Electrochemically modulated liquid chromatography using a boron-doped diamond particle stationary phase

Abstract

This paper reports on preliminary tests of the performance of boron-doped diamond powder (BDDP) as a stationary phase in electrochemically modulated liquid chromatography (EMLC). EMLC manipulates retention through changes in the potential applied (E_appl) to a conductive packing. Porous graphitic carbon (PGC) has routinely been utilized as a material in EMLC separations. Herein the utility of BDDP as a stationary phase in EMLC was investigated and its stability, both compositionally and microstructurally, relative to PGC was compared. The results show that BDDP is stable over a wide range of E_appl values (i.e., −1.2 to +1.2 V vs. Ag/AgCl, sat’d NaCl). The data also reveal that electrostatics play a key role in the adsorption of the aromatic sulfonates on the BDDP stationary phase, and that these analytes are more weakly retained in comparison to the PGC support. The potential for this methodology to provide a means to advance the understanding of molecular adsorption and retention mechanisms on carbonaceous materials is briefly discussed.

1. Introduction

Carbonaceous packings (e.g., porous graphitic carbon, PGC) have been used in a wide variety of liquid chromatography separations [1–9]. Carbon-based materials differ microstructurally and chemically from commonly used reversed-phases. From a retention perspective, carbonaceous packings have a π-electron selectivity that is superimposed on classical reversed-phase characteristics [1–9].

Electrochemically modulated liquid chromatography (EMLC) is a separation technique that manipulates retention by changes in the potential applied (E_appl) to a conductive stationary phase. In other words, the packing functions as both a stationary phase and a working electrode in a chromatographic column designed to function as a three-electrode electrochemical cell [8,10–23]. Past work has shown that carbon materials, like PGC and glassy carbon (GC) [13], are viable stationary phases for EMLC and there are several intriguing examples in which these packings have been coated with conductive polymers, like polyvinylferrocene and polypyrrole [24–26]. Recent work has also demonstrated that, by monitoring the influence of E_appl on the capacity factor (k′) for analyte retention, a chromatographic tool for the examination of electrosorption phenomenon can be devised [10,11,21].

Stationary phases in EMLC must possess several properties: high surface area, compositional and microstructural stability, and high electrical conductivity. Note that stability must also include the inertness of the packing material over a wide range of E_appl. To date, the greatest degree of success in EMLC has been achieved using PGC. This material has a high surface area (120 m²/g) [9,10] and a high inherent electrical conductivity (≥100 S/cm) [22], which is requisite for functioning as a working electrode. PGC is also available as micron-sized particles with a narrow dispersity. However, the tendency of this material to slowly oxidize at high anodic potentials (>+0.5 V) over a few days of continuous use can prove problematic with respect to compositional stability and, therefore, retention reproducibility [27,28]. This oxidation not only alters the surface chemistry of the packing but also can change its microstructure and electronic properties (i.e., density of states), both of which can have a significant influence on retention.

This paper describes initial tests of an alternative carbonaceous packing for EMLC: boron-doped diamond powder (BDDP). The conducting powder was formed by overcoating insulating diamond powder with a layer of boron-doped diamond [29,30]. In this core-shell approach, a boron-doped microcrystalline diamond layer is deposited on 8–12 μm diamond powder particles using microwave chemical vapor deposition (CVD) and a conventional CH₄/H₂ source gas mixture. The electrical conductivity is controlled by the doped-diamond layer (carrier concentration and mobility) rather than by an adventitious non-diamond sp² carbon impurity phase. With respect to GC-based materials, BDDP is more hydrophobic, is much more compositionally and morphologically stable, exhibits a wider potential window before the onset of solvent-electrolyte electrolysis in conventional aqueous media, and is less prone to surface fouling [27–35].

At a more detailed level, electrically conducting diamond thin films can be formed with a resistivity as low as 0.01Ω-cm by introducing dopants, such as boron, into the source gas mixture. Doping produces a highly conductive material that functions effectively as a working electrode. When deposited on a suitable substrate, these coatings are chemically inert and dimensionally stable in aggressive chemical (e.g., high temperature and extremes in pH) and electrochemical (e.g., large anodic and cathodic potentials and high current densities) environments [36–38]. Indeed, reports over the past decade have established the superior morphological and microstructural stability of diamond thin films with respect to several sp² carbon materials, including GC and highly oriented pyrolytic graphite [38–43]. Recent work has also investigated the electrochemical stability of BDDP [29]. In that report, BDDP was anodically polarized at +1.6 V (vs. Ag/AgCl sat’d NaCl) at room temperature for 1 h with no apparent changes in microstructure, as determined with scanning electron microscopy (SEM), or in electrical conductivity, as measured by a four-point conductivity probe. By contrast, SEM images for GC particles were diagnostic of marked morphological changes (e.g., pitting) [29].

The goal of this preliminary work was to prepare columns packed with BDDP and to examine the possibility of probing molecular interactions at this material via retention changes with respect to E_appl. Experiments with BDDP were carried out to: (1) characterize its microstructure and electrical conductivity; (2) assess its utility in the reproducible manipulation of separations by using a mixture of aromatic sulfonates (ASFs) via changes in E_appl; and (3) to compare its retention properties with respect to that of PGC. This paper reports these initial findings.

2. Experimental

2.1. Boron-doped diamond growth and characterization

Prior to deposition of the boron-doped diamond coating, 8–12 μm diameter diamond powder (8–12 μm RVM, Diamond Innovations, Worthington, OH) was cleaned in aqua regia (concentrated hydrochloric and nitric acids, 3:1) for 30 min, and then in 30% hydrogen peroxide for 30 min. The particles were subsequently rinsed with ultrapure water, isopropyl alcohol, and acetone. The boron-doped diamond overlayer was formed by microwave plasma-assisted CVD (1.5 kW, 2.45 GHz, Seki-ASTeX, Tokyo Japan). The specific surface area, measured by the Brunauer-Emmett-Teller (BET) method, was 1.0 m²/g. Procedurally, a 50-mg sample of diamond powder was spread out on a silicon wafer and mounted in the reactor. The deposition was carried out using a source gas mixture of 1% CH₄/H₂ and 10 ppm B₂H₆ at a microwave power of 1.0 kW, a pressure of 45 Torr, and deposition time of 4 h.

The crystallinity of the coated diamond particles was evaluated using powder X-ray diffraction analysis (XRD, 1.54 Å, Rikagu Rotaflex RTP300 RC) by scanning 2θ from 30° to 100°. The microstructure of the particles was examined by Raman spectroscopy. Raman spectra were recorded in a backscattering collection geometry using a 100-mW, Melles Griot CW argon ion laser at 514.5 nm, an Olympus BH-2 microscope assembly, and a Spex 1250 spectrograph (600 grooves/mm holographic grating). The detector was a Horiba Jobin Yvon Symphony 2000 × 800 CCD with a pixel size of 15 μm. Spectra were collected with an incident power density of 1.4 kW/cm². The spectrometer was calibrated using the first-order phonon peak of cubic diamond at 1332.6 cm⁻¹. The powder morphology of the particles was probed by field-emission scanning electron microscopy (FE-SEM, JSM-6300F, JEOL, Ltd., Tokyo, Japan).

Electrical resistance measurements of the particles were made before and after overcoating with conducting diamond using an in-house fabricated test fixture. The measurements were carried out by placing a fixed quantity (~30 mg) of the powder between two metal plates and measuring the contact-to-contact resistance with an ohmmeter. The top metal plate was attached to a copper rod (0.114 cm²), which fit inside the glass tube used to contain the powder sample between the two plates. A 240-g weight was placed on the top metal plate to ensure that a constant force was applied during the measurement. The powder resistance was calculated by using Ohm’s law as a function of three different dc currents (±6, ±10, and ±20 mA). All powder resistances prior to doped diamond growth were >40 MΩ. Ohmic behavior was observed after coating. The electrical conductivity of the BDDP was 2.4 S/cm.

2.2. Chromatographic column construction

The design of the EMLC column has been described elsewhere[11]. Briefly, the column consists of a Nafion cationic-exchange membrane in a tubular form (Perma Pure) that is placed inside a porous stainless steel cylinder. The Nafion membrane serves as a container for the stationary phase. The stainless steel cylinder prevents the deformation of the Nafion tubing under the high pressure of chromatographic flow and acts as the auxiliary electrode in a three-electrode electrochemical cell. BDDP was dispersed in a dibromomethane/acetonitrile (1:1, v/v) slurry and then packed in an EMLC column at 6000 psi using methanol for ~30 min. The same column and packing procedures were used to prepare a PGC (5-μm particles) column. The length and the inner diameter of the column were 7.8 and 0.3 cm, respectively. The actual surface area of the BDDP and PGC stationary phase is 1.0 and 18 m², respectively. These values were calculated based on the BET adsorption measurements (1.0 m²/g for BDDP and 120 m²/g for PGC), and the amount of material loaded into the column (1.0 g for BDDP and 0.15 g for PGC).

2.3. Instrumentation

Chromatographic experiments were performed using an Agilent 1200 Series system, equipped with solvent cabinet, autosampler, quaternary pumping system, and UV/Vis diode array detector. This unit was interfaced to a Pentium IV 600 MHZ computer, equipped with Chemstation software for control of injection sequences, data acquisition, and flow rate parameters. The potential of the working electrode/chromatographic packing was controlled using an AMEL Instruments Model 2055 high power potentiostat (Milan, Italy).

2.4. Mode of operation

After packing, both EMLC columns were equilibrated with degassed mobile phase (0.10 M LiClO₄) at a flow rate of 0.40 mL/min until a stable detector response was obtained (~30 min). After each change in E_appl, the system was allowed to achieve a stable baseline, which typically required ~10 min for a BDDP-packed column and ~30 min for a PGC-packed column.

Injection volumes were set at 5 μL, and injections of the analyte mixtures were repeated four times at each value of E_appl. Individual injections of each analyte were performed in triplicate for peak identification and for retention time determination when elution bands overlapped. Injections of water (5 μL) were used to determine the void volume of the column for calculation of k′. The void volume represents the interstitial (both interparticle and intraparticle) volume of the column [44]. The void volume was 0.28 and 0.41 mL for the BDDP- and PGC-packed column, respectively. The absorbance of all analytes was monitored at 210 nm. All experiments were performed at room temperature.

2.5. Chemicals and reagents

Sodium 4-toluenesulfonate (TS), disodium 1,3-benzenedisulfonate (BDS), sodium benzenesulfonate (BS), dibromomethane, lithium perchlorate, and acetonitrile were purchased from Aldrich. Disodium 1,5-naphthalenedisulfonate (NDS) was acquired from Eastman Kodak. All chemicals were used as received. All aqueous solutions were prepared with HPLC grade water.

3. Results and discussion

3.1. Material characterization

Fig. 1 presents an SEM image of BDDP. Though clearly not ideal from a separation perspective, the image reveals that the particles are irregularly shaped, with sizes ranging between ~8 and 12 μm. The particles possess sharp jagged edges, which hinder efficient packing. The small spots observed on some particles are probably due to the secondary nucleation of diamond on the much larger particles.

Fig. 1.

SEM image of BDDP.

The crystal structure of the BDDP was investigated by XRD. Fig. 2 shows an XRD spectrum for the powder sample. Three reflections are observed at 2θ values of 43.9°, 75.5°, and 91.5°, and are assigned to the (1 1 1), (2 2 2), and (3 1 1) crystallographic planes of cubic diamond, respectively. This conclusion is based on a comparison with a reference diamond material: ASTM 6-0675 [34]. The diffraction data reveal that the bulk film structure is sp³-bonded carbon.

Fig. 2.

XRD pattern of BDDP.

Raman spectroscopy was used to examine this material because of its sensitivity to the microstructure (sp² and sp³ carbon bonding), boron-doping level, and residual stress (intrinsic and thermal) [45–48]. Fig. 3 presents a Raman spectrum for the BDDP. The spectrum shows the distinctive first-order diamond phonon peak at 1330 cm⁻¹. The diamond particles are free of detectable sp² non-diamond carbon impurities, as indicated by the absence of scattering between 1350 and 1580 cm⁻¹. The basis of this conclusion can be qualified by recognizing that the scattering cross-section for graphite is ~50 times larger than that for diamond at visible excitation wavelengths [45,46].

Fig. 3.

Raman spectrum of BDDP.

3.2. Retention as a function of E_appl

The influence of E_appl on the retention of ASFs has previously been examined on both GC and PGC [10–14]. The change in retention generally follows predictions based on electrostatic forces but varies slightly between solutes. These differences reflect a complex mixing of donor-acceptor, dispersive, and solvophobic interactions between the solute and the solid phase [49]. Past work has also found that, if present, the small amounts of oxygen-containing surface groups (e.g., carboxylic acids, quinones, and phenols) can contribute to retention through dipole and hydrogen bonding interactions [50–55].

The importance of the above interactions is likely to differ at BDDP, which is devoid of an extended π-bonding system, and has a hydrogen-terminated surface. Termination renders the surface nonpolar and, therefore, highly hydrophobic [56–59]. In other words, chemisorbed hydrogen replaces carbon-oxygen functionalities on the surface during hydrogenation [60,61]. Oxygen-containing groups can form at surface sites (both diamond and non-diamond carbon), and are usually present at low levels in high-quality hydrogen-terminated diamond thin films [29].

The effect of E_appl on the separation of a four-component ASF mixture (BDS, BS, TS, and NDS) is presented in Fig. 4 for values ranging from −1.2 to +1.2 V (vs. Ag/AgCl, sat’d NaCl) on a BDDP-packed EMLC column. This span in E_appl is much larger than the usable range for PGC, which is roughly from −1.0 to +0.5 V. Above +0.5 V, the PGC surface can slowly oxidize over a few days of continuous use at edge planes and defects sites to form carbon–oxygen surface groups that may be redox-active and/or ionizable [60,61]. These changes lower the reproducibility of retention and the overall effectiveness of the separation.

Fig. 4.

Chromatograms illustrating the effect of E_appl (V vs. Ag/AgCl sat’d NaCl) on the retention of a mixture of BDS (1, 5 μM), BS (2, 5 μM), TS (3, 10 μM) and NDS (4, 2 μM) on a BDDP-packed EMLC column. The mobile phase consisted of 0.10 M LiClO₄ (aq). The flow rate is 0.4 mL/min.

As is evident, the retention of all four analytes is strongly dependent on E_appl. At −1.2 V, the ASFs co-elute with the void volume. However, as E_appl becomes more positive, the retention of some of the analytes begins to extend beyond that of the void volume. At −0.9 V, the k′ values for TS (k′ = 0.34) and NDS (k′ = 0.32) are smaller than those at −0.15 V by factors of 1.7 and 3.6, respectively. The application of more positive values of E_appl results in further increases in retention. At +0.45 V, the k′ values for BDS/BS (BDS and BS were not resolved at any value of E_appl, which we ascribe to the low surface area for the BDDP packing), TS, and NDS are 0.28, 0.84, and 4.57, respectively. These values represent an increase by a factor of 1.3 or more over those at 0 V.

The modulation of ASF retention is summarized in Fig. 5 through ln k′-E_appl curves. The error bars for ln k′ are roughly the size of the data points and represent the average of four replicate separations. The plots of TS and NDS extend from +1.2 to −0.9 V. The curves for BDS/BS cover +1.2 V to 0 V because BS and BDS eluted with the void volume (k′ ≈ 0) at more negative values of E_appl. Each compound follows a linear dependence, which is consistent with our past findings, and reflects the role of electrostatics on the EMLC-based manipulation of adsorption [10]. Though not yet understood, the slope of the plots in Fig. 5 do not all follow the observed dependence found when using PGC as a packing in EMLC [11]. In particular, the slopes for BDS and BS are indistinguishable from each other despite the difference in ionic charge. This observation, which suggests a possible mechanistic difference in electrosorption phenomenon at the two materials, merits further investigation.

Fig. 5.

Plots of ln k′ vs. E_appl for BDS (5 μM), BS (5 μM), TS (10 μM) and NDS (2 μM). Four replicate injections were made at each E_appl, and the data points are roughly the size of the error bars.

Together, these data demonstrate the ability of EMLC to alter molecular adsorption over a wide potential range when using BDDP as a stationary phase. This ability is highlighted by the retention of NDS, which undergoes an increase in k′ by a factor of ~53 between −1.2 and +1.2 V. The next subsection briefly details the findings from a study of retention reproducibility and packing stability.

3.3. BDDP stability

The stability of BDDP was examined by checking the day-to-day reproducibility of retention times of the four ASFs at several values of E_appl, especially above +0.5 V. This study was accomplished by calculating the relative standard deviation (%RSD) of eight runs obtained in 2 days (four consecutive injections of the mixture on each day) at different E_appl values. Retention times typically exhibited less than a 5% RSD for ASFs from day-to-day comparison. For example, the RSDs of TS and NDS at +0.8 V for 2 consecutive days were 3.6 and 1.6%, respectively. This finding demonstrates that BDDP microstructure remains comparatively stable over a wide range of E_appl values. The basis for this conclusion arises from our earlier experiments using PGC. Excursions to values of E_appl in excess of +0.5 V often led to decreases in retention. This situation proved particularly problematic when carrying out EMLC separations at elevated temperatures in order to measure thermodynamic parameters (e.g., enthalpy and entropy) of adsorption. In these cases, the retention times of the ASFs decreased as much as 20% [62]; characterization by X-ray photoelectron spectroscopy revealed that these decreases (both in room and elevated temperature studies) corresponded to an increase in the amount of oxygen-containing surface functional groups, which, at least quantitatively render the carbon surface less hydrophobic. Thus, the morphological and microstructural stability of BDDP point to its potential use as a model material to study the thermodynamics of electrosorption phenomenon (e.g., enthalpy and entropy of adsorption) by examining the dependence of retention with respect to temperature and pressure.

3.4. Comparison of BDDP to PGC

To develop a very brief comparison of the molecular adsorption characteristics of BDDP, a separation of a mixture of BDS, BS, TS, and NDS using 0.10 M LiClO₄ (aq) was carried out at different values of E_appl (+0.3, +0.1, and 0.0 V) using a PGC-packed column. As reported previously [10,14], the time for all compounds to elute increased with E_appl. Fig. 6 shows a chromatogram of the ASFs using the PGC-packed column at +0.3 V. The analysis took ~65 min for BDS, BS, and TS to elute from the PGC-packed column, but only ~3 min from the BDDP-packed column (Fig. 4). NDS did not detectably elute from the PGC-packed column after ~80 min at any value of E_appl; longer times were not tested. The long elution times in the aqueous mobile phase are consistent with stronger interactions of the π-system of the analytes with PGC, which in nearly all cases dictates the use of a mixed (e.g., acetonitrile/water) mobile phase in order to decrease retention [1].

Fig. 6.

Chromatogram showing the retention of a 20 μM mixture of BDS (1), BS (2), and TS (3) on a PGC-packed EMLC column. The E_appl was +0.3 V (vs. Ag/AgCl sat’d NaCl). The mobile consisted of 0.10 M LiClO₄ (aq). The flow rate is 0.4 mL/min.

Table 1 compares the electrosorption of TS on BDDP- and PGC-packed columns at the three values of E_appl after accounting for the difference in the absolute surface area of the packings and void volume of the two different columns. This approach develops via Eq. (1), which defines the partition coefficient, K, for a chromatographic column as

Table 1.

Comparison of retention of 4-toluenesulfonate at BDDP and PGC-packed columns^a,^b

Eappl (V)b	k′P	k′B	Kratio
+0.30	54.83	0.79	5.6
+0.15	46.66	0.70	5.4
0	37.49	0.60	5.1

^aAverage values from four replicate injections. The values for k′ have less than a 2% RSD.

^bE_appl given with respect to Ag/AgCl (sat’d NaCl). Data for benzene sulfonate and 1,3-benzene disulfonate also showed a comparable qualitative trend, but are not included in the tabulation because of the small changes in their k′ values.

$$K=k^{\prime }\frac{V_{\text{mp}}}{A_{\text{sp}}}$$ (1)

where V_mp is the volume of the mobile phase and A_sp is the surface area of the column packing. Thus, by dividing K_P by K_B, the impact of V_P, V_B, A_P, and A_B can be taken into consideration to give K_ratio:

$$K_{\text{ratio}}=\frac{K_P}{K_B}$$ (2)

where the subscripts B and P link the appropriate variables to either BDDP or PGC, respectively. Based on this treatment, the data in Table 1 show that TS is retained more strongly on a PGC-packed column by at least a factor of five when compared to BDDP. While much more in-depth investigation is needed and planned, we presently attribute the differences in retention strength, at least in part, to the importance of π-system interactions in the retention at PGC.

4. Conclusions

In this paper, we presented initial findings on the performance of BDDP as a stationary phase in EMLC separations. The work herein demonstrates the utility of EMLC for fine-tuning separations with BDDP. Due to the inert nature and a wide potential window of conductive diamond, EMLC separations were carried out at high positive and negative potentials without observable limitations due to surface oxidation and the microstructural changes associated with it, or solvent electrolysis common with other carbon-based materials. These data also argue that the morphological and microstructural stability of BDDP support its potential use as a model material to study the thermodynamics of electrosorption phenomenon (e.g., enthalpy and entropy of adsorption) by examining the dependence of retention as a function of temperature and pressure.

We are presently pursuing the creation of conductive diamond coated particles with higher surface area, and the results from EMLC experiments using these particles will be reported on in future work [30]. These studies will focus particularly on measurements of interfacial excesses and the potential of zero charges, both of which are central to engineering insights into the many chemical factors of importance to molecular adsorption.