ 18 ] black curve. again when the growth temperature is surpassing the point where the longest chains are stable (>1460 °C): As the thermal energy is increased, tubes with a diameter deviating from the ideal value (≈0.71 nm [ 19 ] ) are less and less suited for LCC containment since the energy gain for the LCC provided by the nanotube environment is reduced. The optimum diameter for PL enhancement then shifts toward the ideal diameter for LCC encapsulation and therefore to smaller inner tubes when the growth temperature is increased, as seen for the tubes annealed at 1500 °C.
The next correlation that has been observed is by fi rst plotting the PL amplifi cation as function of the growth temperature which is depicted in Figure 5 . There, the curves can be separated in three different groups according to their diameter and temperature dependence: "Small" inner tubes like the (6,4) and (6,5) tubes with diameters ≤0.76 nm; "medium"-sized inner tubes with diameters ranging from 0.78 to 0.89 nm and "large" inner tubes with diameters >0.9 nm. Additionally, in the lower graph the result of the Raman analysis on the chain length distribution obtained in the previous section is plotted; important to note for this context is that the chain signal observed by Raman spectroscopy originates from the whole sample, i.e., LCC incorporated in different species are detected simultaneously, in contrast to the chirality sensitive PL measurements. Eventually, this enables us to evaluate the evolution of the PL intensity for each measured inner tube species with the length of the chain.
Starting with the smallest inner tubes, these species show interesting evolutions in the PL signal in terms of stability. The PL signal for the (6,4) species stays almost constant over the whole temperature range. Their diameter d t = 0.69 nm is very close to the reported optimum diameter for LCC encapsulation (≈0.71 nm), indicating that the (6,4) provide a very stable environment for the longest chains. This is substantiated by Raman measurements depicted in Figure 1 a, as the peak observed at 1830 cm −1 does not vanish even at 1530 °C. The (6,5) species (d t = 0.76 nm) start off similarly, as the PL enhancement is constant up to a growth temperature of 1460 °C. As their diameter deviates from the ideal value, increasing the temperature leads to instability of the longest chains. This in return results in the fabrication of shorter chains that are more suited for PL amplifi cation as mentioned before, so the PL intensity goes up. Eventually, at even higher growth temperatures such as 1530 °C, these chains also get unstable inside the (6,5) so the PL decreases again.
The medium sized tubes (0.78 nm ≤ d t ≤ 0.89 nm) show an unanimous behavior over the whole temperature range. Apparently, these types of tubes are capable of growing LCC with similar length but also, since the highest amplifi cations observed are found in this range, that the charge transfer established by the presence of the LCC reaches an optimum level for chains containing between 20 and 40 carbon atoms. As the chains grow longer when the growth temperature is increased from 1400 up to 1460 °C, the resulting charge transfer exceeds the optimum value and the PL intensity decreases. By surpassing 1460 °C, the chains become unstable and start breaking up, i.e., the charge transfer goes back in the direction of the optimum value. At growth temperatures higher than 1480 °C, this point has already been passed for this type of tubes and the PL intensity decreases monotonically. Furthermore, we can also see an overall decrease in the PL intensity when the diameter is increased from 0.78 to 0.89 nm which can be related to the decreasing interaction energy between chains and inner tubes. [ 19 ] For inner tubes with a large diameter (d t > 0.91 nm), the amplifi cation is already very weak. The fi rst maximum is only reached at 1430 °C which is related to the LCC formation process: As the diffusion of the carbon atoms is not restricted to a 1D path any more but small radial movements are also possible, the diffusion length along the nanotubes axis is reduced. Therefore, higher thermal energy is needed. After trespassing 1430 °C the progression follows the PL response of the medium sized tubes.

Conclusions
In summary, we present an effi cient and simple method to increase the low PL quantum yield of inner tubes from DWCNT. Double-walled carbon nanotubes were fi lled with differently long linear carbon chains and were subsequently investigated by resonant Raman, optical absorption, and photoluminescence spectroscopy. The optical properties of the inner tubes are altered greatly by the presence of the LCC. Depending on the length of the chain and the inner tube's diameter, the PL signal is increased up to a factor of 6 for tubes with a diameter close to ≈0.8 nm, while hosting carbon chains containing between 20 and 40 carbon atoms. Concomitant to previous studies, this enhancement can be assigned to a local charge transfer from the inner tube to the carbon chain, balancing out the DWCNTs intrinsic charge transfer from outer tube to inner tube. The method presented here demonstrates a pathway to untether Adv. Funct. Mater. 2016, 26, 4874-4881 www.afm-journal.de www.MaterialsViews.com Figure 5. Amplifi cation of the PL intensity as function of the annealing temperature. For clarity, the species are separated in three groups according to their diameter and set in comparison to the Raman result as shown earlier. Small diameter tubes like the (6,4) and (6,5) are closest to the ideal diameter suitable for LCC fi lling (≈0.71 nm), whereas medium sized tubes mostly contain shorter chains. Inside large tubes, LCC formation is hindered due to the lower stability given by the nanotube and therefore the amplifi cation is very low. the inner tube's optical response from its protective outer tube, leading to carbon nanotubes with optical properties much more resilient against environmental infl uences than singlewalled carbon nanotubes, which are crucial for their eventual applications, e.g., as biosensors. Ultimately, resolving the longstanding question, this selective enhancement serves as a fi nal proof for their capability to exhibit photoluminescence.

Experimental Section
Synthesis and Growth of Linear Carbon Chains : DWCNT were synthesized with the method of high-vacuum chemical vapor decomposition (HV-CVD) using identical parameters and purifi cation steps as reported earlier for the formation of LCC. [ 19 ] The carbon source for HV-CVD growth was ethanol, the growth temperature was 875 °C for 10 min. For purifi cation, the DWCNT samples were fi rst exposed to hydrochloric acid to remove remaining catalysts followed by annealing in air fl ow at 400 °C for 2 h to remove amorphous carbon and after another acid treatment annealed at 500 °C in air for 2 h to remove SWCNT impurities (pristine sample). The purifi ed DWCNT powders were annealed after synthesis in high vacuum (<10 −6 mbar) at different temperatures between 1370 and 1530 °C for 45 min to grow linear carbon chains with different growth yield inside the inner tubes of the DWCNT.
Characterization by Raman Spectroscopy : Raman spectroscopy was used in this report to characterize the annealed DWCNT samples to evaluate the LCC length distribution, using a multifrequency Raman microscopy system (Horiba LabRAM HR) in backscattering geometry. A laser from an external source (Coherent Innova 70c) with a wavelength of 568.2 nm was coupled into the spectrometer. The samples themselves were placed in a thermalized cryostat to ensure a constant (room) temperature and the laser power was 200 µW to avoid sample heating so the shape and size of the carbon chain signal solely depends on the postsynthesis annealing temperature. Additionally, Raman spectroscopy was also used to determine the inner tube diameter distribution within the sample by exciting the sample at 488.0, 530.9, 568.2, and 647.1 nm. For optimized conditions (i.e., the growth of DWCNT matching the nearinfrared PL window) we fi nd a diameter distribution of the DWCNT of 1.45 ± 0.25 nm and a corresponding inner tube diameter between 0.6 and 1.1 nm.
The estimation of the bulk fi lling ratio for this kind of samples is based on a combination of X-ray diffraction (XRD), Raman and electron spectroscopy, as reported earlier. [ 19 ] The determination of the bulk fi lling ratio from XRD relies on a form factor analysis applied for C 60 peapods [ 39 ] and ferrocene molecules inside SWCNT. [ 40 ] The relative intensity of the diffraction peak between the molecules and the SWCNT diffraction peaks allows an accurate determination of the bulk fi lling ratio. Adopting this method for LCC, the relative intensity of the sp 1 hybridized chain peak compared to the intensity of the sp 2 hybridized carbon peak of DWCNT allows us to pin down the bulk fi lling ratio of LCC inside DWCNT, complementary to the relative yield determined from the intensities in the resonance Raman response (see Supplementary Information in ref. [ 19 ] ). The optimal growth window for bulk fi lling with LCC was observed for inner tube diameters between about 0.6 and 0.9 nm. [ 19 ] We observed that more than 90% of those DWCNT are fi lled with LCC.
Individualization of Double-Walled Carbon Nanotubes : To obtain wellisolated DWCNT without extracting their inner tubes, a method similar to Procedure A given by Rohringer et al. [ 14 ] was used. The DWCNT sample was individualized via ultrasonication in deionized water with 2% w/v sodium deoxycholate (DOC, Sigma-Aldrich), using a steel tip with a diameter of 1/4 in, 60 W of power, for 4 h (Branson Sonifi er 450). The sample was subsequently centrifuged at 10.000 x g for 30 min (Thermal Scientifi c) to remove metallic particles, a speed low enough to ensure inner tube containment. After that, the supernatant (top 50% of the centrifuged solution) rich with isolated DWCNT was collected in a quartz cuvette for further investigation.

Characterization by Optical Absorption and Photoluminescence
Spectroscopy : Optical absorption spectroscopy was performed by using a Bruker VERTEX 80v Fourier transform spectrometer with the resolution set to 2 cm −1 . For the PL measurements, different external lasers were coupled into a NanoLog spectrometer (Horiba Jobin Yvon): A tunable dye laser equipped either with Rhodamine 6G (565 nm < λ < 605 nm) or DCM (640 nm < λ < 680 nm), or a tunable titanium sapphire laser (λ > 695 nm). All measurements were done at room temperature. Normalization of the different suspensions with respect to their optical density was achieved by taking the minimum between the fi rst and second optical transition of the inner tubes. [ 14,18 ]