Does halloysite behave like an inert carrier for doxorubicin?

Abstract

Halloysite (Al₂Si₂O₅(OH)₄·nH₂O), a kind of natural mineral with unique tubule nanostructure, has great potential application as a drug carrier, due to its good biocompatibility, large surface area, inherent large aspect ratio and low-cost. However, the surface of natural clay may have strong interactions with some adsorbates, especially with heterocyclic molecules that are common in chemical drugs. Indefinite interaction between drug molecules and the halloysite surface significantly hampers the application of halloysite in drug-delivery. In this study, doxorubicin and halloysite were chosen as models to investigate the interaction between drugs and carriers. We demonstrated the impact of this interaction on the thermal stabilities of both doxorubicin and halloysite nanotubes, based on the characterization using thermogravimetric analysis and thermogravimetric analysis-mass spectrometry (TGA-MS). TGA-MS revealed more details about doxorubicin adsorbed on halloysite which is unobserved in other characterizations. We also compared the inherent nanostructures, reactivity and doxorubicin-loading capacities of two kinds of halloysite nanotubes, to clarify that neither structural features nor modification of halloysite dominates the interaction of doxorubicin and halloysite. This investigation deepens our understanding of halloysite nanotubes as drug carriers, which will benefit their further application in biomedical and pharmaceutical areas.

---

Introduction

Nanomaterials with a large aspect ratio in one or two dimensions, such as nanotubes,¹ nanorods,² and nanodiscs,³ show extraordinary advantages of delivering drugs into cells due to their particular shapes.^4–6 Halloysite, as a kind of natural clay mineral composed of aluminosilicate, has a hollow inner lumen with a diameter of about 10–20 nm. Vergaro et al. showed that halloysite exhibits high level of biocompatibility and very low cytotoxicity.⁴ In vitro assessment of the cytotoxicity of halloysite nanotubes against HCT116 (colorectal carcinoma), HepG2 (hepatocellular carcinoma) and human peripheral blood lymphocytes confirmed that they are generally safe for oral dosage.⁵ In vivo study also demonstrated that halloysite nanotubes are nontoxic to nematodes.⁶ Chitosan-modified halloysite showed improved hemocompatibility and stability.⁷ More importantly, halloysite nanotubes, as abundant natural minerals, are much cheaper than the man-made nanostructures as ingredients in pharmaceutical products. Therefore, their application as nano-vehicles for drug delivery is very promising, and has been widely studied for years.^8–17 Besides, halloysite can also be used as scatterers for safe ultrasound-based molecular imaging.¹⁸ Many studies have reported that drug molecules or proteins loaded on halloysite nanotubes presented a sustained release, by reason of the restricted diffusion of molecules while they are located inside lumen of halloysite nanotubes.^19–24 Specific case of extraordinary elongated release has been observed and it was attributed to specific interaction between drug molecules with halloysite nanotubes.⁸ However, there is insufficient investigation of the interaction between halloysite nanotubes and drug molecules. Previous studies have demonstrated that the surface of natural clay may have strong interaction with some adsorbates, which is fundamental for recognition and adsorption of organic molecules.¹¹ Because direct theoretical investigation of drug molecule-natural halloysite system is too complex, researchers usually have to take individual groups into accounts in their studies to simplify it. For example, indole is strongly adsorbed on the aluminum hydroxide surface of kaolinite via hydrogen bond.¹⁴ Various functional groups, such as pyrrole, pyridine, thiophene, and carboxylic acid can be bonded onto aluminum hydroxide surface by strong dipolar and hydrogen bonding.^12,17 Supramolecular assembly model combining cooperative bonds of Brönsted acid–base interactions, hydrogen bonding, aromatic π–π stacking, metal coordination complexes and hydrophobic interactions, is also used to interpret the aggregation and adsorption of heterocyclic molecules on minerals.¹³ In general, most drug molecules are heterocyclic compounds and have complex molecular structures, which may help in bounding drug molecules to the surface of halloysite nanotubes via multiplex interactions. However, the integrated effect of functional groups and other structural factors of halloysite, such as surface area and pore size, is hard to be predicted all round. Thus experimental studies show comparative advantages in systematically comparing different experimental factors in drug–halloysite system.²⁵ The interaction strength, as well as other structural parameters, collectively affects the loading capacity and releasing dynamics of drugs. Growing knowledge is gradually obtained through experiments based on different characterization techniques. Using new technique will reveal hidden details of drug–halloysite interactions. On the other hand, being a natural mineral, halloysite nanotubes have a critical shortage as drug carriers: their original nanostructure and composition may be varied with their origins. The original nanostructure of halloysite nanotubes directly influences their surface area, pore size and chemical activity that may strongly change their reactivity and surface charge. Besides, choosing a proper modification strategy for halloysite nanotubes should also be based upon clear understanding of their inherent nanostructures. However, the comparative study of different kinds of halloysite minerals in the application of drug carriers is lacking.

In this study, we chose doxorubicin (DOX), a commercial available and effective antitumor drug, as a model drug, to investigate its interaction with halloysite nanotubes. In particular, thermogravimetric analysis-mass spectrometry (TGA-MS) has been firstly used in the analysis of doxorubicin–halloysite interaction. Mass spectrometry following thermogravimetric analysis provides detailed information in thermal decomposition, which is directly related to the interaction between absorbent and surface. To avoid that the observed interaction is only due to uncommon specificity of halloysite nanotubes generated in unique mineral deposits, we have comparatively studied halloysite nanotubes originated from two different areas. Extensive characterizations including of morphology, size, crystallization degree, surface charge, surface functional groups of halloysite nanotubes were conducted to clearly identify the structures and properties of halloysites, before and after alkaline treatment. We investigated the interaction between doxorubicin and halloysite through the study of thermal decomposition behaviors of doxorubicin loaded on halloysite nanotubes. We observed that: (1) specific interaction indeed exists between DOX and halloysite surfaces, which strongly influences the thermal behavior of DOX@halloysite and enhances the loading capacity of DOX, but may retard the release of DOX; (2) optimized chemical treatment will generate more porous niches to load DOX; (3) original nanostructure of halloysite nanotubes is also critical to their drug-delivery performance, but shows no dominating impact on the interaction between DOX and halloysite, indicating that the interaction is a typical result for DOX@halloysite. Such a fundamental knowledge of halloysite nanotubes is important to their potential applications in biomedical and pharmaceutical areas.

Experimental section

Materials

Two kinds of halloysite nanotubes, originated from Zhengzhou, China (named as zHNTs) and Shandong Province, China (named as sHNTs), were used in our comparative study. zHNTs was purchased from Zhengzhou Jinyangguang Ceramics Co. Ltd., sHNTs was obtained from Dongmingtianhe High-temperature Materials Co. Ltd. Doxorubicin hydrochloride (DOX·HCl, ≥99%) was obtained from Beijing Zhongshuo Pharmaceutical Technology Development Co. Ltd., and was used as received. Sodium hydroxide (NaOH) was purchased from Sinopharm Chemical Reagent Co. Ltd. Deionized water (Millipore Q ≥18 MΩ cm) was used in preparing aqueous solutions.

Chemical treatment of halloysite nanotubes

Halloysite nanotubes have been grinded to fine powder before further chemical treatment. Then, they were well dispersed in NaOH aqueous solution (5 wt%) by stirring and soft sonication. The concentration of HNTs was kept at 10 mg mL⁻¹. HNTs suspensions were kept in 60 °C water bath for varied times from 5 to 15 hours. After that, HNTs were collected by centrifugation and washed by deionized water several times until the pH of washed solution was neutral. Then HNTs products were dried in oven (100 °C) overnight and ready for further characterizations and experiments.

Morphological characterizations

Scanning electron microscopy (SEM, JEOL 7401) was conducted at an accelerating voltage of 5 kV. The samples were coated with a thin layer of gold before SEM characterization. Chemical composition of HNTs samples were examined using a scanning electron microscopy (FEI Nova Nano SEM 450) equipped with an energy dispersive spectrometer (EDS, Oxford Instrument, X-MAX 50 mm²). Transmission electron microscope (Hitachi TEM, H-7650B) was operated at an accelerating voltage of 80 kV. A droplet of halloysite nanotubes suspension in water or ethanol was placed onto copper grids for TEM analysis.

Chemical and structural characterizations

Infrared spectra were collected on a Nicolet Magna IN10 FTIR spectrometer from 4000 to 650 cm⁻¹. Wide angle X-ray diffraction was performed with XRD-7000 diffractometer (Shimadzu) in the reflection mode using Cu target (λ = 0.15418 nm) as incident X-ray. The scan step was 0.02°, scan scope was from 5° to 80°.

Zeta potential and particle size distribution measurements

The zeta potential values and particle size distribution of halloysite nanotubes in aqueous solution at different pH values were measured using a Zetasizer (Nano-ZS, Malven Instruments) at 25 °C. Measurements were performed five times for each sample.

BET measurement

Brunauer–Emmett–Teller (BET) surface areas of halloysite nanotubes were determined from N₂ adsorption isotherm data collected at 77 K by using Micrometric TriStar II 3020 surface area and porosity system. Before measurement, 100 mg samples were outgassed at 120 °C for 12 h. The surface areas were calculated according to the BET method. Mesoporous volume and pore size distribution were calculated using the Barrett, Joyner and Halenda (BJH) method assuming cylindrical pores in the Kelvin equation.

Doxorubicin loading and in vitro release

Drug loading was performed by incubating 2 mg of HNTs with 4.0 mL DOX stock solution (0.5 mg mL⁻¹) prepared in PBS solution (pH = 6.5). Sonication was conducted several times in the incubating period to encourage the release of gas bubble confined inside halloysite lumen. After stirring at 37 °C for 18 h, DOX-loaded HNTs were collected by centrifugation, and then washed several times with PBS solution (pH = 6.5) to remove the physically adsorbed DOX residues on the surface. Finally, the drug content was determined by UV-Vis spectrometer (Cary 50 UV-Vis instrument) at 479 nm. The loading capacity of DOX in HNTs was calculated according to the following equation:

Moreover, the loading capacity of all DOX@HNT samples was calculated based on thermogravimetric analysis. They were presented in ESI Fig. S9† for a comparison. The release kinetics of DOX from HNTs was measured by using dialysis method. 4 mL of DOX@HNTs (4 mg)/PBS buffer was loaded in dialysis bag (cut-off M_w = 3500) and then the bag was dipped in PBS buffer solution (80 mL) with different pH values (5.0 or 7.4). The samples were incubated at 37 °C under mild shaking. At certain time intervals, 3 mL of the release medium was taken out to measure the released drug concentration and then was returned to the original release medium. For the measurement of released DOX concentration, the absorbance of the release medium at 479 nm was recorded on UV-Vis spectrometer.

The investigation of interactions between doxorubicin and halloysite nanotubes

Thermogravimetric analysis was conducted by using Q600 (TA Instrument, USA). The samples were subjected to heating at a rate of 10 °C min⁻¹ under N₂ (flow rate of 100 mL min⁻¹) from room temperature to 600 °C. Thermogravimetric analysis-mass spectrometry (TGA-MS) was performed by using TG-DSC-QMS (STAA449C16/G+QMS403, Netzsch, Germany). The samples were subjected to heating at a rate of 10 °C min⁻¹ under Ar (flow rate of 30 mL min⁻¹) from room temperature to 600 °C. Mass spectra of six specific mass-ions (m = 15 for CH₃⁻, m = 17 for NH₃ or OH⁻, m = 18 for H₂O, m = 35 for Cl⁻, m = 37 for HCl and m = 44 for CO₂) were detected within temperature range.

Results and discussion

Generally, halloysite nanotubes have been modified by acid or alkaline treatments.^15,26,27 The unique advantage of halloysites as a drug carrier is their shape, their tubular structure with hollow lumen, so we need to balance the etching effect to avoid severe structure damage. Based on the reaction mechanism, alkali treatment can react with both Al–O and Si–O, and it is reported that the hydrolysis of silica happens at defects or edge sites,¹⁵ we suppose that optimal alkaline treatment will create more pores on halloysite walls in keeping their basic structural feature. Although acid treatment disaggregates clay particles, removes impurities and increases the surface area, it also causes the generated amorphous silica nanoparticles aggregated inside halloysite.^15,28 Moreover, Wang et al. indicated that although acid-treated halloysite showed increase surface area, the loading capacity of drug (ofloxacin) has not been enhanced.²⁷ In our study, all halloysite nanotubes were modified by alkaline solution. Fig. 1 presents the morphological characterization of zHNTs and sHNTs with or without etching treatment. Tubular structures of both zHNTs and sHNTs are generally kept after alkaline etching for 8 h, longer etching time (10 h or 15 h) will induce the breakage of tubular structure (Fig. 1 and S1 and S2†). The inset picture of Fig. 1c clearly shows a cavity of broken nanotubes. For tracking the composition change of halloysite in alkaline treatment, EDS characterizations of sHNTs were conducted (Fig. S3†). The atomic percentages of both aluminium and silicon decreased with the elongation of etching time, showing that both of them have been etched in alkaline condition. TEM images demonstrated the structural change of zHNTs and sHNTs after etching process more clearly (Fig. 2). Severe structural damage of halloysite nanotubes is observed in both 10 h and 15 h treated zHNTs and sHNTs (Fig. 2 and S1†). Some outer layers were peeled off (Fig. 2c and f). The size distribution of zHNTs and sHNTs after treatment were shown in Fig. S4.† The length of original sHNTs is longer than that of zHNTs, so they are much more stubborn under NaOH treatment. Previous study showed that alkaline molten-salt (Na₂CO₃ and NaNO₃) treatment can etch surface of tube wall and increase the roughness of halloysites.¹⁸ Their BET surface area was changed from 44.98 to 51.07 m² g⁻¹. In another report, the BET surface area of halloysite treated in NaOH solution (1 mol L⁻¹) at room temperature for 28 days is 45 m² g⁻¹.¹⁵ We noticed that in our solution-etching process, the increase of surface area of halloysite after treatment is significant (Fig. 3, S5† and Table 1): the increasing ratio of surface area for zHNTs reaches 53.4% (5 h treated) and 64.9% (8 h treated), that for sHNTs is 49.8% (5 h treated) and 58.7% (8 h treated). Samples showing highest surface areas for both zHNTs and sHNTs are that after 8 h treatment, 58.81 m² g⁻¹ (sHNTs) and 84.17 m² g⁻¹ (zHNTs), respectively. A further elongation of treating time will induce a decrease of surface areas (47.18 m² g⁻¹ for 10 h and 44.90 m² g⁻¹ for 15 h), which maybe because of the blocking of generated micropores by debris of etched halloysites (Fig. S1†). In literature, it is also reported that the nanoparticles and nanosheets originated from the collapsed fragile nanotubes, will aggregate outside the nanotubes, decreasing the accessibility of inner lumen of halloysite nanotubes.¹⁵ The optimal etching time for enhancing surface area has also been observed in acid treatment of halloysite.²⁸ In addition, the alkaline solution treatment also enhances pore volume from 0.168 cm³ g⁻¹ (sHNTs, original) to 0.233 cm³ g⁻¹ (sHNTs after 5 h treatment) and 0.259 cm³ g⁻¹ (sHNTs after 8 h treatment), which will benefit the loading of drugs. The most-fascinating structural feature of halloysite nanotubes is their empty lumen, which directly influences the loading capacity of target materials. Researchers demonstrated varied pathways to modify or enlarge the inner lumen for enriching applications of halloysite nanotubes.^16,29–33 Lvov et al. have enlarged the inner lumen by selective etching of aluminum oxide of sulfuric acid.³⁴ The enlargement of inner lumen significantly increases the loading capacity on halloysite nanotubes. We found that after 8 h treatment, the average size of inner lumen of zHNTs was changed from 12 ± 2 nm to 23 ± 2 nm, that of sHNTs was varied from 10 ± 5 nm to 20 ± 5 nm (Fig. 2b and e, S6†). These values are close to the average pore sizes measured in BET measurements (Table 1). The chemical and structural changes of halloysite after alkaline treatment have also been characterized. FTIR characterizations of zHNTs and sHNTs, before and after alkaline treatments, showed that such treatments have not affected the chemical nature of halloysite significantly, which is in agreement with previous report (Fig. S7†).²⁰ WXRD demonstrated the typical crystallized structure of halloysite before and after etching (Fig. 4). The change of layer distances of zHNTs and sHNTs after etching are minimal. But some impurity in halloysite was removed after alkaline treatment, which is the possible reason why the peak at 31° in original two kinds of HNTs disappeared after alkaline treatment. But quartz (2θ ∼ 27°, d = 3.34 Å)³⁵ in all treated samples still exists, because it has high stability, alkaline treatment fails to eliminate it. Based on the d₀₀₁ peak, we can find that the layer distance between original zHNTs (7.26 Å) is slightly larger than that of original sHNTs (7.18 Å). This is an inherent structural feature of zHNTs.

Fig. 1 SEM images of zHNTs and sHNTs samples after various etching times: 0 h, 8 h, and 15 h, respectively. Scale bars are 500 nm.

Fig. 2 TEM images of zHNTs and sHNTs samples after various etching times: 0 h, 8 h, and 15 h, respectively. All scale bars are 100 nm.

Fig. 3 BET characterizations of original zHNTs (a) and sHNTs (c) samples. The change of surface areas of zHNTs (b) and sHNTs (d) after various etching times: 0 h, 5 h, 8 h, 10 h, and 15 h, respectively.

Table 1 BET measurements of sHNTs and zHNTs before and after different reaction times

Sample		Surface area (m^2 g^−1)	Pore volume (cm^3 g^−1)	Average pore size (nm)
0 h	zHNTs	51.05	0.296	22.99
	sHNTs	37.04	0.168	15.85
5 h	zHNTs	78.32	0.413	18.11
	sHNTs	55.50	0.233	14.47
8 h	zHNTs	84.17	0.475	20.75
	sHNTs	58.81	0.259	15.79
10 h	zHNTs	80.85	0.481	22.51
	sHNTs	47.18	0.241	16.74
15 h	zHNTs	80.28	0.465	21.58
	sHNTs	44.90	0.233	17.13

---

Fig. 4 WXRD patterns of zHNTs (a) and sHNTs (b) after various etching times. The reflection marked with * corresponds to quartz.

Surface charge of carriers shows influence to drug-loading. Previous studies indicated that negatively and positively charged molecules are nearly equally adsorbed on halloysite: 39 mg g⁻¹ for Chromeazurol S (negative) and 44 mg g⁻¹ Rhodamine 6G (positive), owing to the co-existence of negatively charged external surface and positively charged inner lumen on halloysite.⁸ We noticed that the original negatively charged halloysite changes to positive surface charge after DOX-loading, indicating that DOX molecules have been adsorbed on halloysite nanotube surface (Fig. 5). How about the filling of inner lumen of HNTs? TEM images of DOX@HNTs clearly demonstrate the strong adsorption and slight filling of DOX into halloysite (Fig. 6 and S8†). A thick DOX layer covers the halloysite surface as indicated by arrows in Fig. 6. Though the positive inner lumen has repulsive effect on DOX molecules, the filling of DOX still happens mainly because of the strong capillary force of HNTs.^24,36 The majority of drug is located outside of HNTs. The significant increase of the diameter of DOX@HNT tubes (ca. 2 times) confirmed it. Solute (DOX) is soaked into HNTs with liquid and is left inside in drying process, but it may mix with debris of HNTs and is hard to be clearly assigned (Fig. 6b and S8†). DOX is concentrated inside HNTs and will be rarely re-dissolved and removed in washing steps because of the strong spatial confinement. Besides, the structural defects on aluminosilicate layers of HNTs appeared after alkaline treatment, which provide more “niches” for DOX-loading, resulting in a slight increase of loading capacity of DOX on treated-halloysites (Fig. S9†). The loading capacities of zHNTs and sHNTs were calculated based on UV-Vis spectra and TGA characterizations (Fig. S9†). The values (DOX@zHNTs) obtained from TGA curves showed a slightly larger change with batches because of the higher water contents in zHNTs. The loading capacity of zHNTs is integrally higher than that of sHNTs, because zHNTs have larger surface areas. The inherent structure of halloysite is the dominant factor in loading capacity. In view of the severe structural damage of longer treatment to halloysite, 10 h-treated and 15 h treated halloysite showed a decrease either in their surface area (Fig. S5†) or in loading capacities.

Fig. 5 Zeta potential of DOX@zHNTs, DOX@sHNTs compared with that of pure zHNTs and sHNTs at different pH values. (a) zHNTs 0 h and DOX@zHNTs. DOX was loaded on zHNTs samples treated with NaOH aqueous solution for different times. (b) sHNTs 0 h and DOX@sHNTs. DOX was loaded on sHNTs samples treated for different times.

Fig. 6 TEM images of DOX@zHNTs (0 h). DOX covers halloysite surface and is trapped inside their inner lumen as indicated by arrows.

Acidity of solutions shows significant effect on DOX release which indicates that the adsorption of DOX is determined by electrostatic force. In all four cases, the releasing percentages of DOX@HNTs at pH 7.4 are lower than 5%, but these values increase to over 30% at pH 5.0 (Fig. 7). We suppose that the following two factors may contribute to the low release percentage of DOX@HNTs. (1) Spatial confinement strongly blocks the release of those DOX molecules which are located inside HNTs inner lumen. Previous study reported that release time of water soluble tetracycline extends more than 25 times as loading in halloysite.³⁷ That is an identical case of diffusion-blocking by tubular nanocarriers. (2) The interaction between DOX·HCl and HNTs is a dominating factor because DOX filled inside HNTs inner lumen is not the majority in total loading amount. A similar release behaviour has been observed in protein/halloysite case.³⁸ Tully et al. recently have reported that only 50% proteins were release in 4–6 h, another half was staying in the tubes. They also assigned it to the contribution of strong binding of drugs to the tube surface. For the negatively charged HNTs surface, the interaction nature is electrostatic force between HNTs surface and positively charged DOX·H⁺ ions.

Fig. 7 Releasing kinetics of DOX@zHNTs (a and b) and DOX@sHNTs (c and d) at pH 5.0 and 7.4. (a) DOX@zHNTs (0 h), (b) DOX@zHNTs (8 h), (c) DOX@sHNTs (0 h), (d) DOX@sHNTs (8 h).

How strong is this interaction? Does this interaction change properties of loading DOX? We hope to deepen our understanding of this interaction through characterization of thermal decomposition of DOX@HNTs (Fig. 8). Differential thermogravimetry is a convenient method to determine the rate of thermal decomposition at various decomposition stages. The temperatures at which maximum-decomposition rate is observed were listed in Table 2 based on differential thermogravimetry analysis. We found that all HNTs samples with different treatments keep stable from room temperature to 460 °C when de-hydroxylation in HNTs structures occurs and their multilayered packing of tube wall disappears after losing structural water.¹¹ Pure DOX shows two decomposition stages, one is 207–249 °C, the other one is around 356 °C. So we separated the thermal decomposition of DOX@HNTs to three stages, the two stages in the range of 200–400 °C are derived from DOX decomposition, the last one around 460 °C is from HNTs. Being loaded on halloysite has obvious effect on DOX thermal stability: it makes the DOX less thermally stable. The first thermal decomposition stage of pure DOX (207–249 °C) was shifted to lower temperature in all DOX@HNTs samples. It seems that the interaction between treated HNTs and DOX is stronger than that of original HNTs and DOX, because original two peaks of DOX loaded on treated HNTs merged to one peak, but they were at least kept two peaks in DOX loaded on unmodified HNTs. This is partly because alkaline treatment can introduce more ionic defects that enhance the electrostatic interaction between HNTs and DOX. In addition, we noticed that the second decomposition stage is very weak for pure DOX, but it changes to comparable magnitude with first decomposition stage in all DOX@HNTs samples, and shifts to lower temperatures. That may be due to some changes in decomposition mechanisms. Moreover, the third decomposition stage belonging to the decomposition of HNTs, is also varied because of the loading of DOX. We suppose that in loading process, the intercalation of small ions into halloysite layered structures may change structural water in halloysite, leading to a change of de-hydroxylation process. It further confirms that after DOX is loaded on halloysite, they form a “complex” in which both DOX and HNTs are changed.

Fig. 8 Differential thermogravimetric analysis of DOX@zHNTs (a) and DOX@sHNTs (b). Pure HNTs and DOX were shown as references.

Table 2 Temperatures of maximum decomposition-rate of DOX@HNT and HNT samples in different decomposition stages, derived from derivative weight change with temperatures

Sample name	Peak temperature
	First stage (°C)	Second stage (°C)	Third stage (°C)
DOX·HCl	207–249	356	—
zHNTs (0 h)	—	—	474
zHNTs (8 h)	—	—	466
zHNTs (15 h)	—	—	474
DOX@zHNTs (0 h)	186–214	330	472
DOX@zHNTs (8 h)	207	341	485
DOX@zHNTs (15 h)	206	340	479
sHNTs (0 h)	—	—	491
sHNTs (8 h)	—	—	480
sHNTs (15 h)	—	—	478
DOX@sHNTs (0 h)	190–211	331	472
DOX@sHNTs (8 h)	197	335	481
DOX@sHNTs (15 h)	197	338	461

---

Thermogravimetry-mass spectrometry (TGA-MS) is a hyphenated thermal analytical technique in which released volatiles of testing objects are transferred to a dedicated quadrupole mass spectrometer. The additional mass information of released gas products allows researchers to accurately deduce the thermal decomposition process of tested objects. In this study, we performed TGA-MS characterizations of pure DOX, DOX@zHNTs (0 h) and DOX@sHNTs (0 h). On the basis of chemical structure of DOX and FTIR spectra of thermally-decomposed DOX products at different temperatures (Fig. S10†), we set six ions as targeting ions for mass spectrometer, that are CH₃ (m = 15), NH₃ or OH⁻ (m = 17), H₂O (m = 18), Cl⁻ (m = 35), HCl (m = 37) and CO₂ (m = 44). Fig. S11† presents the change of ion current intensity of six detected ions (CH₃, OH⁻, H₂O, Cl⁻, HCl and CO₂) with temperatures. The intensity of CH₃ of DOX shows increase after 200 °C, but those for DOX@zHNTs and DOX@sHNTs became undistinguishable because of small amount (Fig. S11a†). The ions of (OH⁻ and H₂O) may come from both DOX and halloysite in DOX@HNTs samples (Fig. S11b and c†). DOX@zHNTs and DOX@sHNTs presented two stages in their releases, the first one may come from loading DOX that is similar with pure DOX, the second one may be originated from halloysite. However, the releases of HCl (or Cl⁻) and CO₂ have single origin (DOX), which makes it easy to observe the change of DOX before and after loading on halloysite (Fig. S11d–f†). Fig. S11e† presents the results of released HCl during thermal analysis of DOX and DOX@HNTs. In the thermal decomposition of pure DOX, intensity of HCl increases sharply in the range of 200 °C to 300 °C; however, for DOX@zHNTs and DOX@sHNTs, the intensity significantly increases from 450 °C to 550 °C. This phenomenon is somehow opposite to our prediction. Based on the molecular structure of DOX, the Cl⁻ ion is connected with amine groups by electrostatic force, and it should not been changed much after DOX adsorbed on halloysites. But the TGA-MS result revealed that Cl⁻ ion may be an important participant in the adsorption. We found some supports from theoretical studies. Monte Carlo simulation indicates that chloride ion has specific absorbability on α- and γ-alumina surface,³⁹ which may assist the intercalation of DOX·HCl into HNTs layers. If chloride of DOX·HCl was intercalated into HNTs layers, its thermal release is confined, and as a consequence, the maximum of HCl and Cl⁻ intensity shifts to higher temperatures. In addition, the intensities of CO₂ from DOX, DOX@zHNTs and DOX@sHNTs show clear difference (Fig. S11f†). The release of CO₂ from DOX has two stages, first small peak is at 250 °C, the second huge peak starts around 390 °C; but for DOX@zHNTs, the detected CO₂ started from 250 °C; for DOX@sHNTs, the intensity of CO₂ showed a large peak from 300 °C to 380 °C. The release of CO₂ in DOX@zHNTs and DOX@sHNTs happens at lower temperatures compared with that of DOX, in another word, DOX@HNTs are more unstable compared with DOX. This observation coincides with our differential thermogravimetry analysis. TGA-MS characterization presents unique details of DOX@HNTs system, which may provide more clues of the adsorption of DOX molecules on halloysite surface, helping us analyze it in-depth.

Above characterizations indicate that there is strong interaction between DOX·HCl and halloysite, which makes a significant change in their thermal behaviours. The existence of strong interaction assists us in understanding the low releasing percentage of DOX@HNTs, although it is only somehow a superficial investigation. This observation reminds us of the interaction between drug carriers and loading drugs which may directly influence the adsorption and desorption mechanism, especially in case of nano-sized carriers, because nano-carriers have large surface areas, complicated and intense surface interactions. Further research of interaction mechanism of DOX and HNTs requires more accurate characterizations combining surface chemistry and organic chemistry. By the way, we noticed that in literature some inorganic nanoparticles such as silica nanoparticles as drug carriers, often show similar low releasing percentage.^40,41 The study of interaction between inorganic nanoparticles and drugs is fundamental and application-targeting. This knowledge will assist us to develop cheap and practical drug carriers based on natural clays or other inorganic nanoparticles.

Clay nanotubes are un-biodegraded that may induce suspicions for their application as injected-delivery system.^20,42 Similar discussions have also been conducted in the study of delivering-system based on carbon nanotubes. However, the tissue biodistribution, blood clearance and urinary excretion of un-biodegradable nanoparticles strongly relate with their surface chemistry, particle size, shape and colloidal stability. Recent study reported that chitosan-modified halloysite nanotubes have improved hemocompatibility and stability.⁷ Detailed studies on optimizing particle size and surface chemistry of halloysite nanotubes may answer this question that will stimulate their application in drug delivery.

Conclusions

Here we demonstrated our investigations of natural halloysite nanotubes as drug carriers for doxorubicin. Two kinds of halloysite nanotubes were treated in alkaline environment at 60 °C. The changes of their morphologies, structures and surface properties were carefully characterized by using SEM, TEM, XRD, BET and FTIR spectrometry. The loading and releasing of DOX@HNTs were measured for all samples with different origins and chemical treatments. In particular, we observed significant changes of thermal decomposition of doxorubicin before and after loading on halloysite, as well as halloysite itself, by using differential thermogravimetric analysis and TGA-MS. TGA-MS characterizations present detailed and useful clues of the adsorption of DOX on halloysite surface, guiding us understand it in-depth. Our studies indicated that the interaction between halloysite and DOX is strong as forming a kind of complexes, which is beneficial to high loading-capacity and sustained release, but at the same time, it brings side-effect as low release percentage (∼30%) in all loading-DOX. This interaction is ubiquitous and independent of origins of natural halloysite and the treatment of halloysite. A further study of interaction mechanism between DOX and halloysite will benefit the practical application of natural halloysite as drug carriers.

Acknowledgements

The authors gratefully acknowledge the National Natural Science Foundation of China (Grants 21374132, 51173201) for financial support.

Notes and references

1. G. Cavallaro, G. Lazzara, M. Massaro, S. Milioto, R. Noto, F. Parisi and S. Riela, J. Phys. Chem. C, 2015, 119, 8944–8951.
2. S. Barua and S. Mitragotri, ACS Nano, 2013, 7, 9558–9570.
3. S. Wang, F. Zheng, Y. Huang, Y. Fang, M. Shen, M. Zhu and X. Shi, ACS Appl. Mater. Interfaces, 2012, 4, 6393–6401.
4. V. Vergaro, E. Abdullayev, Y. M. Lvov, A. Zeitoun, R. Cingolani, R. Rinaldi and S. Leporatti, Biomacromolecules, 2010, 11, 820–826.
5. F. R. Ahmed, M. H. Shoaib, M. Azhar, S. H. Um, R. I. Yousuf, S. Hashmi and A. Dar, Colloids Surf., B, 2015, 135, 50–55.
6. G. I. Fakhrullina, F. S. Akhatova, Y. M. Lvov and R. F. Fakhrullin, Environ. Sci.: Nano, 2015, 2, 54–59.
7. M. X. Liu, Y. Z. Chang, J. Yang, Y. Y. You, R. He, T. F. Chen and C. R. Zhou, J. Mater. Chem. B, 2016, 4, 2253–2263.
8. Y. M. Lvov, D. G. Shchukin, H. Mohwald and R. R. Price, ACS Nano, 2008, 2, 814–820.
9. S. R. Levis and P. B. Deasy, Int. J. Pharm., 2002, 243, 125–134.
10. C. Viseras, P. Cerezo, R. Sanchez, I. Salcedo and C. Aguzzi, Appl. Clay Sci., 2010, 48, 291–295.
11. P. Boulet, H. Greenwell, S. Stackhouse and P. Coveney, J. Mol. Struct.: THEOCHEM, 2006, 762, 33–48.
12. T. Zoungrana, A. Berrada, J.-M. Douillard and S. Partyka, Langmuir, 1995, 11, 1760–1767.
13. M. R. Gray, R. R. Tykwinski, J. M. Stryker and X. Tan, Energy Fuels, 2011, 25, 3125–3134.
14. Q. Wang, Y. Wang, Y. Zhao, B. Zhang, Y. Niu, X. Xiang and R. Chen, CrystEngComm, 2015, 17, 3110–3116.
15. R. D. White, D. V. Bavykin and F. C. Walsh, Nanotechnology, 2012, 23, 065705.
16. E. Abdullayev and Y. Lvov, J. Mater. Chem. B, 2013, 1, 2894–2903.
17. J. Fafard, O. Lyubimova, S. R. Stoyanov, G. K. Dedzo, S. Gusarov, A. Kovalenko and C. Detellier, J. Phys. Chem. C, 2013, 117, 18556–18566.
18. G. Soloperto, F. Conversano, A. Greco, E. Casciaro, A. Ragusa, S. Leporatti, A. Lay-Ekuakille and S. Casciaro, IEEE Trans. Instrum. Meas., 2014, 63, 1423–1430.
19. R. Qi, R. Guo, F. Zheng, H. Liu, J. Yu and X. Shi, Colloids Surf., B, 2013, 110, 148–155.
20. Y. Lvov, W. Wang, L. Zhang and R. Fakhrullin, Adv. Mater., 2016, 28, 1227–1250.
21. Y. Lee, G.-E. Jung, S. J. Cho, K. E. Geckeler and H. Fuchs, Nanoscale, 2013, 5, 8577–8585.
22. H. Li, X. Zhu, H. Zhou and S. Zhong, Colloids Surf., A, 2015, 487, 154–161.
23. S. S. Zargarian, V. Haddadi-Asl and H. Hematpour, J. Nanopart. Res., 2015, 17, 218.
24. Y. Lvov and E. Abdullayev, Prog. Polym. Sci., 2013, 38, 1690–1719.
25. C. Aguzzi, P. Cerezo, C. Viseras and C. Caramella, Appl. Clay Sci., 2007, 36, 22–36.
26. Q. Wang, J. Zhang and A. Wang, Appl. Surf. Sci., 2013, 287, 54–61.
27. Q. Wang, J. Zhang, Y. Zheng and A. Wang, Colloids Surf., B, 2014, 113, 51–58.
28. A. B. Zhang, L. Pan, H. Y. Zhang, S. T. Liu, Y. Ye, M. S. Xia and X. G. Chen, Colloids Surf., A, 2012, 396, 182–188.
29. F. Arcudi, G. Cavallaro, G. Lazzara, M. Massaro, S. Milioto, R. Noto and S. Riela, J. Phys. Chem. C, 2014, 118, 15095–15101.
30. G. Cavallaro, G. Lazzara, S. Milioto, F. Parisi and V. Sanzillo, ACS Appl. Mater. Interfaces, 2014, 6, 606–612.
31. G. Cavallaro, G. Lazzara, S. Milioto and F. Parisi, Langmuir, 2015, 31, 7472–7478.
32. O. Owoseni, E. Nyankson, Y. Zhang, S. J. Adams, J. He, G. L. McPherson, A. Bose, R. B. Gupta and V. T. John, Langmuir, 2014, 30, 13533–13541.
33. W. O. Yah, H. Xu, H. Soejima, W. Ma, Y. Lvov and A. Takahara, J. Am. Chem. Soc., 2012, 134, 12134–12137.
34. E. Abdullayev, A. Joshi, W. Wei, Y. Zhao and Y. Lvov, ACS Nano, 2012, 6, 7216–7226.
35. B. Szczepanik, P. Słomkiewicz, M. Garnuszek, K. Czech, D. Banaś, A. Kubala-Kukuś and I. Stabrawa, J. Mol. Struct., 2015, 1084, 16–22.
36. W. Wei, R. Minullina, E. Abdullayev, R. Fakhrullin, D. Mills and Y. Lvov, RSC Adv., 2014, 4, 488–494.
37. R. Price, B. P. Gaber and R. Y. Lvov, J. Microencapsulation, 2001, 18, 713–722.
38. J. Tully, R. Yendluri and Y. Lvov, Biomacromolecules, 2016, 17, 615–621.
39. L. J. Criscenti, R. T. Cygan, A. S. Kooser and H. K. Moffat, Chem. Mater., 2008, 20, 4682–4693.
40. Y. Gao, Y. Chen, X. Ji, X. He, Q. Yin, Z. Zhang, J. Shi and Y. Li, ACS Nano, 2011, 5, 9788–9798.
41. Y. Chen, Q. Meng, M. Wu, S. Wang, P. Xu, H. Chen, Y. Li, L. Zhang, L. Wang and J. Shi, J. Am. Chem. Soc., 2014, 136, 16326–16334.
42. Y. Lvov, M. DeVilliers and R. Fakhrullin, Expert Opin. Drug Delivery, 2016, 13, 1169271.

---

Footnote

† Electronic supplementary information (ESI) available: Supporting figures. See DOI: 10.1039/c6ra09198a

---