Article pubs.acs.org/Langmuir
Adsorption of Superparamagnetic Iron Oxide Nanoparticles on Silica and Calcium Carbonate Sand Yoonjee C. Park,† Jeffrey Paulsen,‡ Rikkert J. Nap,§ Ragnhild D. Whitaker,† Vidhya Mathiyazhagan,∥ Yi-Qiao Song,‡ Martin Hürlimann,‡ Igal Szleifer,§ and Joyce Y. Wong*,† †
Department of Biomedical Engineering, Boston University, Boston, Massachusetts 02215, United States Schlumberger-Doll Research, Cambridge, Massachusetts 02139, United States § Department of Biomedical Engineering and Chemistry of Life Processes Institute, Northwestern University, Evanston, Illinois 60208, United States ∥ Department of Chemistry, Boston University, Boston, Massachusetts 02215, United States ‡
ABSTRACT: Superparamagnetic iron oxide (SPIO) nanoparticles have the potential to be used in the characterization of porous rock formations in oil fields as a contrast agent for NMR logging because they are small enough to traverse through nanopores and enhance contrast by shortening NMR T2 relaxation time. However, successful development and application require detailed knowledge of particle stability and mobility in reservoir rocks. Because nanoparticle adsorption to sand (SiO2) and rock (often CaCO3) affects their mobility, we investigated the thermodynamic equilibrium adsorption behavior of citric acid-coated SPIO nanoparticles (CA SPIO NPs) and poly(ethylene glycol)-grafted SPIO nanoparticles (PEG SPIO NPs) on SiO2 (silica) and CaCO3 (calcium carbonate). Adsorption behavior was determined at various pH and salt conditions via chemical analysis and NMR, and the results were compared with molecular theory predictions. Most of the NPs were recovered from silica, whereas far fewer NPs were recovered from calcium carbonate because of differences in the mineral surface properties. NP adsorption increased with increasing salt concentration: this trend was qualitatively explained by molecular theory, as was the role of the PEG grafting in preventing NPs adsorption. Quantitative disagreement between the theoretical predictions and the data was due to NP aggregation, especially at high salt concentration and in the presence of calcium carbonate. Upon aggregation, NP concentrations as determined by NMR T2 were initially overestimated and subsequently corrected using the relaxation rate 1/T2, which is a function of aggregate size and fractal dimension of the aggregate. Our experimental validation of the theoretical predictions of NP adsorption to minerals in the absence of aggregation at various pH and salt conditions demonstrates that molecular theory can be used to determine interactions between NPs and relevant reservoir surfaces. Importantly, this integrated experimental and theoretical approach can be used to gain insight into NP mobility in the reservoir.
1. INTRODUCTION
Superparamagnetic iron oxide (SPIO) nanoparticles have been used successfully as magnetic resonance imaging (MRI) contrast agents in biomedical applications because they are relatively easily produced and have excellent nuclear magnetic resonance (NMR) T2 contrast at low concentrations.3 These contrast agents are detected indirectly through their effects on the signal of the bulk fluid, shortening the relaxation time of NMR signal decay.4,5 Furthermore, nanoparticle quantification is possible as the change in relaxation rate ΔR2 (reciprocal of the relaxation times, 1/T2) is proportional to nanoparticle concentration.4,5 Our long-term goal is the use of the nanoparticles as NMR nanosensors in oil fields, where we envision NMR relaxometry directly either in the formation at the well bore6 or in extracted fluid samples at the well site. In other words, the nanoparticles are transported to an arbitrarily
In order to meet rising global energy demand and consumption, it is critical to achieve sustained production and increased oil recovery from existing oil fields. Despite the efforts of enhanced oil recovery using techniques such as water, chemical, gas, or thermal recovery, 60% of oil still remains in reservoir rocks.1 In these cases, additional economic extraction of the remaining oil is frequently limited by a lack of knowledge of the particular reservoir. It is challenging to characterize interwell matrix and fracture properties because conventional applied deep sensing techniques such as seismic and crosswell electromagnetics yield very limited details about the formation, while micrometer-scale sensors are larger than the typical throat size range of conventional reservoir rock (0.03−10 μm).2 On the other hand, nanoscale sensors have the potential to provide improved reservoir characterization, which allows for more efficient enhanced oil recovery and better oil field development with reduced cost. © 2014 American Chemical Society
Received: November 14, 2013 Published: January 6, 2014 784
dx.doi.org/10.1021/la404387t | Langmuir 2014, 30, 784−792
Langmuir
Article
Figure 1. Recovery percentages of CA SPIO NPs (left panels) and PEG SPIO NPs (right panels), determined by UV−vis absorbance of iron concentration in supernatant extracted from silica sand after 2 and 24 h at different pH and salt conditions. The salt concentration for top panels is 10 mM, and the pH value for bottom panels is 7.5.
which influences nanoparticle mobility in porous media.9 Citric acid-coated iron oxide nanoparticles (CA SPIO NPs) and poly(ethylene glycol) (PEG)-grafted nanoparticles on the CA SPIO NPs (PEG SPIO NPs) were used to study the effect of the nanoparticle surface coating of the nanoparticles on the adsorption. Surface coating of iron oxide nanoparticles highly affects their adsorption behavior. Bagaria et al. have reported that the degree of adsorption of iron oxide nanoclusters can be controlled by the nature of the surface coating of the nanocluster in brine conditions.10,11 In order to predict the adsorption behavior of the nanoparticles, the interaction between the nanoparticles and the sand surface was calculated using a molecular theory (described in ref 12) that explicitly incorporates the molecular details of each of the different species in the system treated. Results from characterization of the nanoparticles, such as size and surface coating coverage, were used as inputs for the molecular theory. The theoretical predictions were then compared to the experimental results (UV−vis and NMR measurements). The long-term goal of this study is to elucidate and understand the fundamental mechanisms controlling nanoparticle mobility or retention in the oil reservoir and to iterate and optimize nanoparticle design based on our findings.
distance within the reservoir rock containing two immiscible fluids, and their signal results in illuminating the reservoir. One can also monitor the interwell changes in fluid properties that occur as the reservoir is developed and produced. In order to achieve the goal of enhancing reservoir characterization, the nanoparticles should have high mobility. This can be attained by achieving significant colloidal stability of the nanoparticle dispersion and by minimizing adhesion to the mineral surfaces under conventional reservoir environments. The conventional reservoir conditions range from pH 6.5 to 8.5 and salinity from 50 000 to 150 000 ppm (850 mM to 2.5 M of a NaCl solution), with approximately 2 wt % divalent ions such as Ca2+.7 In addition, basic conditions of pH 11−12 are required for hydraulic fracturing of low permeability formations. The study of nanoparticle stability is critical because changes such as aggregation or alterations to the coating will modulate or hinder their transport through a porous rock formation. In a previous study, theoretical and experimental study of the colloidal stability of the SPIO nanoparticles at various pH conditions from pH 5 to 11,8 we concluded that the SPIO nanoparticles have the potential to be used to characterize the rock formation at the conventional reservoir conditions, pH 6.5 to 8.5, over relevant time scales (30 days). There are many factors to consider, such as transport, sedimentation, or diffusion, to understand particle mobility in porous structure. The purpose of our study here is to isolate two factors, thermodynamic adsorption and sedimentation, in order to understand those effects on the particle mobility. Sedimentation is mainly due to particle aggregation, which we studied previously.8 In addition to the previous study, here, we examine the adsorption behavior of the nanoparticles to silica and calcium carbonate sand, the main composition of common reservoir rocks. Adsorption occurs due to the attractive interactions between the nanoparticle and the sand surface,
2. METHODOLOGY 2.1. Materials. Iron tri(acetylacetonate) (99.9%), citric acid (CA, 99.5+%), methanol (99.8%), and acetone were purchased from Sigma (St. Louis, MO). Benzyl ether (99%), N-hydroxysuccinimide ester (NHS ester, 98%), oleic acid (90%), and oleylamine (70%) were purchased from Aldrich (St. Louis, MO). 1,2-Dichlorobenzene (DCB, 99%), N,N′-dimethylformamide (DMF, 99.8%), diethyl ether (99.9%), and hexane (99.9%) were purchased from Acros (Morris Plains, NJ). 1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC, 97%), was purchased from Fluka. Ethanol (ACS grade) was purchased from Pharmco (Lees Summit, MO). Amino end-functionalized poly(ethylene glycol) (NH2-PEG) with molecular weights of 2000 785
dx.doi.org/10.1021/la404387t | Langmuir 2014, 30, 784−792
Langmuir
Article
provide very good agreement with experimental observation, including the structure and charge state of grafted poly(acrylic acid) layers15 and effective charged state of ligand modified gold nanoparticles16 and the electrochemical behavior of electrodes modified with redox polybases.17 We previously applied the molecular theory to describe the interactions between two NPs both coated with either citric acid molecules or PEG polymer molecules.8 The predictions of the theory concerning the stability of NP solutions was found in good agreement with the experimental observations, which showed the ability of the theory to account for the essential features of the system such as bulk pH and solution ionic strength. In the accompanying paper,12 we continued the development of the molecular theory to describe the adsorption of the coated nanoparticles onto mineral surfaces. Briefly, the model takes into account the polymer conformations of the PEG coating and the charge regulating ability of the citric acid molecule as well as the finite concentration of the nanoparticles in solution. Derivation and technical details of the molecular theory used to study NP adsorption are presented in the accompanying paper. Here we will only present relevant results concerning the understanding of the NP adsorption. The adsorption of the NPs onto surfaces is theoretically given as Γ
Da was purchased from Laysan Bio, Inc. (Arab, AL). Silica pure sand with mesh size 40−100 (0.152−0.422 mm) was purchased from Acros Organics (Geel, Belgium). Calcium carbonate (average diameter = 1.7 μm, assuming spherical particles, polydispersity = 60%) was obtained from Schlumberger. The size distribution of the calcium carbonate particles was characterized optically using a phase contrast microscope (Axiovert 25, Zeiss, Oberkochen, Germany) and ImageJ software. The polydispersity, σ, was calculated as σ = δ/davg × 100%, where δ is the standard deviation and davg is the sand particle average diameter. Thioglycolic acid, hydroxylamine, phenanthroline, and sodium citrate were purchased from Sigma-Aldrich (St. Louis, MO). Six-well plates were purchased from Corning (Lowell, MA). Water used for all experiments was distilled by Myron L Company Series 750 (Carlsbad, CA) and had a resistivity of 16 MΩ at 25 °C. 2.2. Synthesis of CA SPIO NPs and PEG SPIO NPs. The citric acid-coated SPIO nanoparticles (CA SPIO NPs) and the PEG-grafted SPIO nanoparticles (PEG SPIO NPs) were synthesized8 according to the method of Sun et al.13 and Lattuada et al.14 Briefly, iron(III) tri(acetylacetonate) [Fe(acac)3] (2 mmol), benzyl ether (40 mL), 1,2tetradecanediol (10 mmol), oleic acid (6 mmol), and oleylamine (6 mmol) were mixed and stirred magnetically under a flow of nitrogen. The mixture was heated at 2 °C/min to 100 °C and kept for 45 min, followed by heating to 200 °C at 2 °C/min and kept for 2 h. Subsequently, the reaction mixture was heated to reflux (∼300 °C) and held for another 30 min to 1 h. The reaction mixture was cooled to room temperature and washed with ethanol, and precipitates were collected using a magnet. The precipitates were dissolved in hexane (10−15 mL) and centrifuged at 3000 rpm for 5 min. The oleic acidcoated iron oxide nanoparticles were washed in an excess of ethanol and then dried in a vacuum oven before storage. In order to obtain water-soluble NPs, oleic acid capping of the NPs was exchanged with citric acid. The oleic acid-coated nanoparticles were dissolved in a DCB/DMF mixture, citric acid was added, and the mixture was heated to 100 °C for 24 h. Diethyl ether was added upon cooling the mixture to room temperature in order to precipitate light brown particles. Finally, citric acid-coated iron oxide nanoparticles (CA SPIO NPs) were extracted upon washing with acetone (10 mL) and diethyl ether several times to remove unreacted citric acid. The CA SPIO NPs were further functionalized with amineterminated poly(ethylene glycol) (PEG) 2000 Da. PEG was grafted on to the surface using N-hydroxysuccinimide (NHS) ester and 1-ethyl-3(3-(dimethylamino)propyl)carbodiimide (EDC) in pH 9 for 24 h on an agitator at room temperature. For 50 mg of CA SPIO NPs in 10 mL of a pH 9 solution, 50 mg of EDC, 60 mg of NHS ester, and 10−5 mol of each PEG were used. The PEG SPIO NPs were obtained after they were dialyzed for 48 h in pH 9 water. The characterization of the CA SPIO NPs and the PEG SPIO NPs, such as size distribution, elemental composition, and surface charge, is introduced in ref 8. Briefly, the CA SPIO NPs and the PEG SPIO NPs with MW 2000 of PEG (PEG 2K SPIO NPs) consisted of iron oxide cores of 6−11 and 5−11 nm in diameter, respectively, based on TEM images (Figure 1). Average diameters from the TEM images for CA SPIO NPs and PEG 2K SPIO NPs were both 8 nm. The hydrodynamic effective diameter of CA SPIO NPs, measured by dynamic light scattering (DLS) with intensity-weighted, was 60 ± 2 nm, and the size range between 20% and 80% of the total distribution calculated based on Gaussian distribution was 52−68 nm. The zetapotential value of CA SPIO NPs was −8 ± 1 mV at pH 7. The density of the CA or PEG coatings on the nanoparticles, characterized by CHN combustion analysis, was 4.33 molecules of CA per nm2 and 1.42 molecules of PEG 2K per nm2. 2.3. Adsorption Behavior of SPIO NPs. 2.3.1. Molecular Theory. The theoretical approach used to study NP adsorption is a generalization of a molecular theory, which explicitly incorporates the molecular details of each species in the system treated. Hence, the size, shape, conformation, and charge distribution of every molecule type are accounted for exactly. The theory has originally been developed to describe various interfacial polymer systems and was extended to study the dispersion of polymer-coated carbon nanotubes and the behavior of coated weak polyelectrolyte layers.15 The theory has been shown to
Γ=
∫0
∞
dz (ρ(z) − ρ bulk )
(1)
which represents the excess density of molecules at the surface as compared to the bulk concentration. 2.3.2. Experimental Conditions for Recovery Percentage of SPIO NPs. Prior to using the silica sand, it was soaked in 25% HCl for 24 h in order to remove nanosized sand particles from the pores of sand particles. Then, the HCl-treated sand was subsequently rinsed with DI water until the pH is the same as the DI water and dried at 80 °C for 24 h. 2 g of the sand was spread evenly across the bottom of a 6-well plate, and 1 mL of 200 ppm of the SPIO NPs in each buffer was applied to the layer of sand. This results in about 1 mm in combined height of sand and water. The buffer solutions tested were 10 mM phosphate solutions at pH 5, 7, 9, and 11 for the pH series and 10, 50, 100, 150, 500, and 850 mM at pH 7.5 for the NaCl salt concentration series. The SPIO NP dispersions in the buffer solutions were prepared immediately before the dispersions were applied to the sand. After 2 or 24 h, the supernatant was extracted without disturbing the surface of the sand. The extracted supernatant was analyzed both by UV−vis absorbance and by NMR 1/T2 relaxivity to obtain recovery percentage of the SPIO NPs. Measurements were performed three or more times for each sample. For calcium carbonate sand, pretreatment was not performed, and 1 g was used because it is finer and its bulk density lower than the silica sand. 2.3.2.1. UV−vis Absorbance. Thioglycolic acid (2 μL) was added to the extracted supernatant solution (100 μL) from the adhesion tests as described in section 2.3.2 to reduce iron (Fe3+) in Fe3O4 to Fe2+. After 2 h, 200 μL of 10% hydroxyl amine, 300 μL of 0.25% phenanthroline, 15 μL of 2.5% sodium citrate, and DI water up to 1 mL were added, resulting in a compound [Fe(Phen)3]2+. The concentration of Fe2+ was measured by UV−vis absorbance at 510 nm. The recovery percentage was determined by the [Fe2+ of supernatant (ppm)]/[Fe2+ of initial SPIO NP solution (ppm)] × 100 (%). 2.3.2.2. NMR Relaxation Measurements. All NMR measurements were performed on the extracted solutions to analyze the recovery percentage from sand. Because of variability in the NMR response of different particle types, each set of measurements for a particular particle and batch included a standard concentration series spanning 0 to 175 ppm by mass from the same batch. To maintain adequate signal-to-noise ratio (SNR) from the small sample volumes, low field NMR measurements were not used, and instead measurements were performed on an available 1 T (42.58 MHz) horizontal bore magnet. Only T2 measurements with a standard CPMG sequence were acquired.4 NMR relaxation measurements were collected in the target oil field application and were sensitive enough to detect the nanoparticles at the relevant concentrations (
