1 Introduction

The investigations of diamond and diamond containing materials as selective adsorbents for application in solid phase extraction (SPE) and chromatography have intensified significantly during the last decade (Nesterenko and Haddad 2010; Peristyy et al. 2014). This trend is due to superior chemical and mechanical properties of diamond and relatively low cost of production of synthetic diamond. Chemical inertness, excellent mechanical and hydrolytic stability, extraordinary high thermal conductivity in combination with low thermal expansion makes diamond a very promising material for the use in chromatography (Peristyy et al. 2014; Nesterenko and Haddad 2010; Chen et al. 2006). However, there is a lack of information on the adsorption properties of diamond. Especially, this is true for detonation nanodiamond (DND), which has very complex surface chemistry defined by synthesis and purification methods applied during production (Mochalin et al. 2012; Kulakova 2004).

The adsorption of inorganic ions on DND surface is of great importance for various applications as it affects surface charge and, therefore, colloidal stability of its suspensions (Chiganova et al. 1993) and aggregation of DND (Dolenko et al. 2014). Therefore, adsorption of ionic impurities is crucial for DND performance in various applications including galvanic coatings (Chiganova et al. 2013), lubricating compositions (Ivanov et al. 2010), fabrication of electronic devices (Williams 2011) and chromatographic applications (Nesterenko and Haddad 2010; Peristyy et al. 2014; Wiest et al. 2011). Also, ion-exchange properties of DND may be responsible for the negative effects in drug delivery associated with the excessive delivery of sodium ions into the cells and related implications for tumour therapy (Zhu et al. 2012) or for so called “Trojan horse” effect, connected with intracellular delivery of Cu2+, Cd2+, Cr3+ and Ni2+ by DND particles and triggering cytotoxicity (Zhu et al. 2015). On the contrary, the controlled adsorption of lanthanides, transition metals and precious metal ions on DND has been also used for the preparation of new useful magnetic and optically active materials for medical diagnostics (Manus et al. 2010), and efficient catalysts (Chiganova et al. 2012).

Undoubtedly, cation-exchange properties of DND are due to the presence of negatively charged carboxylic and phenolic groups at its surface (Paci et al. 2013). For the purified DND the typical value of dzeta potential (ζ) is about −(42–45) mV in suspensions with pH >10 (Mitev et al. 2014a, b). However, higher values of the negative ζ-potential up to −70 mV may be achieved after extensive surface carboxylation. The adsorption of divalent and trivalent metal cations due to complexation with surface carboxylic groups is also possible for DND (Nesterenko et al. 2007; Buchatskaya et al. 2015). At the same time, the preparation of DND samples with high positive values of ζ-potential (50–55 mV) has been also reported by Ozawa et al. (2007) and Petit et al. (2013). In this case, positive ζ-potential is responsible for anion-exchange interactions between this type of DND and inorganic anions (Sakurai et al. 2006).

The number of publications on systematic investigation of adsorption of inorganic ions by DND is rather limited and non-informative. As a rule, the authors of these publications reported only values of adsorption capacity for various cations. For example, Bogatyreva et al. (2008a, b) reported the adsorption capacities of 0.20 and 0.16 mmol g−1 for Fe3+ and Ni2+, respectively, for the DND sample with specific surface area S = 172 m2 g−1. The adsorption capacities of 0.49, 0.44, 0.30 and 0.10 mmol g−1 are reported for DND (no presented data on surface area) by Zhu et al. (2015) for Cr3+, Cu2+, Ni2+ and Cd2+, respectively. Significantly higher adsorption capacity of 3.37 mmol g−1 was measured for Al3+ by Chiganova et al. (1993), which can be explained by hydrolysis of aluminium salt and precipitation of aluminium hydroxide under experimental conditions. Chukhaeva and Cheburina (2000) measured adsorption capacity for Cs+ for 12 DND samples produced using different purification technologies. The obtained values varied from 0.005 to 0.7 mmol g−1, while the maximum value was obtained for DND sample purified with chromic anhydride. Nesterenko and co-authors reported adsorption capacity 2.0 mmol g−1 for Na+ (Mitev et al. 2014a, b) by using direct ICP-MS analysis for determination of elemental impurities in DND (Mitev et al. 2013).

These capacities are in a good agreement with the reported data on surface concentration of acidic carboxylic functional groups in DND. The values of 0.15–0.81 COOH groups per nm2 and up to 0.20 lactone groups per nm2 (Schmidlin et al. 2012) are obtained by Boehm titration, while acid–base potentiometric titration revealed the presence of 1.3–1.4 carboxylic groups per nm2 (Chiganova et al. 2012; Nguyen et al. 2007; Mitev et al. 2014a, b). Assuming that DND particles have a spherical shape of average diameter d p = 4.5 nm with density ρ = 3.5 g cm−3, the maximum value of specific surface area (S = 6/ρd p) could be as high as 380 m2 g−1, and the corresponding ion-exchange capacity due to presented in DND carboxylic groups could be as high as 0.89 mmol g−1. It should be noted that this value of ion-exchange capacity is comparable with that reported for commercially available cation-exchange resins (Haddad et al. 2008).

There is also communication on adsorption of inorganic anions by DND. Sakurai et al. investigated adsorption of AsO3 3−, AsO4 3−, Cr2O7 2−, ReO4 2−, SeO3 2−, SeO4 2−, MoO4 2−, and WO4 2− using DND treated by ball milling and containing 0.4 % of Zr. The quantitative adsorption of WO4 2− from aqueous solution with maximum capacity of 0.064 mmol g−1 was found, while only partial adsorption was noted for AsO4 3−, Cr2O7 2− and MoO4 2− (Sakurai et al. 2006) with adsorption capacities of 2.4, 3.0 and 6.5 µmol g−1, respectively. It was proposed that the positive surface charge of this type of DND (ζ = 55 mV at pH = 6) and anion uptake could be associated with the presence of protonated primary amino groups at the DND surface. However, no adsorption of Cu2+ owing to complexation with primary amino groups was noted at these conditions.

Interestingly, despite the presence of negatively charged functional groups at the surface, the capability of DND to adsorb anions has been also demonstrated. For example, adsorption of anionic complexes AuCl4 − and [Rh(H2O)6−n Cl n ]3−n (n = 4, 5) was studied by Skorik et al. (2011), and maximum adsorption capacity of 1.1 mmol g−1 was found for the latter complex.

Several mechanisms have been proposed for the retention of cations and anions on the surface of diamond, but the majority of them are inconclusive and unable to explain important experimental details. For example, a possibility of co-joint adsorption of cations and anions on polyampholyte surface of DND has not been considered as an option. Therefore, the goal of this work is investigation of adsorption of metal cations and inorganic anions on MSDN with the focus on zwitter-ionic properties of DND.

2 Experimental

2.1 Materials and reagents

All adsorption experiments were carried out using additionally fractionated MSDN sample (PDD 3–6) purchased from ALIT Co. (Kiev, Ukraine). The preparation, properties of MSDN particles and chromatographic performance have been described in the literature (Bogatyreva et al. 2000; Yushin et al. 2005; Fedyanina and Nesterenko 2010, 2011; Nesterenko and Fedyanina 2010). Before the use MSDN was purified by washing with 5 M HNO3 in order to remove metal impurities, followed by washing with 1 M NaOH to remove adsorbed anions, and then again with 5 M HNO3 to ensure all surface groups are protonated. Finally, the excess of HNO3 was removed by flushing MSDN with DIW until the pH of washings was in the range 4.5–5.0.

The following salts MnSO4, NaClO4 (all from AJAX Chemicals, Scoresby, VIC, Australia), Zn(NO3)2, CoSO4, CuCl2, CuSO4, Na2SO4, Cu(NO3)2, Na2C2O4, NaCl (Sigma-Aldrich, Castle Hill, NSW, Australia), Cd(NO3)2, NiSO4, Al(NO3)3, Na3PO4, NaI, Fe(NO3)3 (99 %, all from VWR International, Murarrie, QLD, Australia) were used for the preparation of stock solutions of cations and anions. Solutions of hexamine, Na2B4O7, CH3COOH were used as pH buffers, while concentrated HNO3, H2SO4 (99 %, all from Chem-Supply, Gillman, SA, Australia), H3PO4 (99 %, all from Univar, Ingleburn, NSW, Australia) and NaOH (99 % Scharlau, Sentmenat, Barcelona, Spain) were used for the pH adjustment. Deionised water (DIW) was obtained from a Milli-Q system (Millipore, USA) and used for the preparation of all solutions.

For the adsorption experiments, 0.2 g of MSDN and 10 mL of the ion-containing solution of required concentration and adjusted pH were placed in polypropylene test tubes. After shaking for at least 30 min (shorter times were applied in kinetics study of adsorption), the test tubes were centrifuged, and the supernatant was collected and filtered through a membrane filter (pore size 0.45 µm). The determination of ion concentrations in supernatant was performed using either ion chromatography (IC) or spectrophotometrically with o-cresolphtalein (99 %, Sigma-Aldrich, Castle Hill, NSW, Australia), 4-(2-pyridylazo)resorcinol (PAR) (98 %, Kodak, Rochester, NY, USA), or chromazurol S (85 %, Riedel-de-Haёn, Sigma-Aldrich, Seelze, Germany). 0.1 M HNO3 was used to desorb metal cations from MSDN.

2.2 Instrumentation

A Nanoseries zetasizer Model Zen3600 (Malvern Instruments, Malvern, Worcestershire, UK) was used for ζ-potential measurements. The Smoluchowski approximation for large particle sizes (1–2 µm) and high ionic strengths (>100 mM) was used according to Hunter (1981). SEM images of MSDN particles were obtained with Hitachi SU-70 (Hitachi Ltd., Chiyoda, Tokyo, TKY, Japan) field emission scanning electron microscope and 1.5 keV electron beam. All samples were platinum-sputtered prior to analysis. BET surface area measurements and porous structure of MSND were measured by adsorption of nitrogen at low temperatures using Micromeritics TriStar II analyser (Norcross, GA, USA).

FTIR spectra of MSDN were obtained using Bruker MPA spectrometer (Billerica, MA, USA). Prior to measuring FTIR spectra, MSDN powder was dried overnight in a vacuum oven at 100 °C and 0.4 atm. Potassium bromide tablets containing 5 % w/w of MSDN were used as samples.

2.3 Experiments on adsorption of ions

2.3.1 Determination of metal cations

IC system comprising AS50 autosampler, IP25 isocratic pump, and AD25 photometric detector (all from Thermo Scientific, Sunnyvale, CA, USA) was used for the determination of the transition metal cations. IonPac CS16 column (250 mm × 3.0 mm ID) was used for the determination of metal ions (Zn2+, Mn2+, Co2+, Ni2+, Cd2+, Cu2+, Fe3+) with 0.1 M HNO3 as eluent at flow rate of 0.36 mL min−1. A 1.0 × 10−4 M solution of PAR in 0.1 M sodium tetraborate buffer (pH 9.18) was used as post column reagent for the photometric detection of metals at 490 nm. Similarly, 1.0 × 10−4 M solution of chromazurol-S in 0.1 M hexamine buffer (pH = 6.0) was used for the photometric detection of Al3+ at 545 nm using UV–Vis spectrophotometer Metertech SP-8001 (Nankang, Taipei, Taiwan).

2.3.2 Determination of inorganic anions

ICS-2000 ion chromatograph (Thermo Scientific, Sunnyvale, CA, USA) equipped with an AS50 autosampler, conductivity detector and IonPac AS19 column (250 mm × 4.0 mm ID) was used for the determination of inorganic anions. 40 mM KOH eluent was generated by eluent generator EGC-II and used for the separation of NO3 −, SO4 2−, ClO4 −, Cl−, I−, and PO4 3− at 30 °C and 1 mL min−1 flow rate. B4O7 2− and C2O4 2− were quantified using ion-exclusion IC mode with an IonPac ICE-AS6 column (250 mm × 9.0 mm ID) and 10 mM H2SO4 eluent at 1 mL min−1.

2.3.3 Boehm titration

Acid–base potentiometric titration was accomplished using Metrohm 809 Titrando autotitrator with Tiamo 1.2 software (MEP, Mitcham, VIC, Australia). Vent capsule filled with Lecosorb NaOH/SiO2 adsorbent (Leco Corporation, St. Joseph, MI, USA) was attached to titrant container in order to prevent CO2 uptake by NaOH titrants. Prior to titration, all acidic groups at the surface of MSDN were protonated by consecutive washing with 5 mM HCl and DIW, and drying at 100 °C. 4.8 mM HCl solution used in titration experiments was normalised with 10 mM standard solution of potassium hydrogen phthalate prepared by dissolving a precisely weighted amount of anhydrous salt in water using volumetric flask. 0.5 g of dried MSDN was placed in three 50 mL polypropylene vials and 20 mL of solutions of NaOH, NaHCO3 or Na2CO3 (~10 mM) were added according to the original work of Boehm (1994). Vials were kept for 1 h being sporadically shaken. After centrifugation the supernatant was collected and filtered through 0.22 µm pore size Nylon filter. Aliquots of 5 mL of filtrates were titrated with 4.8 mM HCl solution. Both initial (C 0) and equilibrium concentrations (C eq ) of NaOH, NaHCO3 and Na2CO3 were determined by titration. Uptakes of NaOH, NaHCO3 or Na2CO3 (U, mmol g−1), which indicate the concentrations of acidic groups (C) of different acidity at the surface of MSDN, were calculated as following:

$$U = \frac{{V_{base} \times (C_{0} {-}C_{eq} )}}{m}$$
(1)

where V base is the volume of the added alkaline solution (20 mL) and m is the mass of the titrated MSDN (0.5 g).

2.4 Adsorption isotherms

Solutions of increasing concentrations (0–100 µmol mL−1) were used to study the adsorption of ions on MSDN (0.2 g) from solutions of constant volume (5 mL). Adsorption (A, µmol g−1) was calculated as follows:

$$A = \frac{{(C_{0} {-}C_{eq} ) \cdot V}}{m}$$
(2)

where C 0 and C are original and equilibrium concentrations (µmol mL−1) of the ion, V is volume of solution (mL) and m is weight (g) of MSDN. Adsorption values (µmol g−1) were plotted versus equilibrium concentration of ion after adsorption.

Langmuir and Freundlich models (Eqs. 3 and 4, respectively) were applied to describe the adsorption isotherms:

$$A = \frac{{A_{\hbox{max} } \cdot K_{L} \cdot C}}{{1 + K_{L} \cdot C}}$$
(3)

where A max is the limit of adsorption achieved by increasing concentration C and K L is coefficient equal to the slope of adsorption isotherm at low ion concentrations. K L is also proportional to the distribution coefficient (K D ). Freundlich isotherm of adsorption is described as follows:

$$A = K_{F} \cdot C^{1/n}$$
(4)

where n and K F are constants. Correlation factors (R 2) between the experimental and simulated isotherms were calculated in order to check which model fits better the experimental data.

3 Results and discussion

3.1 Fractionation, purification and porous structure of MSDN

The original batch of MSDN contained particles with broad distribution on size from 10 nm to 30–40 µm. The presence of fine particles of size <200 nm may cause some problems in handling MSDN during experiments and could decrease the accuracy of the results. Therefore, the fine particles were removed from the original batch of MSDN by repetitive sedimentation from isopropanol. SEM image of finally isolated fraction (see Fig. 1) shows the presence of particles with the sizes between 0.5 and 20 µm. Each of these particles is composed of nanoparticles of size 5–30 nm.

Fig. 1
figure 1

SEM images of MSDN (see Sect. 2.2 for details)

Full size image

According to producer data (Mitev et al. 2014a, b) the supplied MSDN contained 12.59 mg g−1 (or 1.259 mass%) of elemental impurities (Ca, Cr, Ni, Mn, Sn, Fe, Si and others) including 4.27 mg of Fe and 1.43 mg of Si per gram, respectively. The presence of elemental impurities in MSDN could effect on adsorption data for selected ions, so fractionated MSDN particles were additionally purified as described in Sect. 2.1. The accurate control of impurities in purified MSDN was performed using inductively coupled plasma – mass spectrometry (ICP-MS) as described (Mitev et al. 2013). The results of ICP-MS analysis of MSDN after purification (see Sect. 2.1) showed the significant decrease in amount of metal impurities including 556.5 µg g−1 of Cr, 169.1 µg g−1 of Fe, 141.8 µg g−1 of Cu, 111.5 µg g−1 of Al, 29.6 µg g−1 of Zn, 18.8 µg g−1 of Ni, 3.0 µg g−1 of Mn and others (Mitev et al. 2014a, b). Obviously, these amounts of impurities are due to metals entrapped into MSDN matrix during detonation synthesis, and therefore they should not influence the results on adsorption of ions.

The isotherm of nitrogen adsorption at low temperatures has characteristic Type IV shape with hysteresis (Fig. 2a), which is typical for mesoporous adsorbents. Brunauer–Emmett–Teller (BET) and Barrett–Joyner–Halenda (BJH) methods were used for the evaluation of surface area (S) and pore size (D pore) and pore volume of MSDN. The obtained value of surface area 191 m2 g−1 was slightly higher than 153 m2 g−1 reported by the manufacturer (Bogatyreva et al. 2008a, b). As shown in Fig. 2b, the major pores have diameter about 4.0 nm, but significant amount of smaller mesopores and micropores is also present.

Fig. 2
figure 2

Adsorption isotherm of nitrogen (a) and dependence of pore volume on pore width (b) for fractionated MSDN sample

Full size image

3.2 Surface characterisation

3.2.1 Surface chemistry

According to the literature, various oxygen containing functional groups are present at the surface of DND (Schmidlin et al. 2012; Kulakova 2004; Paci et al. 2013; Nesterenko and Fedyanina 2010) including carboxyls, carbonyls, hydroxyls and lactones. To evaluate the amount of these functional groups the FTIR spectrum of MSDN was obtained (Fig. 3).The broad absorption band at 3000–3600 cm−1 corresponds to O–H stretching mode, and can be attributed to adsorbed water, hydroxyl groups and the OH of the carboxyl groups. The related shoulder at 2890–2970 cm−1 can be assign to C–H symmetric and asymmetric stretches. The characteristic band at 1735 cm−1 is due to the C=O stretching of the carbonyl. The band at 1000–1320 cm−1 is related to C–O stretches in the hydroxyl groups. The spectrum confirms the presence of oxygen-containing functional groups in MSDN and indicates the hydrophilic character of this material. Overall, the observed FTIR spectrum is similar to reported previously for MSDN (Nesterenko and Fedyanina 2010).

Fig. 3
figure 3

FTIR spectrum of purified MSDN

Full size image

3.2.2 Surface charge

As discussed in Sect. 1, ζ-potential of DND may vary in the range from −70 to +55 mV depending on the type of applied purification, disaggregation and modification (Paci et al. 2013; Gibson et al. 2009; Petit et al. 2013; Ozawa et al. 2007). The occurrence of negative ζ-potentials in alkaline conditions is associated with the presence of dissociated hydroxyl- and carboxyl groups, but the reasons for the positive ζ-potential can be more complex. The simple explanation is based on the presence of protonated nitrogen-containing functionalities. However, neither N–H nor C–N related bands can be found at detectable levels in FTIR spectrum of MSDN (Fig. 3). Another possible explanation is connected with the presence of sp 2-carbon layer with various functional groups at the external surface of DND (Mochalin et al. 2012). Therefore, the positive ζ-potential can be attributed to the protonation of pyrone, chromene and phenol functional groups at graphitic surface of MSDN (Paci et al. 2013). It should be noted that Penner and Nesterenko (2000) have observed similar effect for hyper crosslinked polystyrene, when protonation of carbonyl groups was responsible for anion-exchange of inorganic anions. The maximum of anion-exchange capacity was achieved at pH 2. Finally, the presence of positive net charge at (100) facets and neutral or negative net charges at (111) facets in diamond nanoparticles was proposed by Chang et al. (2011). However, no clear experimental proofs were provided for the last hypothesis.

Figure 4 presents the dependence of ζ-potential on pH of aqueous suspension for the prepared MSDN sample. At pH 2 MSDN surface has a maximum positive ζ-potential of +17 mV. This value gradually reduces with the increase in pH, with an isoelectric point observed at pH 8.0. The further rise in pH increases negative ζ-potential of MSDN surface up to −23 mV at pH 12. Typical value of negative ζ-potential for untreated DND with fully dissociated functional groups is about 30–35 mV, which is slightly higher than that measured for MSDN sample.

Fig. 4
figure 4

pH profile of ζ-potential for MSDN. Measured for 0.1 mg mL−1 aqueous suspension of MSDN (µ = 0.01 M). pH adjusted with either NaOH or HNO3 1 M solutions

Full size image

3.3 Adsorption of ions

Very little attention has been paid to the investigation of adsorption mechanism for inorganic ions on DND in the past. Due to the presence of both negative and positive charges at the surface, MSDN may exhibit zwitter-ionic properties, so the adsorption of both anions and cations was studied in this work. The adsorption properties of MSDN were investigated using different buffers, pH of solutions, solute concentrations and sample volumes.

3.3.1 Metal cations

3.3.1.1 pH dependence and adsorption selectivity

As shown in Fig. 5, MSDN adsorbs metals cations from 0.05 M sodium phosphate buffer over pH range from 1 to 6 and the following affinity row Al3+ > Fe3+ > Cu2+ > Mn2+ > Zn2+ > Co2+ > Ni2+ > Cd2+ was obtained. This selectivity is similar to Fe3+ > Ni2+ > Cr3+ > Pb2+ > Cd2+ reported by Bogatyreva et al. (2008a, b; 2010a, b). Surprisingly, the adsorption of metal cations takes place in acidic solutions with pH <7 despite the expected repulsion of cations from positively charged surface as shown in Fig. 4. The obtained selectivity order differs from the selectivity series Cu2+ > Zn2+ > Cd2+ > Ni2+ > Mn2+ and Cu2+ > Ni2+ > Co2+ > Zn2+ > Mn2+ previously reported for common carboxylic cation exchangers (Hubicki and Kolodynska 2015) and oxidised active carbon, respectively (Strelko and Malik 2002). This difference can be explained by dual nature of adsorption mechanism for metal cations on MSDN (Nesterenko et al. 2007). First option includes a combination of electrostatic interactions between negatively charged carboxyls and protonated carbonyls located at MSDN surface and inorganic cations and anions from the solution. The other option is coordination of metal cations with either two carboxylic groups or with one carboxylic and one hydroxyl group located at the surface in a close proximity, when the formation of stable surface complexes is possible due to the chelation effect (Nesterenko et al. 2007). Also, it was found that the presence of sp 2-carbon at the surface of DND has a strong influence on the adsorption of transition metals (Bogatyreva et al. 2010a, b). The maximum adsorption of Fe3+, Ni2+, Cr3+, Pb2+, and Cd2+ was obtained for DND with 80:20 ratio of diamond core to sp2-carbon shell. Additionally, the hydrophobicity of the DND has a strong effect on adsorption (Bogatyreva et al. 2010a, b), which is relevant to the concentration of polar carboxyl and hydroxyl- groups at the surface.

Fig. 5
figure 5

Adsorption of various metals on MSDN versus pH of solutions. Sample volume: 10 mL of 10 µg mL−1 of corresponding salts. Solutions in 0.05 M sodium phosphate buffer and HNO3 solutions of varied concentration were used for pH >3.0 and pH ≤3.0, respectively

Full size image

As expected, MSDN demonstrated a higher affinity towards triple charged Fe3+ and Al3+, which can be adsorbed from acidic solutions with pH from 1 to 3 (Fig. 5). This is in agreement with the results on investigation of adsorption of triple charged lanthanides and actinides cations on DND (Watanabe and Kimura 2012). In this pH range the surface of MSDN is positively charged (see Fig. 4), and the adsorption of Fe3+ and Al3+ can be only due to the formation of relatively stable complexes with oxygen-containing functional groups at the surface (Nesterenko and Jones 2007).

High affinity of MSDN towards Cu2+ and Mn2+ was also noted (Fig. 5) at pH >3. It is known that Cu2+ can form stable octahedral complexes with carboxylic groups presented on the surface of oxidised active carbon (Strelko and Malik 2002) or sp 2 carbon coated MSDN (Shames et al. 2010). However, the reason for the high affinity of MSDN towards Mn2+ is not completely clear. Probably, this is due to co-adsorption of Mn2+ and phosphate ions from phosphate buffer used in this set of experiments.

For the rest of cations (Zn2+, Ni2+, Co2+ and Cd2+) the adsorption increased gradually with the increase of pH from 3 to 5, while substantial adsorption was observed only for Zn2+. A weak adsorption of Ni2+, Co2+ and Cd2+ (~1 µg g−1) by MSDN was noted at pH >4.0. It should be underlined that the adsorption of metal cations at pH above 5–6 may be affected by hydrolysis and precipitation of insoluble phosphates or hydroxides, therefore the adsorption was not studied at pH >6, depending on the cation. Fig. 1S in supporting information provides the solubility product constants (K sp , see Table 1S in supplementary) for the corresponding metal phosphates (Cu2+, Cd2+, Co2+, Zn2+, Ni2+ and Mn2+) and hydroxides (Al3+, Fe3+) and calculated pH values at which precipitation of the corresponding species begins under given conditions (metal concentration 10 mg L−1, 0.05 M phosphate buffer). It should be noted that adsorption of trivalent metals at pH 1–3 from diluted nitric acid solutions.

3.3.1.2 Adsorption capacity

The adsorption capacity of MSDN depends on the type and surface concentrations of various groups such as carboxy-, hydroxy-, carbonyls and others. Boehm titration is used for the determination of functional groups in carbonaceous adsorbents according to their acidity by reaction with HCl, sodium bicarbonate, sodium carbonate and sodium hydroxide. The results of Boehm titration of MSDN, diamond of static synthesis at high temperature and high pressure, activated carbon and carbon black are presented in Table 1. The relatively high density of 0.45–0.56 acidic functional groups per nm2 of the surface of different types of diamond was obtained, which is 3–10 times higher than values reported for activated carbon and carbon black (Kalijadis et al. 2011). The total ion-exchange capacity of MSDN (0.177 mmol g−1) is comparable with the capacity of 0.4–1.0 mmol g−1 for commercial ion-exchange resins (Haddad et al. 2008). As above noted the highest adsorption capacity of DND measured for caesium was 0.7 mmol g−1 (Chukhaeva and Cheburina 2000), but an absolute maximum of ion-exchange capacity of 2.0 mmol g−1 was reported for DND after the treatment with concentrated sodium carbonate solution (Mitev et al. 2014a, b).

Table 1 Experimental and literature data for Boehm titration of various carbonaceous materials
Full size table

In the case of adsorption of transition metals, the highest value of adsorption capacity of 9 µmol g−1 was observed for Mn2+. This value is significantly less than the values of 200–300 µmol g−1 reported for activated carbon (Strelko and Malik 2002). However, it should be taken into account, that the surface area of MSDN used in this work is 191 m2 g−1, which is approximately four times less than the values for activated carbon (850 m2 g−1) and carbon black (up to 1500 m2 g−1) as presented in the Table 1.

It is likely that the diamond surface possesses diphilic character and adsorption of both cations and anions can occur (Dolenko et al. 2014; Sakurai et al. 2006; Nesterenko et al. 2007). However, the direct titration with HCl shows the presence of only minor anion-exchange capacity (Table 1).

3.3.2 Kinetics of adsorption

The adsorption kinetics of Cu2+ and SO4 2− on MSDN was studied. The maximum of adsorption was achieved after 5 min of shaking and the obtained curves show typical first order adsorption kinetics (Fig. 6a). MSDN demonstrates relatively fast kinetics of adsorption due to dominance of electrostatic interactions as compared with the other types of microporous and mesoporous carbonaceous adsorbents (Chen and Lin 2001). Figure 6b presents the results on adsorption of 0.625 µmol of Cu2+ and SO4 2− on 0.2 g of MSDN from continuously increasing volumes of phosphate buffer at pH 5. The calculated distribution constant (K D ) for Cu2+ was equal to 280 mL g−1, which is comparable with the K D values obtained for the carboxylic cation exchangers (Hubicki and Kolodynska 2015).

Fig. 6
figure 6

Dependence of Cu2+ and SO4 2− adsorption on MSDN (0.2 g) on time (a), and on phase ratio (b). a obtained with 10 mL of 0.0625 mM CuSO4 in 0.05 M sodium phosphate buffer (pH 5.0) and b obtained with 0.625 µmol of CuSO4 in 0.05 M sodium phosphate buffer (pH 5.0)

Full size image

3.3.3 The effect of anions on adsorption of copper

As discussed in Sect. 1, MSDN can possess both cation and anion exchange properties. On this reason, the adsorption of anions and cations at different pH can be dependent on each other as proposed by Dolenko et al. (2014). To investigate this option, the effect of anionic component from pH buffers on adsorption of Cu2+ was studied. The corresponding dependences obtained for sodium acetate and sodium phosphate buffers are shown in Fig. 7. The adsorption of Cu2+ on MSDN occurred at pH >3.0 in both buffers and increased with pH growth. The complete adsorption of Cu2+ was achieved from phosphate buffer at pH 5.0, but only partial adsorption of Cu2+ was obtained from acetate buffer even at higher pH 6.0. Therefore, this happens due to a stronger adsorption of phosphate, that resulted in increase of negative charge for MSND surface. The adsorbed phosphate acts as an additional adsorption site for metal cations. Another reason for the observed effect is connected with an ability of acetate to form neutral copper acetate complexes [log K = 3.4; (Martell and Smith 2004)] that competes to interaction of Cu2+ with carboxyl groups at the surface of MSDN. It should be noted that strong interaction between organophosphonates and hydroxylated diamond surface has been previously used for functionalisation of diamond electrodes (Caterino et al. 2014).

Fig. 7
figure 7

Adsorption of Cu2+ on MSDN (0.2 g) depending on the pH of sodium phosphate and sodium acetate buffers. Experiments were done from 10 mL of 10 µg mL−1 Cu2+ solution in 0.05 M buffer

Full size image

The concentration of ions in pH buffers is significantly higher than the concentration of the analyte; hence, the effect of anions on the adsorption of Cu2+ is expected. However, the study of Cu2+ adsorption on MSDN from solutions of different copper salts in absence of pH buffers can illustrate better the role of anions. Figure 8 presents the corresponding adsorption isotherms obtained with the solutions of copper chloride, nitrate and sulphate salts. All isotherms have three distinctive “steps”, but adsorption of Cu2+ from copper sulphate solution was higher than from solutions of corresponding chloride and nitrate salts. The presence of several “steps” in the isotherms is quite unusual and indicates the occurrence of several types of interactions around Cu2+ in the system. Since the type of anion does not influence the position of “step”, the corresponding interactions must take place at different adsorption sites in MSDN. It should be noted that adsorption isotherms for heavy metals on graphite adsorbents show only one type of adsorption site (Seco et al. 1999) indicating a substantial difference between adsorption properties of graphite and MSDN, despite the presence of thin sp2 carbon layer at the DND surface (Aleksenskii et al. 1999).

Fig. 8
figure 8

Adsorption isotherms for Cu2+ on MSDN (0.2 g) obtained with 10 mL of Cu2+ containing solutions of different salts

Full size image

A possible explanation for the observed steps on the adsorption isotherms can be derived from the ζ-potential–pH dependence for MSDN (see Fig. 4). The pH of native MSDN suspension in water is about 4.5–5.0, that means that the surface is positively (+14 mV) charged, and Cu2+ adsorption is reduced due to electrostatic repulsion at these conditions. At the same time, counter ions (SO4 2−, NO3 −, and Cl−) are adsorbed at this pH and neutralise the surface charge of MSDN, while for doubly charged SO4 2− counter ion it happens at lower concentrations than for the singly charged Cl− or NO3 −. Thus, the interaction of MSDN with anions provides an additional step on adsorption isotherm and lead to the higher Cu2+ adsorption capacity in the case of CuSO4, as compared to CuCl2 and Cu(NO3)2. Maximum adsorption capacity was found to be 5 µmol g−1 for CuSO4 and 3.5 µmol g−1 for CuCl2 and Cu(NO3)2. To confirm the proposed mechanism the adsorption of Cu2+ was additionally investigated from solutions of copper sulphate and copper nitrate with added sodium sulphate and sodium nitrate, respectively. Figure 9a shows the growth in adsorption of Cu2+ on MSDN with increase of SO4 2− concentration in the solution, but steady low adsorption with varied concentration of NO3 −. Therefore, the adsorbed SO4 2− ions can act as an additional adsorption sites for Cu2+.

Fig. 9
figure 9

Influence of nature of anion (a) and ionic strength (b) on Cu2+ adsorption on MSDN (0.2 g). 10 mL of 10 µg mL−1 Cu2+ solution were used. a Na2SO4 or NaNO3 were added to increase counter ion/Cu2+ ratio. b Ionic strength was maintained with NaNO3

Full size image

The ionic strength of the solution influences the adsorption of Cu2+ too. As shown in Fig. 9b the adsorption of Cu2+ is notably growing with increase of ionic strength, especially in the range of 0–0.05 M. This indicates the suppression of electrostatic repulsion between Cu2+ and positively charged MSDN surface, which results in elevated adsorption of Cu2+ from concentrated solution of sodium nitrate due to chelation mechanism. This effect is well described for complexing ion-exchangers in the literature (Nesterenko and Jones 2007).

3.4 Adsorption of anions

As counter ions have influence on the metal cation adsorption on MSDN, the adsorption of inorganic anions was carefully studied. Figure 10 shows adsorption isotherms for eight inorganic anions on MSDN and calculated values of distribution coefficients (K D ) as well as an approximation of adsorption isotherms to Langmuir and Freundlich models (Eqs. 2 and 3) are presented in Table 2. The obtained data show that the adsorption of anions is better described by the Langmuir equation except of PO4 3−, which behaviour will be explained later. This is in an agreement with assumption of monolayer adsorption of anions on the MSDN surface. The maximum adsorption capacities for anions vary from 50 to 150 µmol g−1, which are significantly higher than the values obtained for cations. This is the first time when superior anion adsorption capacities are revealed for DND.

Fig. 10
figure 10

Isotherms of adsorption on MSDN for eight inorganic anions from 5 mL solutions of sodium salts of increasing concentration. (bottom figure gives a magnified view in the area of low initial concentration of anions)

Full size image
Table 2 Adsorption capacity and distribution coefficients for anion adsorption on MSDN and adsorption modelling with Langmuir and Freundlich equations
Full size table

The following selectivity order CH3COO− < Cl− < B4O7 2− < ClO4 − < I− < SO4 2− < C2O4 2− is obtained for adsorption of anions on MSDN. This sequence is different from common anion-exchange selectivity series B4O7 2− < CH3COO− < Cl− < I− < C2O4 2− < SO4 2− < ClO4 − known for common anion-exchangers resins with quaternary ammonium and tertiary amino- functional groups (Haddad et al. 2008). MSDN possesses a higher affinity towards B4O7 2− and C2O4 2− and lower affinity for ClO4 −, which means that the presence of not only ion-exchange adsorption mechanisms for this adsorbent. For example, FTIR spectrum (Fig. 3) shows the presence of high concentration of hydroxyl groups at the surface of MSDN particles, which can form polyol complexes with B4O7 2− (Simonnot et al. 2000). Also, ICP-MS analysis of purified MSDN confirms the presence of significant amount of metal impurities (Mitev et al. 2013), such as Ca2+, Sr2+ and Ba2+, which contribute into adsorption of C2O4 2− or PO4 3−.

The shape of adsorption isotherm for PO4 3− (see Fig. 10) differs from that observed for other anions. Its part obtained for low anion concentrations matches the adsorption isotherms obtained for the other anions, but with further increase of phosphate concentration in solution the adsorption suddenly decreases and remains at very low level. A possible explanation for this behaviour is that both the MSDN suspension and Na3PO4 solution used in this experiment have pH buffering capacity around pH of ~5 and 12, respectively. Therefore, at low Na3PO4 concentrations the pH of the system is defined by the pKa of MSDN particles and is equal to ~5. It means a positively charged (see Fig. 4) surface of MSDN particles which provides adsorption of phosphate as has been reported by Caterino et al. (2014). However, a further increase in the concentration of Na3PO4 causes a substantial rise in pH due to strong basicity of PO4 3− and switches the ζ-potential to negative values. Correspondingly, the adsorption of phosphate decreases due to raising electrostatic repulsion from the MSDN surface. A strong influence of PO4 3− concentration on the ζ-potential of nanodiamond was also confirmed in work of Zhu et al. (2004). The described effect was not observed for other anions as they have no buffering capacity at high pHs, except of B4O7 2−. However, the adsorption mechanism for B4O7 2− is different and related to the complexation with diol groups. In this case, increase of pH provides higher adsorption values as stability of B4O7 2−–diol complexes is growing with pH (Simonnot et al. 2000). To confirm the hypotheses the dependence of adsorption on pH of the solutions was obtained for borate, sulphate and phosphate (see Fig. 11). The concentration of solutes was kept constant and pH of the solutions was adjusted by addition of small volumes of concentrated HNO3 or NaOH. For H2PO4 − and SO4 2− a remarkable drop in adsorption is observed at pH 7, which is close to the isoelectric point for MSDN. Obviously, it follows to sharp decrease in electrostatic attraction for these anions, while the adsorption of B4O7 2− gradually grows with pH increase.

Fig. 11
figure 11

pH influence on phosphate, sulphate and borate adsorption on MSDN (0.2 g) from 10 mL of 100 µmol mL−1 solution. NaOH/HNO3 was used to adjust pH and ionic strength (µ = 0.01 M)

Full size image

The kinetics of adsorption of SO4 2− is similar to that discussed before for Cu2+ (Fig. 6a), when adsorption equilibrium can be reached in 10–20 min. K D value 563 ± 16 mL g−1 calculated for adsorption of SO4 2− from diluted solutions (Fig. 6b) is consistent with the value obtained from the SO4 2− adsorption isotherm (552 ± 44 mL g−1, see Table 2).

4 Conclusions

The adsorption of di- and tri-valent metals cations and inorganic anions from aqueous solutions on MSDN is investigated. For the first time for DND based material, the significantly higher adsorption capacities (50–150 µmol g−1) are obtained for anions as compared with adsorption capacities (<9 µmol g−1) calculated for cations. It is shown that adsorption of anions is mainly due to electrostatic ion-exchange interactions, while adsorption of cation is mainly due to complexation with carboxyl- and hydroxyl-functional groups at the surface of MSDN. The specific or chemisorption type of interactions are noted for adsorption of borate and attributed to the formation of borate–polyol type of complexes with hydroxyl groups at the surface of MSDN. Interactions with metal impurities can impact in the adsorption SO4 2− and C2O4 2−. The effect of conjoint type of simultaneous adsorption of metal cations and doubly and triply charged anions is also observed. Adsorption of ions on the MSDN surface obeys Langmuir law. The distribution coefficients for the calculated for adsorption of anions on MSDN are ranging between 10 and 870 mL g−1.