Carbon nanotube based biomedical agents for heating, temperature sensoring and drug delivery

Abstract

Due to their extraordinary physical and chemical properties carbon nanotubes reveal promising potential as biomedical agents for heating, temperature sensoring and drug delivery on the cellular level. Filling carbon nanotubes with tailored materials realises nanoscaled containers in which the active content is encapsulated by a protecting carbon shell. We describe different synthesis routes and show the structural and magnetic properties of carbon nanotubes. In particular, the filling with magnetic materials offers the potential for hyperthermia applications while the insertion of NMR active substances allows the usage as markers and sensors. The potential of carbon nanotubes for biomedical applications is highlighted by hyperthermia studies which prove their applicability for local in situ heating. In addition we have shown that a non-invasive temperature control by virtue of a carbon-wrapped nanoscaled thermometer and filling with anti-cancer drugs is possible.

• Carbon nanotubes
• hyperthermia
• magnetic nanoparticles
• temperature control
• NMR

Introduction

Strong adverse effects on the healthy tissue in the vicinity of a tumour are a major drawback in current cancer therapies. One innovative technological approach to solve this problem focuses on therapies on the cellular level by applying intracellular probes, i.e. the transfer of nano-sized biocompatible devices into the cells. Most promising materials in this regard are magnetic nanoparticles which can be addressed by external magnetic fields. A great advantage of magnetic fields is their biocompatibility, i.e. they penetrate tissue non-invasively and without known adverse effects, and their weak interaction with organic matter so that deep layers of (human) tissue can be reached. In particular, magnetic nanoparticles can be localised in deep tissue, external static magnetic fields can fix them at a precise position, gradient fields can move them and alternating (AC) fields lead to local heating. The latter can be utilised for hyperthermia applications, shown, for example, by the use of superparamagnetic nanoparticles for magnetic fluid hyperthermia [1]. Much of the current research is focused on iron oxide nanoparticles which have proven their feasibility in animal experiments [2], [3] and are now under clinical trials [4]. Other magnets usually present limitations due to biocompatibility issues, since most of them contain potentially toxic elements.

Due to its particular magnetic properties, i.e. large anisotropy and saturation magnetisation, metallic iron could generate heat more efficiently in comparison to iron oxides since various mechanisms yield dissipative effects such as domain wall motion, etc. [5]. In practice, higher heating efficiency means that less nanoscaled material would have to be introduced into the biological system in order to achieve the targeted hyperthermia effect. The use of nanoparticles made of iron, however, is hindered by the fact that oxidation in ambient or biological conditions has to be avoided. A promising way to overcome this problem appears to be the coating of the iron with a carbon shell by insertion of the material inside of carbon nanotubes (CNTs) and thereby protecting the biological environment and the filling material against each other. Degradation of filling materials is avoided and their potential toxicity and adverse effects are suppressed so that CNTs provide a smart carrier system on the nanometer scale.

A targeted hyperthermia therapy using iron filled CNTs and, in general, utilisation of tailored carbon nanotube based biomedical agents for therapeutic and diagnostic purposes can therefore be envisioned which will be detailed in the following sections. After introducing some basic properties of CNTs, a short review of biofunctionalisation, biocompatibility and toxicity aspects is provided. The synthesis of iron filled CNTs as well as magnetic and hyperthermia studies are discussed. Adding additional material such as a thermometer in Fe filled CNTs increases the potential of CNTs as hyperthermia inducing agents since simultaneous local heating and temperature control might be feasible. Moreover, the container feature of CNTs offers the step beyond the controlled heat treatment but also suggests the additional local release of therapeutics.

Biocompatibility and toxicity

CNTs are hollow carbon structures with one or more walls, a small diameter on the nanometre scale and a large length in comparison. Their mechanically and chemically stable carbon shells can be opened, filled and closed again without losing their stability. Experiences in filling CNTs range back to their discovery in 1991 [6]. Since then, extensive work has been performed to synthesise CNTs and to functionalise them both exohedrally, i.e. by attaching functional elements to the outer shell, and endohedrally by filling with various materials. CNTs can be filled with metals, semiconductors, salts, organic materials, fullerenes, etc., either during the synthesis process or through subsequent opening, filling, and closing of the CNTs. In particular, ferromagnetic materials such as Fe, Ni, and Co can be encapsulated which is relevant to hyperthermia applications. Introducing fluorescent markers offers a route for visualising CNTs [7], which is also possible by exohedral functionalisation [8], [9]. The container feature of CNTs allows, in principle, simultaneous filling of CNTs with different materials thereby combining multiple functionalities in one kind of carrier (Figure 1). In this way CNTs provide a smart carrier system on the nanometre scale which can be filled with tailored materials to address specific purposes. The container function is underlined by recent model calculations for encapsulation of the anticancer drug cisplatin. Interestingly, all three orientations of cisplatin might be transferred into a CNT if its minimum radius is at least 4.785 Å while the maximum uptake occurs for radii of approximately 5.3 Å. These results therefore suggest that multi-walled CNTs with a relatively large inner diameter are particularly appropriate for drug delivery of common therapeutics like cisplatin or carboplatin [10].

Figure 1. Sketch of a filled carbon nanotube serving as multi-functional container for in vivo applications. A ferromagnet can induce heat in AC magnetic fields and a material with strongly temperature dependent nuclear magnetic resonance (NMR) signal might serve as a thermometer. Additional drug delivery can be envisaged. Exohedral functionalisation to achieve biocompatibility is sketched. Note that single-walled as well as multi-walled CNT can be realised.

Beyond the shielding effect against a biological environment, the carbon coating offers an interface for further exohedral functionalisation with suitable (bio-) molecules. A major task of such functionalisation is the stable dispersion of the nanoparticles in aqueous media which still is a major challenge. Both non-covalent and covalent modification strategies can be applied among which the former preserve the pristine CNT structure while covalent modification introduces partial damage of the outer wall but in general yields better dispersion. Dispersion becomes even more crucial if ferromagnetically filled CNTs are envisaged which exhibit an increased tendency to agglomerate due to magnetic interactions. In addition, exohedral functionalisation is needed to get the highly symmetric carbon structures compatible to actual biological environments. Nowadays, also functionalisation aiming to perform dynamic tasks such as target recognition, target transformation, transport, or electrical conduction in the living cell is addressed. These functions can be provided by biomolecules, like DNA and enzyme proteins. Importantly for any therapeutical or biomedical usage, functionalised CNTs can effectively cross biological barriers such as the cell membrane and penetrate the individual cell. Recently, it was shown that DNA wrapped SWNTs were enveloped by cancer cells and they were used to deliver a lethal dose of microwave radiation to the cancer cells [11]. In contrast to such wrapping of CNTs with biomolecules, any chemical exohedral functionalisation needed to improve the biocompatibility of CNTs will structurally perturbate the external wall which, however, in the case of multi-walled CNTs does not affect the overall carbon shielding. A variety of methods exist for the exohedral functionalisation of carbon nanotubes [12], [13], some of which have been successfully applied for the conjugation of proteins, drugs and fluorescent dyes [14], [15], [16], [17] and active tumour targeting in vivo [18]. Successful biofunctionalisation is demonstrated for example by Pantarotto et al. who observed that functionalised CNTs can cross the cell membrane and accumulate in the cytoplasm or reach the nucleus without being toxic for the cell [15]. Soluble CNTs can be coupled with amino acids and bioactive peptides [19] for further derivatisation. It was also shown that stable complex formation of CNTs and cationic lipid may accelerate the delivery of nanotubes into the bladder cancer human cells [20]. For the actual mechanism of internalisation, i.e. endocytosis or phagocytosis, which has been discussed contradictorily, the actual morphology of the CNTs seems to be crucial [14], [15], [21].

Despite various industrial applications of CNTs and their large scale synthesis only little is known about the interaction of CNTs with biological environments. CNTs and other nanomaterials when placed in contact with human (or other mammalian) body fluids and tissues are recognised by the immune system complement proteins, and may also interact with other systems, such as coagulation proteins and cell-surface proteins. Binding of complement proteins activates the complement system via both classical and alternative pathways and results in strong binding of several complement proteins to the CNTs [22]. Such complement activation influences subsequent interaction of the CNTs with cells and tissues and is a predictor of potential toxicity in animal models. Salvador-Morales et al. report that such protein binding to CNTs is highly selective. In particular, fibrinogen and apolipoproteins were the proteins that bound to CNTs in greatest quantity. Among proteins contained in lung surfactant, SP-A and SP-D selectively bind to CNTs so that chronic level exposure may result in sequestration of these proteins [23].

Although a variety of toxicity studies has been published (for recent reviews see for example [24], [25]), no clear picture evolves for CNTs in general. This is, for example, illustrated by contrasting in vitro studies showing CNTs to be safe [26], [27] or to induce significant toxic effects [28], [29], [30]. One reason for this ambiguity is the fact that–similar to other nanoparticles–a large number of specific factors govern the toxicity of CNTs such as their shape and size (diameter, length), the number of shells (single- or multi-walled CNTs), agglomeration state and surface chemistry [31]. In particular, the concentration of CNTs is not necessarily a main parameter. On the other hand, ambiguity also results from a lack of standardisation and of thorough characterisation of, for example, defects of the outer shell or potentially toxic contaminants. Such non-carbon material originating from the synthesis process of pristine empty CNTs might amount to 5–10% of the total mass. This presumably accounts for much of the reported differences so that details of synthesis, choice of catalyst particles, washing procedures and dispersion methods have to be considered thoroughly.

In the following we concentrate on multi-walled Fe filled CNTs for which synthesis and properties are described in §3 and §4. We recall the feasibility of an efficient delivery of Fe filled CNTs into human cancer cell lines which has been shown in cell culture experiments [20]. A pre-treatment of Fe filled CNTs with cationic lipid was found to cause a qualitative delivery of the complexes into the cytoplasm but not into the nucleus. A recent study on cytotoxic effects of Fe filled CNTs in vitro addressed metabolic activity, cell proliferation, apoptosis and cell cycle distribution of a malignant (PC-3) and a non-malignant (fibroblasts) cell lineage. The data imply that CNTs strongly associate with cells within a short incubation period. The presence of CNTs in cells did not pose any significant toxic effect [32]. This observation is corroborated by an animal study on mice which indicated no remarkable toxic or adverse effects over a period of 440 days after the intraperitoneal or intravenous injection of pristine Fe filled CNTs. In particular, based on TEM and histological analyses after 6 weeks Mönch et al. [33] report the absence of any indication on inflammation. In one case, CNTs have been administered up to a total of >1 g Fe filled MWCNTs/kg body weight over a period of 3 months. Interestingly, in agreement with the fact that pristine CNTs were used, large agglomerates have been observed in various organs in case of the intravenous treatment. However, such agglomerates seem to have no drastic effects. All animals survived for more than one year and showed no abnormalities in behaviour or weight suggesting a general biocompatibility of the CNTs for the applied doses.

The long-term absence of significant toxicity effects in the mentioned animal study also implies the expected insignificant degradation of CNTs in the biological environment. Regarding their fate in vivo, excretion in faeces and urine has been reported after injection of empty CNTs intravenously or intraperitoneally in mice [34], [35]. Studies on the biodistribution and translocation pathways of empty CNTs in mice, however, indicate accumulation predominately in liver and retain for long time while low acute toxicity is confirmed [36].

Synthesis of ferromagnetically filled multi-walled carbon nanotubes

Encapsulation of iron nanowires in CNTs realises highly anisotropic ferromagnetic nanoparticles which are discussed for a wide range of potential applications [37]. Among a variety of methods to synthesise these filled carbon structures such as arc discharge [38] and laser ablation [39], the chemical vapour deposition (CVD) is applied when a high yield uniform multi-walled CNTs is aimed for [40].

A suitable method for synthesising CNTs filled with metals as Fe, Co or Ni is the so-called ‘in situ method’, a special CVD method, in which the formation of CNTs and their filling with ferromagnetic elements or compounds take place simultaneously. Here, relevant precursors are needed, which for example provide iron for the filling as well as carbon for the shells. The most common precursors are metallocenes [Me(C₅H₅)₂; Me = Fe, Co, Ni] which supply a large scale of well-defined multi-walled (MW) CNTs with a high filling yield (about 50%) of the ferromagnetic material. Many groups world-wide use this method in different equipments. Sen et al. reported the first experimental results for the synthesis of ferromagnetic MWCNTs [41]. In a typical experiment the ferromagnetically filled MWCNTs are grown in a quartz tube reactor inside a dual zone furnace system. The main parameters of the deposition process are the sublimation temperature of the metallocene in the first furnace, the gas flow rate, and the temperature of the second hot furnace zone where the pyrolysis of metallocene and the deposition of filled tubes take place. CNTs are formed on oxidised silicon substrates uncoated or coated (2 nm layer of different materials such as Fe, Co and permalloy) placed inside the reaction zone. In this case they grow in perpendicular alignment to the substrate surface. Many more nanotubes are formed on the wall of the quartz tube reactor (which is a mixture of bundles of well-aligned Fe filled MWCNTs) [42], [43], [44], [45].

An additional technique is the liquid source CVD (LS-CVD) [40]. This method is characterised by a constant and reproducible transport of the precursor. The metallocene, especially the ferrocene, is dissolved in cyclopentane and dropped on the moving band continuously. In the first part of the system the solvent is vaporised and only the ferrocene is transferred into the reactor at a defined temperature and with a constant transport velocity. In the deposition reactor a second moving band, populated with precoated substrates is positioned. This LS-CVD method is highly suited for the continuous production of defined ferromagnetically filled CNTs. Wang et al. have synthesised Fe filled CNTs with a high filling ratio by using ferrocene and dichlorbenzene as precursor [46]. This solution is injected through a nozzle directly into the reactor so that a spontaneous decomposition reaction occurs. Pichot et al. applied an aerosol assisted CVD (AA-CVD) [47] device to produce Fe filled CNTs. Ultrasonicating a solution of ferrocene in cyclohexane produced an aerosol, which was then carried for 15 min by an argon flow through a quartz reactor placed in a tubular furnace at 850°C. Eventually, MWCNTs carpets were deposited on the reactor walls [48].

Filled CNTs can be also synthesised by a post-synthesis method. This technique includes (1) the synthesis of empty CNTs, (2) the opening of their ends, (3) a filling step with various materials (metals, salts, therapeutics) and (4) the closing step of the filled CNTs. The opening procedure is mostly done by well established wet chemistry techniques [49] or by oxidation in air [50], [51]. The easiest way to incorporate the material into the open ended CNTs is over a gas-phase reaction where the CNTs and the filling material are inserted together into glass ampoules sealed under vacuum and heated beyond the sublimation temperature of the filling material [52], [53]. But also a wet chemical approach can be applied where capillarity is the driving force [54], [55]. The closure of the open-ended and filled CNTs can be realised by a redeposition with a polymer or other carbon-containing phases [56]. Various techniques such as transmission electron microscopy and X-ray diffraction demonstrate the successful synthesis of Fe filled CNTs by our CVD technique [40], [42], [43], [57], [58], [59], [60]. For example, the cross section shown in Figure 2 clearly indicates the single crystalline Fe filling as well as the carbon shell.

Figure 2. Typical SEM and TEM images of well-aligned iron filled multi-walled carbon nanotubes synthesised by LS-CVD.

The length of CNTs is in the micrometer range and outer diameters vary from 20–80 nm. The Fe filling is discontinuous and consists of iron nanowires with a filling yield of ∼50%. Interestingly, by X-ray diffraction the existence of the body-centred structure (α-Fe), the face-centred cubic structure (γ-Fe) and in very low concentration also of Fe₃C could be verified. For a hyperthermia application a high yield of the ferromagnetic α-Fe is requested. By changing details of the synthesis process the relative yields of the different Fe phases can be adjusted. The rate of the ferromagnetic α-Fe is increased by an annealing process below the temperature of the phase transition of γ-Fe to α-Fe which indeed yields a larger saturation magnetisation [59]. To get exact information about the morphology of the iron filled CNTs (γ-Fe/α-Fe ratio), the ⁵⁷Fe Mössbauer spectroscopy is an appropriate method. Transmission Mössbauer spectroscopy (TMS) and backscattered conversion electron Mössbauer spectroscopy (CEMS) were applied in order to distinguish different Fe phases and their spatial distribution within the whole sample and along the tubes’ height together with Ruskov's group [61]. A characterisation (on a large spatial scale) of the aligned CNT samples was performed by obtaining TMS spectra for selected spots positioned at different locations of the sample. While the total Fe content changes considerably from one location to another, the γ-Fe/α-Fe phase ratio is constant onto a relatively large area. Using TMS and CEMS for all aligned Fe MWCNTs samples it is also shown that along the CNT axes, going to the top of the nanotube the relative content of the γ-Fe phase increases. Going to the opposite direction, i.e. towards the silicon substrate, the relative content of the Fe₃C phase increases [61].

Magnetic properties and hyperthermia feasibility

Various studies have shown that, independently of the synthesis technique, the carbon encapsulated iron is efficiently protected by the surrounding shells and its magnetic properties are retained. Often, magnetic characterisation studies have been performed for CNTs fixed to a SiO₂ substrate. For such studies, Fe CNTs are grown directly on the substrate so that a rather dense coating of well aligned nanowires is studied. For such aligned Fe filled CNTs, uniaxial magnetic anisotropy is reported with the easy magnetic axis being parallel to the CNT axis. The anisotropy can be straightforwardly attributed to the shape anisotropy of the magnetic nanowires. However, depending on the alignment [43] and density of CNTs on the substrate magnetic dipole-dipole interactions between the Fe cores have to be considered, too. Interestingly, enhanced magnetic coercive fields of H_C > 100 mT are observed in such ensembles of Fe filled CNTs at room temperature [37], [43] which significantly exceed the value H_C = 9 * 10⁻³ mT observed in bulk Fe. The more representative data for biomedical applications are achieved by powder measurements. Here, depending on the filling ratio, diameter, etc. coercive fields in the range of H_C ∼ 20–50 mT have been found. In a recent study the magnetisation of cells dried after incubation with Fe filled CNTs exhibited a critical field of similar size (∼23 mT) [33].

Magnetic coercivity is even more reduced when the magnetisation is measured in a living cell culture [32]. For such a study, 2 * 10⁵ PC-3 cells were incubated with 50 µg/ml of Fe filled CNTs. After an incubation period of 4 h, the cells were washed two times with PBS, trypsinised and centrifuged at 4°C. This procedure in total yielded approximately 1 * 10⁶ cells which were resuspended in 100 µl for media and measured, at 4°C, in a maximum applied field of H = 0.5 T. Interestingly, the data imply only a vanishing coercivity of H_C ∼ (2 ± 2) Oe.

The magnetisation data suggest the feasibility of Fe filled CNTs by applying an alternating (AC) magnetic field (see for example [62]). The heating effect is based on the physical principle that applied AC magnetic fields induce magnetisation loops. If these loops are not completely reversible (e.g. in the case of ferromagnetic particles or for superparamagnetic particles below the blocking temperature), magnetic energy is transformed into heat. We mention that residual ferromagnetic catalyst particles appear in most synthesis processes which provide superparamagnetism in CNTs (e.g. [63]). Another route to superparamagnetic CNTs employs the insertion of Gd ions to form Gadonanotubes [64].

There are various physical mechanisms which yield non-reversible loops, i.e. which are related to magnetic dissipation processes, and it should be emphasised that in the case of Fe filled CNTs the mechanism has not yet been identified. In general, different processes will be related to different energy scales and resonance frequencies, so that the filling material, its size and its shape have a significant effect on the heating properties. For example there will be strong differences between the case of single-domain particles, where the magnetisation process competes with shape and crystal anisotropies, and the motion of domain walls in multi-domain particles.

The feasibility of Fe filled CNTs for hyperthermia has been demonstrated by a calorimetric study in AC magnetic fields. Our test device consists of a high-frequency generator with an impedance matching network and the magnetic coil system. Water-cooled copper tubes are wound into a coil system (e.g., 4–10 turns, diameter of bore 20–100 mm) in which the sample is placed. Temperature is controlled by a fibre-optic temperature controller (LXT-Luxtron One). The generator can provide alternating magnetic fields in the frequency range of 50–1200 kHz. The data shown in Figure 3 have been obtained at f = 230 kHz. Here, Fe filled MWCNTs have been dispersed by human albumin in PBS. The concentration was 4.2 mg/ml so that about 0.08 mg Fe/ml can be estimated. This concentration value is corroborated by analysis of magnetisation loops. The initial slope of temperature vs. time studies was used to determine the specific absorption rate SAR. In agreement with previous data [33], values of SAR_Fe < 100 W/gFe were found in the field range below ∼20 kA/m which is applicable for medical treatment. We emphasise, however, that the presented first studies do not yet elucidate the actual dissipative mechanism. Detailed information about the magnetisation reversal will help to optimise, for example the geometrical dimensions of the Fe nanowires in order to achieve improved switching behaviour.

Figure 3. Specific absorption rate SAR_Fe of Fe nanowires (∼0.08 g Fe/l) encapsulated in CNT. Measurements have been performed at f = 230 kHz after dispersing the CNT by means of human albumin in PBS (4.2 g/l).

In addition to a magnetic field induced thermal ablation described in detail above, CNTs are also feasible for near-infrared (NIR) light based hyperthermia. This method applies the fact that biological systems are transparent to light in the NIR regime of 0.7–1.1 µm wavelength while the strong absorbance renders single-walled CNTs for hyperthermia agents in living cells [11], [65], [66]. Focused heat transduction and photo-ablative destruction of kidney cancer cells has also been shown for multi-walled CNTs if being nitrogen doped [67]. Although NIR light is capable of passing through several centimetres of tissue, its interaction with biological matter is by far larger than of magnetic fields and the penetration depth is much smaller. For efficient heat transfer to deep tissue and in order to avoid parasitic heating effects in surrounding tissue magnetic agents seem to be advantageous.

Temperature control by NMR on filled CNTs

Accurate control of the tissue temperature is mandatory in any hyperthermia approach. Currently, in the clinical treatments, temperature is controlled by a clinician's intervention by placing thermocouples or fibre-optical thermometers into the tumour, in combination with computer modelling later on. Any model, however, demands estimates of the tissue properties, blood perfusion rate and other dynamic properties [68]. Instead, a continuous non-invasive temperature monitoring appears to be advantageous not only for hyperthermia treatment. Magnetic resonance (MR) thermometry based on a temperature-dependent proton resonance frequency shift of the water molecule provides temperature control in addition to a good spatial localisation, thereby allowing for accurate target identification in ultrasound thermal therapy [69]. However, large doses of unshielded magnets which are present in nanoparticle-based hyperthermia introduce magnetic field inhomogeneities that reduce MRI contrastivity based on the proton relaxation-weighted image and prevent proton-based MR thermometry.

A promising approach for a non-invasive in vivo temperature control relies on the use of a nanoscaled thermometer, which consists of a CNT and a filling material with strongly temperature dependent NMR parameters. In particular, the filling material might exhibit strong T-dependencies of the spin-lattice or the spin-spin relaxation, resonance frequency, dipolar or scalar couplings, and electrical quadrupole coupling at 310°–350°K (ca. 20°–60°C) so that temperature detection is possible with high accuracy (<0.1°). Due to the protecting carbon shell, the number of materials which can be used for temperature sensoring without toxic adverse effects strongly increases. On the other hand, the container feature of CNTs might, in principle, allow simultaneous filling of CNTs with a temperature sensor and another probe such as a ferromagnet (=heater), thereby combining different functionalities in one kind of CNT (Figure 1).

Up to now, the more simple system has been studied, where only the temperature sensor is encapsulated in CNTs [53]. Many alkali and cuprous halides are known to show pronounced temperature dependencies of NMR parameters. From a family of these compounds monovalent cuprous iodide (CuI) turned out to be most suited. Here, both the copper and iodine nuclei have NMR active isotopes with a high natural abundance.

The synthesis of CuI filled CNTs was based on pristine nanotubes consisting of 10 to 40 carbon layers with inner diameters between 5 and 20 nm. For filling with CuI, CNTs were opened using thermal and acid treatment in combination with sonication. The opened CNTs were ground in a mortar gently and put in a silica glass ampoule together with CuI (Aldrich 99.99%) in excess. The ampoule was sealed under vacuum (10⁻³ Torr) and heated at 600°C for 24 h. At this temperature CuI is completely sublimated and transported into the opened CNTs thanks to the capillarity effect. The resultant material was examined by transmission electron microscopy (TEM and HRTEM), X-ray diffraction analysis and energy dispersive X-ray analysis (EDX) and identified as CNTs filled to 80% with single crystalline cuprous iodide.

Although both ⁶³Cu and ¹²⁷I nuclear isotopes possess a quadrupole moment, in CuI copper and iodine atoms each are surrounded tetrahedrally by four atoms of the opposite kind. This leads to a vanishing quadrupolar coupling. Therefore only the resonance frequencies, linewidths and relaxation times can be considered as temperature indicators. The NMR measurements were done in a standard solid state NMR spectrometer in the external magnetic field of 7.05 T. Both ⁶³Cu and ¹²⁷I NMR spectra obtained from the Fourier transformed echo signals represented a single resonance line. The spin-lattice relaxation T₁ was measured employing an inversion recovery pulse sequence. The obtained decay of ⁶³Cu and ¹²⁷I longitudinal magnetisation was analysed following a standard equation for the spin I = 3/2 and I = 5/2, correspondingly. In both cases the magnetisation curves followed a simple exponential form characterised by a single spin-lattice relaxation time.

The analysis of the ⁶³Cu NMR spectra reveals insignificant changes in the resonance frequency in the relevant temperature range, while the resonance frequencies measured on the ¹²⁷I nucleus indicate a stronger dependence on temperature (Figure 4a). Such behaviour is explained by lattice effects which include both the lattice vibration and the lattice expansion. The ¹²⁷I NMR frequency data are well fitted with a linear function over the whole temperature range studied. Thus, at first glance, the ¹²⁷I resonance frequency might be used as a measure for the temperature determination. A relatively small slope of this function leads, however, to an error of 15°K in temperature determination. Therefore the usage of this parameter for an accurate temperature control is not reasonable. Furthermore, the spectral linewidths at half maximum have been analysed for both nuclei and were found to be constant at temperatures below 320°K. Thus, also the linewidths could be ruled out to serve as a temperature control parameter.

Figure 4. Temperature dependencies of ¹²⁷I, ³⁵Cl and ⁶³Cu nuclear magnetic resonance parameters measured on CuI- and AgCl-filled CNT. (a) Nuclear magnetic resonance frequency. (b) Nuclear spin-lattice relaxation rate. The symbols present the experimental data. Solid lines are the fit (see the text).

The temperature dependence of the ¹²⁷I spin-lattice relaxation rates is presented in Figure 4b. The ⁶³Cu dependence is similar but much smaller compared to the ¹²⁷I data. The dependencies for both nuclei are found to be in very good agreement with the law that is expected for a Raman two-phonon quadrupolar process [70]. This behaviour is observed over the entire temperature range implying no contributions from impurities which might appear at low temperatures and from ionic diffusion which might be observed at high temperatures. This is consistent with the view that relaxation is driven by a quadrupolar mechanism in this compound [71]. The ¹²⁷I experimental data are well fitted with a quadratic function = a + bT + cT², where fitting coefficients are a = 1, b = (7 ± 1) * 10⁻² and c = (1.49 ± 0.05) * 10⁻³. The mean squared errors of the fitting coefficients provide an estimate of the accuracy of the temperature determination. In the temperature range of biological interest (i.e. 290° to 320°K) the CuI-CNT nanothermometer can indicate the temperature with an accuracy of 2°K by means of the spin-lattice relaxation measurement. Other Cu-halides filled in CNTs demonstrate a qualitatively similar behaviour but exhibit less temperature sensitivity as shown in Table I. These results, in particular those on CuI-CNTs, provide a good starting point to look for further filling materials of CNTs in order to increase the accuracy of temperature determination [72].

Table I.  Temperature sensitivity parameters of several filled CNTs. The table shows the filling material, the respective nucleus as well as temperature dependence of resonance frequency and T₁-relaxation time in the temperature range of 300°–320°K.

Material	Nucleus	dνres/dT (Hz/K)	d()/dT (Hz/K)
CuI-CNTs	^63Cu	–	0.27
	^127I	14	0.86
CuBr-CNTs	^63Cu	19	0.15
	^81Br	7	0.75
CuCl2-CNTs	^63Cu	15	0.23
	^35Cl	1	–

Conclusions

Due to their extraordinary physical and chemical properties carbon nanotubes reveal a promising potential for applications on the cellular level. Upon filling, nanoscaled containers are realised in which the active material is encapsulated by a protecting carbon shell. We describe different synthesis routes and show the structural and magnetic properties of CNTs. In particular, the filling with magnetic material offers the potential for hyperthermia applications while the insertion of NMR active substances allows the usage as markers and sensors. This potential of carbon nanotubes for biomedical applications is highlighted by hyperthermia studies which prove their applicability for local in situ heating. In addition we have shown that a non-invasive temperature control by virtue of a carbon-wrapped nanoscaled thermometer is possible.

A valuable extension would be spatially resolved NMR. The rapidly growing field of cellular and molecular MR imaging enables the visualisation of cells and insertion into CNTs in order to track cancer cells and to control therapies on the cellular level such as magnetic hyperthermia. The container feature of CNTs is extensively utilised if a heating element (ferromagnet), a temperature sensor and a contrast agent are confined within the same nanocontainer. Such a combination of different functionalities on a nanoscale would provide simultaneous heating, temperature control by means of MRI and high spatial resolution of the image. Furthermore, if clinical usage of the static magnetic field is contra-indicated and MR imaging is no longer suitable then the technology proposed here addresses materials with temperature dependent nuclear quadrupolar resonance NQR (e.g. cuprous oxide) or zero-field NMR (e.g. co-based compounds) parameters that demonstrates the versatility of this approach for biomedical applications. The potential of CNTs for biomedical applications becomes even more evident if their container function is exploited for drug delivery in magnetically functionalised and NMR labelled nanodevices. It has been shown that anti-cancer drugs can be inserted in CNTs [54] for which purpose the multi-walled ones turn out to be particularly appropriate. Combing different functionalities in well shielded containers therefore seems to be the particular advantage of carbon nanotubes.

Acknowledgements

This work was partly supported by the European Community through the Marie Curie Research Training Network CARBIO under contract no. MRTN-CT-2006-035616. The authors thank Diana Haase, Manfred Ritschel, Anastasia Vyalikh, Anja Wolter, Kamil Lipert, Yulia Krupskaya, Christopher Mahn, Thomas Mühl, Dieter Elefant and Hans-Jörg Uhlemann from IFW Dresden as well as Arthur Taylor and Kai Krämer from the Department of Urology, Technical University Dresden. We particularly appreciate Albrecht Leonhardt for support and valuable discussions. Scientific advice by Sabine Achten (Boehringer Ingelheimer Fonds), Francois Rossi (European Commission Joint Research Centre, Ispra) and Andreas Jordan (Magforce Nanotechnologies AG) is gratefully acknowledged.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.