Electrostatically mediated adsorption by nanodiamond and nanocarbon particles

  • Natalie M. Gibson
  • Tzy-Jiun Mark Luo
  • Olga Shenderova
  • Alexey P. Koscheev
  • Donald W. Brenner
Research Paper


Nanodiamond (ND) and other nanocarbon particles are popular platforms for the immobilization of molecular species. In the present research, factors affecting adsorption and desorption of propidium iodide (PI) dye, chosen as a charged molecule model, on ND and sp 2 carbon nanoparticles were studied, with a size ranging from 75 to 4,305 nm. It was found that adsorption of PI molecules, as characterized by ultraviolet–visible spectroscopy, on ND particles is strongly influenced by sorbent-sorbate electrostatic interactions. Different types of NDs with a negative zeta potential were found to adsorb positively charged PI molecules, while no PI adsorption was observed for NDs with a positive zeta potential. The type and density of surface groups of negatively charged NDs greatly influenced the degree and capacity of the PI adsorbed. Ozone-purified NDs had the highest capacity for PI adsorption, due to its greater density of oxygen containing groups, i.e., acid anhydrides and carboxyls, as assessed by TDMS and TOF–SIMS. Single wall nanohorns and carbon onion particles were found to adsorb PI regardless of their zeta potential; this is likely due to π bonding between the aromatic rings of PI and the graphitic surface of the materials and the internal cavity of the horns.


Adsorption Diamonds Surface characterization Surface modification Nanoparticles Drug delivery Biomedical application 


Nanodiamonds (NDs) have recently attracted tremendous attention in the fields of nanotechnology and biomedicine due to their biocompatibility (Schrand et al. 2007b; 2008; Vaijayanthimala et al. 2009; Shenderova 2010), nontoxicity (Liu et al. 2007; Puzyr et al. 2007a; Schrand et al. 2007a; Zhu et al. 2009; Mohan et al. 2010), chemical inertness, and environmental stability (Mohan et al. 2010). NDs have high specific surface areas (SSAs) (typically 300–400 m2/g for NDs produced by detonation synthesis), which can be easily functionalized with a variety of surface groups to enhance binding selectivity toward target molecules (Schrand et al. 2009). Although recent studies of high pressure-high temperature NDs indicate the clearance of NDs from vital organs may be a concern (Yuan et al. 2009), NDs’ inherent features and absence of cytotoxicity (Bakowicz and Mitura 2002; Yu et al. 2005; Chao et. al. 2007) make them a growing area of research. NDs have been considered for a variety of applications including carriers for active molecules (Kossovsky et al. 1995; Grichko et al. 2006), drug delivery agents (Chen et al. 2009; Xing and Dai 2009; Chow et al. 2011), biosensors (Chung et al. 2005; Huang et al. 2004; Raina et al. 2010), and more recently enterosorbents (Gibson et al. 2007, 2009; Puzyr et al. 2010) for toxin binding. Molecular species that have been attached to NDs have included proteins such as obelin (Puzyr et al. 2004), cytochrome c (Huang and Chang 2004) and glycoproteins, ovalbumin and fetuin (Yeap et al. 2008), as well as DNA (Ushizawa et al. 2002; Yang et al. 2002) and fluorescent dyes (Hens et al. 2008; Gibson et al. 2010b). Conjugation of these molecules to NDs was shown to take place via chemisorption (Krueger 2007) or physisorption (Huang and Chang 2004; Chow et al. 2011) for a variety of applications (Schrand et al. 2009).

Conjugation of molecules to NDs is a critical step in the development of successful nanobiotechnology applications. NDs are complex in the fact that they possess many different types of surface groups. Dissociation of these groups in solution produces strong electrostatic potentials, which can dominate the adsorption process. For this reason, this article investigates the adsorption behaviors of several types of NDs for their electrostatic interactions, as well as other adsorption influencing factors, including surface areas and groups. The experiments in this article comprehensively study ND surface properties, including aggregate size, surface charge, specific surface area and functional groups, and their resulting influence on adsorption. To complete a thorough investigation, ND adsorption was compared to adsorption by graphitized nanocarbon particles, such as onion-like carbon (OLC) and single wall nanohorns (SWNH), as well as activated charcoal.

Studies were completed using propidium iodide (PI) as the charged molecule model. PI was selected as it is a well characterized dye, used widely in the field of microbiology to identify dead cells (Schmid et al. 1992; Foglieni et al. 2001; Gitig and Koff 2001). Its strong fluorescence capability allows for straightforward identification and its strong positive charge allows for a clear understanding of electrostatic interactions with ND substrates. Adsorption and desorption of the molecule was characterized using ultraviolet–visible (UV–Vis) spectroscopy. Through subsequent development of the Langmuir isotherm and related transform calculations, the ND substrate that provides the largest adsorption capacity, strongest binding, and greatest control in releasing PI could be identified. Maximum capacities obtained from transform calculations further allowed for estimates of the specific surface area (SSA) covered by the adsorbate for select NDs. Surface characterization using TDMS and TOF–SIMS were also completed to help determine dominant surface groups on the NDs.

Materials and experimental methods

Types of carbon species

An extensive selection of carbon species were used to understand not just the effects of surface charge on PI adsorption, but also the magnitude of that charge, the particle size, surface configuration, and sp 2 versus sp 3 surface hybridization on binding. The complete list of species is given in Table 1. Four classes of carbon species were used: (a) nanodiamonds, (b) OLC nanoparticles, (c) SWNHs, and (d) activated charcoal. NDs, the prime focus of this article, were further subdivided into two classes, those carrying a positive charge and those carrying a negative charge, which depends on manufacturing and treatment processes. For these NDs the role of particle size and surface treatment was also explored. OLC, obtained by annealing NDs, were classified based on annealing temperature and initial type of ND used. Both factors also affect the dispersivity in water and the amount of sp2 coverage on the sorbent.
Table 1

Size (diameter for ND and OLC, nm), zeta potentials (ZP, mV), and processing methods of all nanocarbon aggregates used in the adsorption experiments


ZP (mV)

Size (nm)



46.8 ± 0.9

185.6 ± 0.25

CrO3 in H2SO4; NaOH + H2O2; ion-exchange resin; FC


−44.1 ± 0.66

260.1 ± 1.75

Singlet O in NaOH; HNO3;


−49.3 ± 1.9

75.0 ± 0.35

Singlet O in NaOH; HNO3; NaCl; FC


−48.5 ± 0.2

139.2 ± 0.59

Soot oxidized with Ozone


−51.2 ± 0.5

80.5 ± 0.99

Soot oxidized with Ozone; FC


47.1 ± 0.65

189.2 ± 0.92

ND+180 heated in vacuum at 1,600 K


13.9 ± 0.15

4305.0 ± 252

Nanocraft, Inc. Renton, WA

Activated charcoal


20/40 mesh

Supelco Bellefonte, PA

FC Fractionation by centrifugation


NDs were produced at the manufacturer’s site through detonation using a mixture of trinitrotoluene and 1,3,5-trinitro-1,3,5-s-triazine, which results in “detonation nanodiamonds” (DNDs). DNDs were purified from non-diamond carbon in the soot using liquid- or gas-phase oxidizers. With purification, the removal of metallic impurities and oxidation of non-diamond carbon was achieved. Additional processing methods were used to further reduce the non-diamond carbon content and provide intentional alternation to the ND’s surface group content and composition. The sample ND+180 (“New Technologies,” Chelyabinsk, Russia) served as the standard positively charged ND. The material contained an incombustible impurity content of 0.6 wt%. Details of the detonation and post-synthesis processing of NDs were recently reported (Gibson et al. 2009). Several negatively charged NDs were used that included NDs purified from detonation soot under wet-phase (NDW) and gas-phase (NDG) techniques. NDW (Real-Dzerzinsk, LTD, Russia) was purified from soot using singlet oxygen in NaOH and HNO3 (Puzyr 2003, 2009; Gibson et al. 2009). A portion of NDW was treated in NaCl solution and size fractionated through centrifugation to achieve NDW75. NDG (“New Technologies”) was purified by oxidizing soot with ozone at 200 °C. NDG was also size fractionated by centrifugation to achieve NDG80, which had a smaller, more uniform particle size.

Onion-like carbon, single wall nanohorns, and activated charcoal

The OLC sample (Institute of Catalysis, Novosibirsk, Russia), which was produced by annealing ND+180 at a temperature of 1600 K, was chosen due to its superior colloidal stability based on a high zeta potential (ZP) and small size characteristics. SWNHs (Nanocraft, Inc., Renton, WA.) and activated charcoal (Supelco Inc., 20/40 mesh size) were used as received.

Hydrosol preparation

All ND suspensions were prepared using deionized (DI) water (pH = 5.8 and water resistivity = 18 MΩ cm) followed by sonication with a Cole-Parmer 750-Watt ultrasonic homogenizer (EW-04711-60). Its tapered titanium horn (3 mm diameter) was directly immersed into the polypropylene centrifuge tubes containing the sample suspension. Sonication occurred for 2–4 min at an output power of 10 W and output intensity of 100 W/cm2. All solutions were prepared to a 0.1 wt% concentration for ZP, size, and pH measurements, while a 0.5 wt% concentration was prepared for adsorption measurements. The ND colloids had a natural pH ranging from 5 to 6, related to the 0.1 wt% suspension concentration. OLC and SWNH suspensions were prepared in the same manner but with sonication times of 8–10 min. The natural pH of OLC was measured as 5.7; SWNHs were more acidic with a pH close to 4.6.

Particle size and zeta potential measurements

Particle size was measured using the N5 submicron particle size analyzer (Beckman Coulter Inc.) and the ZetaSizer NanoZS (Malvern Instruments). To prepare samples for particle size measurements, ND suspensions were diluted by adding approximately 10 μL of the concentrated suspension to 3 mL of DI water. Each sample was measured three times and the average particle sizes were calculated based on unimodal intensity. It is worth noting that the ZetaSizer is able to measure particle sizes accurately without diluting the suspension below 0.1 wt%. ZP measurements were completed solely on the ZetaSizer by using 0.1 wt% ND suspensions that were transferred into specialized Malvern zeta cells. Again, three measurements were taken at room temperature; averaged results are reported.

Adsorption studies

All nanocarbon types

PI from Sigma Aldrich was dissolved in DI water to prepare a 0.07 wt% stock solution. All ND adsorption experiments were carried out first by mixing the desired concentration of PI solution and a 0.5 wt% ND suspension at a 7:3 volume ratio. Initial adsorption experiments used final PI concentrations of 0.01 and 0.03 wt%. A ND suspension with 0 wt% PI was used as the control sample. The mixtures were incubated on an orbital shaker at room temperature for 30 min followed by centrifugation (14,000 rpm, 25 min). The remaining PI in each supernatant was measured on a UV–Vis spectrophotometer (Lambda 35, Perkin Elmer). Due to ND’s very high colloidal stability, there is a remaining ND in the supernatant after centrifugation. To eliminate these effects, the spectra of these residual NDs were subtracted from all ND–PI spectra that used PI concentrations below 0.015 wt%. PI concentrations above this concentration caused complete coagulation of the residual NDs. The percentage of unbound PI was determined based on a PI standard absorption curve and this concentration was converted to ND adsorption capacity (μg PI/mg ND). Because SWNH and OLC samples do not form a pellet during centrifugation, microcentrifugal regenerated cellulose filters (Sigma-Aldrich, MWCO 100 kDa) were used to separate nanocarbons from unbound PI. To reduce non-specific binding of PI on the cellulose membrane, all filters were washed with 0.04 wt% PI and DI water twice before performing adsorption experiments.

Activated charcoal

PI is an effective and widely used dye, but it also carries many harmful health hazards as it is a possible carcinogen and mutagen (Waring 1965; Menozzi et al. 1990). For disposal, we compare NDs to traditional techniques that require PI to be poured through activated charcoal and filter paper followed by incineration of the charcoal (Waring 1965) or charcoal filters for simpler, faster methods of disposal (Menozzi et al. 1990). To complete experiments, activated charcoal (6 or 3.5 mg) was added to 1 mL PI suspensions of 0.005–0.025 wt% concentrations. Individual mixtures were vortexed for 1 min and then placed on the orbital shaker for 15 min, 1, 24, and 28 h. PI samples in the absence of activated charcoal were prepared for control samples. After incubation, charcoal was removed and the amount of unbound PI was measured spectroscopically.

Langmuir isotherm

Langmuir isotherms were constructed for the ND+180, NDW75, and NDG80 samples by including adsorption studies at several different PI concentrations ranging from 0.0025 to 0.07 wt%. To obtain the Langmuir curve adsorption percentages were converted to capacities as micrograms of PI adsorbed per milligram of sorbent and plotted against the concentration of equilibrium PI (μg/mL). From this data, the maximum capacity (Q max) and the binding constants (K b) of the sorbents were extracted by linear regression curve fitting using four transform equations: Eadie–Hofstee, Lineweaver–Burk, reciprocal line, and Scratchard.

Specific surface area

SSAs (surface area per gram of ND) covered by PI were calculated based on Q max, obtained through transform equations, and the cross-sectional area of the adsorbed molecule (Å2). The cross-sectional area of PI (128.7 Å2) was estimated from a PI molecule whose structure was determined via energy by energy minimization using the PM6 semi-empirical total energy method in MOPAC.1 The relaxed structure was added with the Connolly surface using a probe size of 1.5 Å on the Visual Molecular Dynamics program (Humphrey et al. 1996). By outputting the image to ImageJ the cross-sectional value was determined. Results based on the Langmuir model were compared to true SSAs measured by the Brunauer, Emmett and Teller (BET) method, which used nitrogen adsorption at 77 K (Quadrasorb from Quantachrome).

Desorption studies

Desorption studies were conducted to detect the controllability of releasing PI from the ND substrates. ND+180, NDW75, and NDG80 pellets, saved from previous adsorption experiments, were used. In the first stage individual pellets were redispersed into 1 mL of DI water followed by vortexing (2 min) and placing on an orbital shaker (20 min). Redispersed samples were then centrifuged (14,000 rpm, 25 min) and supernatants were collected for UV–Vis detection. In the second stage redispersion occurred in 5 mg/mL NaCl solution or 5 mg/mL CaCl2 solution with all remaining steps the same. This stage was repeated for a second time to have a total of 3 redispersion cycles for each ND sample.

Surface characterization

Mass spectrometry

A portion (4.4 ± 0.1 mg) of ND powder was placed into the Ni-envelope (pre-cleaned ultrasonically in acetone and water and dried) and introduced in the high-temperature vacuum oven. The sample was then degassed under evacuation by ion pump for one day at room temperature. The residual pressure at the end of degassing was in the range of 10−7 Torr. TDMS analysis was performed by measuring the mass spectra of gases released from the sample under programmed heating with constant rate 10 °C/min up to 1,100 °C. The mass spectrometer was a quadrupole MX7304 with electron multiplier with mass range 10–120 amu (up to 700 °C) and 2–50 amu (700–1,150 °C). MS scan rate was 1 min per scan (or 10 °C per scan). The evolved gases were pumped permanently with the rate ~1 L/s (relative to nitrogen). In this case the measured partial pressures of gases were proportional to desorption rates at every moment. The total pressure in the vacuum chamber during heating was measured by an ionization gauge. The obtained mass spectra were used to calculate the intensities of characteristic ion fragments at different temperatures, their temperature profiles, and the relative amounts of released species.

TOF–SIMS was used in a complementary analysis. The instrument was equipped with a Bi n m+ (n = 1–5, m = 1, 2) liquid metal ion column. The analysis chamber pressure was maintained below 5.0 × 10−9 mbar to avoid surface contamination. The primary Bi+ ions had a kinetic energy of 25 keV and were contained within a ~10 μm diameter probe beam. For high mass resolution spectra, the Bi+ ion beam was rastered across a 500 × 500 μm2 area divided into 128 × 128 pixels. The electron flood gun was used to neutralize the charge accumulated on the sample surface while acquiring the spectrum. The total accumulated primary ion dose used to acquire spectra was less than 1 × 1012 ions/cm2, which is within the static SIMS regime. The secondary ions were extracted at 2 keV and mass analyzed using a TOF mass spectrometer. Secondary ions were post accelerated to 10 keV before impacting the detector. At least three areas of each sample were analyzed. Positive secondary ion mass spectra were calibrated using H+, C+, CH+, CH2 +, CH3 +, C2H3 +, C2H5 +, C4H9 + and negative spectra were calibrated using C, O, OH, and various C n secondary ions. Peak intensities were reproducible to within 10% from scan to scan. Mass resolution of ~5,000 MM (50% valley definition) at m/z 29 was obtained throughout. Results from the TOF–SIMS studies are described in the Online Resource.


Size and zeta potential

Particle size measurements showed that different processing and treatment methods had largely impacted the aggregate size of the NDs. The size of the primary ND particles is known to be between 4 and 5 nm; however, the primary particles aggregate during synthesis (Chiganova 2000). While beads milling could have been used to reduce the aggregate size back down to the primary size range, for the purposes of this research significant size reduction was not desired, due to the high colloidal stability of small NDs, which makes them difficult to pellet. PI additions at high concentrations to ND suspensions cause immediate and rapid precipitation; this may cause additional PI confinement within ND agglomerates. Since aggregated NDs contain pores, a possible role of the sizes of aggregates (and contribution of the pores network) was another object of the research. The pores may promote internal diffusion and trapping of PI molecules within the aggregates. Surface morphology and cavities of ND aggregates were previously observed through HRTEM and SEM imaging (Petrov et al. 2007; Turner et al. 2009; Shenderova 2010).

ZP measurements indicated that the influence of particle size on the ZP is insignificant. When comparing the non-fractionated NDG to NDW75 it was shown that NDG was nearly twice the size of NDW75 (139 versus 75 nm); however, the ZPs of these two samples were virtually the same (−49 mV). Such a result shows the importance of surface chemistry; NDG has a greater density of negative surface groups which gives rise to higher ZPs. Nevertheless, NDs, processed under identical conditions, can be manipulated to have slightly higher ZP values by decreasing their size through fractionation by centrifugation (Fig. 1).
Fig. 1

Adsorption capacities (μg/mg) of 0.01 wt% and 0.03 wt% PI in DI water on various nanocarbon substrates (1.5 mg/mL)

Classes of commercial NDs, including NDs of the present study, demonstrate positive or negative zeta potential values depending on the process used for detonation soot oxidation (Table 1). NDs processed with soot oxidation using either singlet oxygen in NaOH, or purified with ozone in a gas phase have negative zeta potentials, indicating deep oxidation with formation of acidic groups on the surface. DND oxidized from soot with CrO3/H2SO4, NaOH/H2O2, less strong oxidizers, have positive zeta potentials and prevalence of basic groups (Schrand et al. 2009).

Adsorption for all samples at 0.01% and 0.03% PI


The adsorption studies show drastic differences in adsorption capacities for each ND type (Fig. 1). Interestingly, NDs themselves show notable changes in their adsorption capacity depending on a method of their purification from soot. The data indicates that positively charged NDs do not adsorb the PI, which is also positively charged in DI water (Fig. 2a), due to electrostatic repulsion. However, all negatively charged NDs showed adsorption of PI, though the amount each ND adsorbed was drastically different depending on the purification from soot procedures. As an example, an adsorption spectrum of NDW75 is illustrated in Fig. 2b. It was also concluded that ZP is not a direct indicator of adsorption capacities, as seen with NDW and NDW75. Both NDs had comparable ZPs but vary drastically in the amount of dye adsorbed. More examples of these phenomena were provided in previous work (Gibson et al. 2010a, b). It is understood that surface chemistry of the ND has the greatest influences on adsorption. However, adsorption capacity of NDs can be increased by using fractions of ND with smaller sizes obtained, for example, by centrifugation. Using NDs with smaller average aggregate sizes resulted in increased ZPs and subsequently increased adsorption quantities. This result can be seen for both the NDW and NDG series (Fig. 1).
Fig. 2

UV–Vis absorbance spectra of ND–PI supernatants after incubation and centrifugation as compared to the reference PI concentration (PI + water): ND+180 with 0.009 wt% PI (a) and NDW75 with 0.01 wt% PI (b)


Positive ZPs exist for SWNHs (+14 mV) and OLC (+47 mV), yet unlike observations with positively charged NDs these carbon based particles show adsorption of PI, although the amount of adsorption is low. Both samples adsorbed roughly equal amounts of PI, suggesting that adsorption extends beyond surface functional groups. Annealing ND+180 in vacuum at high temperatures generates the sp 2 surface of the OLC derivative. The sp 2 graphitic surface of the OLC and the SWNH has a greater, though not excellent, ability to adsorb the positively charged molecule likely due to π–π interactions between the aromatic rings of PI and the graphitic surface of the materials. The SWNH also has an internal cavity accessible to PI, which allows entrapment of the molecule and contributes to its increased adsorption characteristics. In each case defects or non-uniformities on the surface may also play a role in adsorption of the like-charged molecule.

Activated charcoal

Activated charcoal is a standard in adsorption applications and is used specifically as a PI adsorbent in disposal protocols (Menozzi et al. 1990; Waring 1965); therefore, it is important to compare its results to that of the other carbon structures listed above. It was found that the activated charcoal adsorbed PI at a much slower rate than NDs even though the charcoal concentration was 6 mg/mL as compared to 1.5 mg/mL of ND samples (Fig. 3). For activated charcoal, the amount of adsorbed PI linearly increased from 15 min incubation to 24 h incubation, while for ND samples the adsorption reached 82% in 30 min and only increased by 5% at the end of the 24 h period. Therefore, it was concluded that 30 min incubation times for all ND samples are sufficient. Furthermore, adsorption rates for activated charcoal are inferior to NDs, due to activated charcoal requiring time for diffusion.
Fig. 3

Total percentage of 0.005 wt% PI adsorbed on activated charcoal (6 mg/mL) as compared to NDW75, NDG and NDG80 (1.5 mg/mL) substrates during 0.25, 0.5, 1, and 24 h incubation times

In another experiment, the amount of activated charcoal was reduced by almost half (3.5 mg/mL) and results of the PI adsorption were compared to the previously used 6 mg/mL. The experiment was completed using 0.025 wt% PI with incubation times of 15 min, 1, 24, and 28 h. The two samples performed similarly until 1 h, after which the 6 mg/mL sample began adsorbing more PI than the 3 mg/mL sample. Remarkably, a noticeable amount of adsorption was seen in the time interval between 24 and 28 h and was significantly higher for the higher concentration of charcoal. Overall, the amount of PI adsorbed over 28 h was about 44% for 6 mg/mL and 21% for 3.5 mg/mL of activated charcoal. For both studies the total amount of PI adsorbed by activated charcoal, which had a SSA of 1040 m2/g, adsorbed far less PI than NDs, consequently, opening up new possible applications for NDs.

Langmuir studies

Langmuir isotherm curves of NDW75, NDG, and NDG80 (Fig. 4) deviate from typical isotherms that demonstrate a plateau, indicating saturation of the substrate’s adsorption sites. In all cases, it was initially observed that as the concentration of PI increased there still remained available active sites for other PI molecules to bind, denoted by the gradual slope early on in the curve. However, at higher concentrations of PI a turning point exists where some sites become free, corresponding to desorption of the dye (“valleys” on the curves, Fig. 4). Such a phenomenon, previously described by Giles et al. (1974) is attributed to desorption of dye molecules due to micelle formations, as a result of dyes having a stronger affinity for themselves than to the adsorbing substrate. Concentrations of PI used in this experiment were very high making micelle formation plausible. However, detection of PI at their micelle formation is still needed. The NDW75 and NDG80 samples with smaller sizes show an adsorption peak at PI concentrations of nearly 150 μg/mL whereas the NDG curve shows a continuous increase that does not reach a maximum until beyond 150 μg/mL. As the concentration of PI is further increased it is seen that the PI molecules are again being adsorbed. Such behavior illustrates the conflicting ability for the material to complete the first monolayer, as the PI molecules are adsorbed, desorbed, and reabsorbed onto the ND substrate. Giles’ theory suggests that this sinusoidal pattern should continue if dye concentrations were increased further. The theory additionally states that the exhibited behavior, seen in certain dyes and detergents, may be caused by the presence of substrate surface impurities.
Fig. 4

Langmuir isotherm plot for adsorption PI to NDW75, NDG, and NDG80

The maximum capacity and binding constants of all three NDs were calculated using the four transforms listed in “Langmuir isotherm” section. Four transforms were selected as each shows different error sensitivities and have biases toward fitting data at low or high concentration ranges. Data beyond the PI concentration corresponding to the adsorption/desorption turning point were excluded because linearity must be achieved to calculate the maximum capacity and binding constants. The transform calculations showed the highest capacity for NDG80 (~200 μg/mg) followed by NDW75 (Table 2). As previously mentioned, both NDs have roughly the same aggregate size and ZPs in water. To understand the 25% difference in the sorption capacity, detailed surface characterization of the ND samples was performed (“Surface characterization” section). Besides surface group differences, internal pores, which contribute to a larger surface-to-volume ratio may contribute to NDG80s higher maximum capacity. Therefore, pores size and total volumes were also characterized with BET methods (Table 3). As seen, NDG80 has a larger average pore size and nearly double the pore volume than NDW75, which helps explain the larger loading capacity.
Table 2

Transforms used in the calculation of maximum capacity and binding constant for PI on (a) NDW75, (b) NDG, and (c) NDG80


Max capacity (μg/mg)

K b (mg/μg)

r 2










 Reciprocal line





















 Reciprocal line





















 Reciprocal line












Table 3

Comparisons of calculated SSAs, m2/g, covered by PI based on maximum capacity computations, in relationship to the true SSA, m2/g (BET)


SSA from Langmuira

SSA from BET

Average pore size (nm)

Pore volume (cc/g)
















Average pore size, nm, and pore volume were also obtained through BET measurements

Specific surface area = (Qmax × N × A)/MW PI

N A Avogadro’s number, A cross-sectional area of the adsorbed molecule (Å2), MW PI molecular weight of the adsorbed molecule

aCalculations carried out using the following relationships

Despite differences in maximum capacities, the binding constant of both NDs are similar (Table 2). The low constants indicate weak binding and explain the observed desorption at higher PI concentrations on the isotherms. If the binding strength of the dye to the ND was increased desorption behavior at higher concentrations could potentially be reduced or eliminated. Interestingly, NDG, which possessed the lowest maximum capacity, had the highest binding constant. Its smaller surface area, as a result of larger aggregate sizes and pore size, indicates the adsorption mechanism is dominated by specific and strong binding sites on the surface while the small pores on NDW75 and NDG80 might help to contribute additional but weak binding mechanisms.

Specific surface areas

SSAs covered by the dye reveal that NDG80 had the largest available area for PI binding (Table 3) though NDG80 and NDW75 had similar aggregate sizes. As expected, comparisons between NDG and NDG80 indicate that as aggregate particle sizes decrease the SSA covered by PI increases. Compared to the true SSA found by BET measurements, approximately 55–65% of SSA is covered by PI, indicating PI adsorption on ND surfaces is highly selective as compared to the non-specific adsorption performed by the BET method. NDG80 exhibited the highest areas covered by PI from the Langmuir calculations (232.29 m2/g) as well as SSA from BET measurements (426.4 m2/g), which follows the previous trend showing it also had the highest capacity. Discrepancies between the measured and calculated areas can be attributed to PI having preferred binding sites on each ND where once these sites are filled no more PI is adsorbed. In addition, during adsorption PI may lay on the ND surface in preferred orientations that may not allow for ordered, high density packing of PI. Inaccuracies in the estimated size of the PI may also contribute to this error.

Desorption experiments

The three ND samples used in the adsorption experiments were also examined for desorption of PI. NDG80 and NDG sample pellets showed no release of PI under the presence of water. NDW75, however, showed approximately 5% of the dye was removed with water additions (Fig. 5). This result is explained by the weaker calculated binding constants of NDW75 as compared to the ozonated samples.
Fig. 5

Total percentage of PI released from ND substrates, NDW75, NDG80 and NDG, during three centrifugation cycles in water (Ctg 1) followed by redispersion and centrifugation in NaCl or CaCl2 (Ctg 2 and 3)

Samples under the influence of NaCl and CaCl2 all showed PI desorption. NDG80 pellets showed ~23% release under the influence of NaCl and a 25% release under the influence of CaCl2. NDG showed slightly lower desorption rates at 19% release with NaCl and 21% with CaCl2, which is attributed to the somewhat higher binding constant. For NDW75 an additional 21% with NaCl and 23% with CaCl2 was released after the 5% in water. NDW75 pellets that were not first redistributed in water showed a 27% PI release when introduced to NaCl and almost a 29% PI release with CaCl2. Upon subsequent redispersion, all samples showed similar PI release as compared to the first salt redispersion; leaving the total amount of PI desorbed after all three rounds as 45 and 49% in NaCl and CaCl2 for NDG80 and 33 and 41% in NaCl and CaCl2 for NDG. For NDW75 the total amount of PI released was 37 and 40% in NaCl and CaCl2, taking into account the 5% initially washed out with DI water. The highest desorption amounts exhibited by NDG80 were attributed to the weak binding constant and higher density of surface charges, i.e., surface groups, which rely on electrostatic interactions for binding; the success of groups are impaired by salt additions. Results also show that NaCl and CaCl2 are equally effective on removing PI from NDW75, while CaCl2 removed more PI from NDG and NDG80 samples than NaCl. This indicates their negatively charged functional groups exist on the ozonated surface that preferentially binds to Ca2+.

Surface characterization

Samples NDW75 and NDG have similar average aggregate sizes and zeta potential values. However, the NDG sample possesses higher sorption capacity, possibly originating from surface group differences. It has previously been observed that NDG is the most acidic of all currently known ND due to its larger concentration of oxygen containing groups stemming from treatment in ozone (Petrov et al. 2007; Cunningham et al. 2008). In the present study, mass spectrometry techniques, which both included TOF–SIMS and TDMS, were used to reveal details of the surface groups on the ND samples.

The main differences between NDG and NDW75 are the shape and intensities of TDMS profiles of H2O, CO CO2, and CxHy (spectra for CO and CO2 are shown in Fig. 6). The measured profiles allowed the estimation of the amounts of desorbed species, calculated as the area under the curve. The total amount of CO and CO2 was highest for NDG indicating surface oxidation was highest for this sample. The complex shape of CO2 and CO release curves indicates the presence of several types of C–O bonds on the ND surface giving rise to CO and CO2 evolution under heating at different temperatures. According to the published data on the thermal decomposition of oxygen containing species on the surface of diamond and carbon materials one can suggest the next simplified explanation of observed release pattern: region 1 (below 400 °C, CO2 only)–carboxyl groups; region 2 (500–600 °C, CO2 and CO)–carboxylic anhydride; region 3 (600–700 °C)–lactone (CO2 release) and hydroxylic groups (CO release); region 4 (above 700 °C, CO only)–ether and/or carbonyl groups. It was concluded that the majority groups on NDG are carboxylic acid anhydride species (Petrov et al. 2007), while carboxylic groups prevail on the surface of NDW75 sample. TOF–SIMS revealed the presence of a Na+ peak in sample NDW75 (Online Resource Fig. 1). The presence of Na+ originates from the treatment procedures on the ND, which used NaCl with the purpose of increased colloidal stability at the vendor site (Puzyr et al. 2007a, b; Gibson et al. 2009). The presence of positively charged Na+ ions on the ND surface can be one of the reasons for the lower adsorption capacity of this sample, since these sites became unavailable for PI adsorption.
Fig. 6

TDMS profiles CO and CO2 desorbed from NDG, ND+180 and NDW75

TOF–SIMS and TDMS were also performed for ND+180 ND with positive zeta potential (see Online Resource for results on TOF–SIMS). ND+180 have a much lower level of surface oxidation as compared to NDW75 and NDG (Fig. 6). Its concentration of carboxyl groups is very low, with the main oxygenated surface species belonging to carbonyl groups. In addition, ND+180 have a large presence of hydrocarbon residues (Fig. 2, Online Resource) probably due to ion-exchange treatment, which is a part of the procedure at the vendor site. Their presence may also inhibit high adsorption capacities in cases where electrostatic charges are not the dominant mechanism of adsorption.


Results of the present study indicate that electrostatic interaction between NDs and molecular species have strong influence on the binding capacity of NDs. When considering NDs as drug delivery platforms, relative charges on the NDs, surface group composition and aggregate size should be taken into account, as well as the molecular species being adsorbed. Experiments show NDs having a like-charge to PI were not able to adsorb the molecule, though SWNHs and OLC could due to π–π interactions between the aromatic rings of PI and the graphitic surface of the nanocarbons and the nanohorns’ internal cavities. While all negatively charged NDs showed uptake, size and surface functionalization differences significantly determined the total loading capacity of the adsorbed molecule. In the specific case of PI adsorption, smallest sized ozone-purified NDs (NDG75) were shown to have the highest loading capacity with fast adsorption rates, which, on both accounts, surpassed that of activated charcoal. Furthermore, studies also demonstrate that PI can be controllably released from NDs under the influence of salt solutions.

Surface characterization using TOF–SIMS and TDMS indicate NDG75 had higher density of oxygen containing groups, including a prevalence of acid anhydrides, whereas ND oxidized using atomic oxygen (NDW75) showed predominately carboxylic groups. In addition, as a result of earlier treatments, Na+ ions were identified on the NDW75 surface. The differences may lead to preferential binding sites or adsorption orientations for the PI on NDs. These assumptions were further examined through comparisons of BET measurements to calculations of SSA covered by the adsorbate. By comparison, positively charged NDs show more complex surfaces with an abundance of carbonyls and hydrocarbons that prevent adsorption of the studied molecule.


  1. 1.

    MOPAC2009, Stewart, J.J.P. Stewart Computational Chemistry, Version 8.318 M web:



This research is supported by the Materials World Network program of the National Science Foundation under Grant No DMR-0602906. O.S. acknowledges the partial support through Air Force Office of Scientific Research under grant N66001-04-1-8933. In addition, we thank V. Kuznetsov, of the Boreskov Institute of Catalysis, Novosibirsk for providing onion-like carbon samples, V. Vorobyev for providing NDW samples, Yury Gogotsi, Department of Materials Science and Engineering and Nanomaterials Group at Drexel University, for the BET analysis, Elaine Chuanzhen Zhou at the Analytical Instrumentation Facility for TOF–SIMS experiments and Zachary Fitzgerald, Department of Materials Science and Engineering at North Carolina State University for his modeling of the PI molecule. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.

Supplementary material

11051_2011_700_MOESM1_ESM.pdf (58 kb)
Supplementary material 1 (PDF 58 kb)


  1. Bakowicz K, Mitura S (2002) Biocompatibility of NCD. J Wide Bandgap Mater 9(4):261–272CrossRefGoogle Scholar
  2. Chao JI, Perevedentseva E, Chung PH, Liu KK, Cheng CY, Chang CC, Cheng CL (2007) Nanometer-sized diamond particle as a probe for biolabeling. Biophys J 93(6):2199–2208CrossRefGoogle Scholar
  3. Chen M, Pierstorff ED, Lam R, Li SY, Huang H, Osawa E, Ho D (2009) Nanodiamond-mediated delivery of water-insoluble therapeutics. ACS Nano 3(7):2016–2022CrossRefGoogle Scholar
  4. Chiganova GA (2000) Aggregation of particles in ultradispersed diamond hydrosols. Colloid J 62:238–243Google Scholar
  5. Chow EK, Zhang X-Q, Chen M, Lam R, Robinson E, Huang H, Schaffer D, Osawa E, Goga A, Ho D (2011) Nanodiamond therapeutic delivery agents mediate enhanced chemoresistant tumor treatment. Sci Transl Med 3(73):1–11CrossRefGoogle Scholar
  6. Chung PH, Perevedentseva E, Tu JS, Chang CC, Cheng CL (2005) Spectroscopic study of bio-functionalized nanodiamonds. Diam Relat Mater 15(4–8):622–625CrossRefGoogle Scholar
  7. Cunningham G, Panich AM, Shames AI, Petrov I, Shenderova O (2008) Ozone-modified detonation nanodiamonds. Diam Relat Mater 17(4–5):650–654CrossRefGoogle Scholar
  8. Foglieni C, Meoni C, Davalli AM (2001) Fluorescent dyes for cell viability: an application on prefixed conditions. Histochem Cell Biol 115(3):223–229Google Scholar
  9. Gibson N, Shenderova O, Puzyr A, Purtov K, Grichko V, Luo TJM, Fitzgerald Z, Bondar V, Brenner D (2007) Nanodiamonds for detoxification. In: NSTI nanotechnology conference and trade show–NSTI nanotech technical proceedings, vol 2, Nanotech, Santa Clara, pp. 713–716Google Scholar
  10. Gibson N, Shenderova O, Luo TJM, Moseenkov S, Bondar V, Puzyr A, Purtov K, Fitzgerald Z, Brenner DW (2009) Colloidal stability of modified nanodiamond particles. Diam Relat Mater 18(4):262–620CrossRefGoogle Scholar
  11. Gibson NM, Luo TJM, Shenderova O, Choi YJ, Brenner DW (2010a) Modified nanodiamonds for adsorption of propidium iodide and aflatoxin. MRS Fall 2009 Proceedings, available online, Paper #1236-SS09-05Google Scholar
  12. Gibson NM, Luo TJM, Shenderova O, Choi YJ, Fitzgerald Z, Brenner DW (2010b) Fluorescent dye adsorption on nanocarbon substrates through electrostatic. Interactions 19(2–3):234–237Google Scholar
  13. Giles CH, Smith D, Huitson A (1974) A general treatment and classification of the solute adsorption isotherm. I. Theoretical. J Colloid Interf Sci 47(3):755–765CrossRefGoogle Scholar
  14. Gitig DM, Koff A (2001) Cdk pathway: cyclin-dependent kinases and cyclin-dependent kinase inhibitors. Mol Biotechnol 19(2):179–188CrossRefGoogle Scholar
  15. Grichko V, Grishko V, Shenderova O (2006) Nanodiamond bullets and their biological targets. NanoBioTechnol 2(1–2):1294–1551Google Scholar
  16. Hens S, Cunningham G, Tyler T, Moseenkov S, Kuznetsov V, Shenderova O (2008) Nanodiamond bioconjugate probes and their collection by electrophoresis. Diam Relat Mater 17(11):1858–1866CrossRefGoogle Scholar
  17. Huang LCL, Chang HC (2004) Adsorption and immobilization of cytochrome c on nanodiamonds. Langmuir 20(14):5870–5884CrossRefGoogle Scholar
  18. Huang TS, Tzeng Y, Liu YK, Chen YC, Walker KR, Guntupalli R, Liu C (2004) Immobilization of antibodies and bacterial binding on nanodiamond and carbon nanotubes for biosensor applications. Diam Relat Mater 13(4–8):1098–1102CrossRefGoogle Scholar
  19. Humphrey W, Dalke A, Schulten K (1996) VMD: visual molecular dynamics. J Mol Graph 14(1):33–38CrossRefGoogle Scholar
  20. Kossovsky N, Gelman A, Hnatyszyn HJ, Rajguru A, Garrell RL, Torbati S, Freitas SSF, Chow MG (1995) Surface-modified diamond nanoparticles as antigen delivery vehicles. Bioconjugugate Chem 6(5):507–511CrossRefGoogle Scholar
  21. Krueger A (2007) New carbon materials: biological applications of functionalized nanodiamond materials. Chem A Eur J 14(5):1382–1390CrossRefGoogle Scholar
  22. Liu KK, Cheng CL, Chang CC, Chao JI (2007) Biocompatible and detectable carboxylated nanodiamond on human cell. Nanotechnology 18(32):325102CrossRefGoogle Scholar
  23. Menozzi FD, Michel A, Pora H, Miller AOA (1990) Absorption method for rapid decontamination of solutions of ethidium bromide and propidium iodide. Chromatographia 29(3–4):167–169CrossRefGoogle Scholar
  24. Mohan N, Chen CS, Hsieh HH, Wu YC, Chang HC (2010) In vivo imaging and toxicity assessments of fluorescent nanodiamonds in Caenorhabditis elegans. Nano Lett 10(9):3692–3699CrossRefGoogle Scholar
  25. Petrov I, Shenderova O, Grishko V, Grichko V, Tyler T, Cunningham G, McGuire G (2007) Detonation nanodiamonds simultaneously purified and modified by gas treatment. Diam Relat Mater 16(12):2098–2103CrossRefGoogle Scholar
  26. Puzyr AP, Bondar VS, (2003) Method of production of nanodiamonds of explosive synthesis with an increased colloidal stability, St. PetersburgGoogle Scholar
  27. Puzyr AP, Pozdniakova IO, Bondar VS (2004) Design of a luminescent biochip with nanodiamonds and bacterial luciferase. Phys Solid State 46(4):761–763CrossRefGoogle Scholar
  28. Puzyr AP, Baron AV, Purtov KV, Bortnikov EV, Skobelev NN, Mogilnaya OA, Bondar VS (2007a) Nanodiamonds with novel properties: a biological study. Diam Relat Mater 16(12):2124–2128CrossRefGoogle Scholar
  29. Puzyr AP, Purtov KV, Shenderova OA, Luo M, Brenner DW, Bondar VS (2007b) The adsorption of aflatoxin B1 by detonation-synthesis nanodiamonds. Doklady Biochem Biophys 417(1):299–301CrossRefGoogle Scholar
  30. Puzyr AP, Burov AE, Bondar VS, Trusov YN (2010) Neutralization of aflatoxin B1 by ozone treatment and adsorption by nanodiamonds. Nanotechnol Russ 5(1–2):137–141CrossRefGoogle Scholar
  31. Raina S, Kang WP, Davidson JL (2010) Nanodiamond macro- and microelectrode array bio-sensor. IEEE Sens Conf 2009:1780–1783Google Scholar
  32. Schmid I, Krall WJ, Uittenbogaart CH, Braun J, Giorgi V (1992) Dead cell discrimination with 7-amino-actinomycin D in combination with dual color immunofluorescence in single laser flow cytometry. Cytom Part A 13(2):204–208CrossRefGoogle Scholar
  33. Schrand AM, Huang HJ, Carlson C, Schlager J, Osawa E, Hussain S, Dai L (2007a) Are diamond nanoparticles cytotoxic? J Phys Chem B 111(1):2–7CrossRefGoogle Scholar
  34. Schrand AM, Dai L, Schlager JJ, Hussain SM, Osawa E (2007b) Differential biocompatibility of carbon nanotubes and nanodiamonds. Diam Relat Mater 16(2):2118–2123CrossRefGoogle Scholar
  35. Schrand AM, Johnson J, Dai L, Hussain SM, Schlager JJ, Zhu L, Hong Y, Osawa E (2008) Cytotoxicity and genotoxicity of carbon nanomaterials. In: Webster T (ed) Safety of nanoparticles: from manufacturing to clinical applications. Springer, ProvidenceGoogle Scholar
  36. Schrand AM, Hens SAC, Shenderova OA (2009) Nanodiamond particles: properties and perspectives for bioapplications. Crit Rev Solid State Mater Sci 34:18–74CrossRefGoogle Scholar
  37. Shenderova OA, Hens SAC (2010) Detonation nanodiamond particles processing, modification and bioapplications, In Ho D (ed) Nanodiamonds: applications in biology and nanoscale medicine, Springer, New York, pp 79–116Google Scholar
  38. Turner S, Lebedev OI, Shenderova O, Vlasov II, Verbeeck J, Van Tendeloo G (2009) Determination of size, morphology, and nitrogen impurity location in treated detonation nanodiamond by transmission electron microscopy. Adv Funct Mater 19(13):2116–2124CrossRefGoogle Scholar
  39. Ushizawa K, Sato Y, Mitsumory T, Machinami T, Ueda T, Ando T (2002) Covalent immobilization of DNA on diamond and its verification by diffuse reflectance infrared spectroscopy. Chem Phys Lett 351(1–2):105–108CrossRefGoogle Scholar
  40. Vaijayanthimala V, Tzeng YK, Chang HC, Li CL (2009) The biocompatibility of fluorescent nanodiamonds and their mechanism of cellular uptake. Nanotechnology 20(42):425103CrossRefGoogle Scholar
  41. Waring MJ (1965) Complex formation between ethidium bromide and nucleic acids. J Mol Biol 13(1):269–282CrossRefGoogle Scholar
  42. Xing Y, Dai L (2009) Nanodiamonds for nanomedicine. Nanomedicine 4(2):207–218CrossRefGoogle Scholar
  43. Yang W, Auciello O, Butler JE, Cai W, Carlisle JA, Gerbi JE, Gruen DM, Knickerbocker T, Lasserter TL, Russell JN, Smith LM, Hamers RJ (2002) DNA-modified nanocrystalline diamond thin-films as stable, biologically active substrates. Nat Mater 1:253–257CrossRefGoogle Scholar
  44. Yeap WS, Tan YY, Loh KP (2008) Using Detonation nanodiamond for the specific capture of glycoproteins. Anal Chem 80(12):4659–4665CrossRefGoogle Scholar
  45. Yu SJ, Kang MW, Chang HC, Chen KM, Yu YC (2005) Bright fluorescent nanodiamonds: no photobleaching and low cytotoxicity. J Am Chem Soc 127(50):17604–17605CrossRefGoogle Scholar
  46. Yuan Y, Chen Y, Liu JH, Wang H, Liu Y (2009) Biodistribution and fate of nanodiamonds in vivo. Diam Relat Mater 18(1):95–100CrossRefGoogle Scholar
  47. Zhu Y, Li W, Li Q, Li Y, Li Y, Zhang X, Huang Q (2009) Effects of serum proteins on intracellular uptake and cytotoxicity of carbon nanoparticles. Carbon 47(5):1351–1358CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media B.V. 2012

Authors and Affiliations

  • Natalie M. Gibson
    • 1
  • Tzy-Jiun Mark Luo
    • 1
  • Olga Shenderova
    • 1
    • 2
  • Alexey P. Koscheev
    • 3
  • Donald W. Brenner
    • 1
  1. 1.Department of Materials Science and EngineeringNorth Carolina State UniversityRaleighUSA
  2. 2.International Technology CenterResearch Triangle ParkUSA
  3. 3.State Scientific Center of Russian FederationKarpov Institute of Physical ChemistryMoscowRussia

Personalised recommendations