Article pubs.acs.org/ac
In Situ Determination of Colloidal Gold Concentrations with UV−Vis Spectroscopy: Limitations and Perspectives Thomas Hendel, Maria Wuithschick, Frieder Kettemann, Alexander Birnbaum, Klaus Rademann,* and Jörg Polte* Institut für Chemie, Humboldt-Universität zu Berlin, Brook-Taylor-Strasse 2, 12489 Berlin, Germany S Supporting Information *
ABSTRACT: This paper studies the UV−vis absorbance of colloidal gold nanoparticles at 400 nm and validates it as a method to determine Au(0) concentrations in colloidal gold solutions. The method is shown to be valid with restrictions depending on the investigated system. The uncertainty of the determined Au(0) concentration can be up to 30%. This deviation is the result of the combined influence of parameters such as particle size, surface modification, or oxidation state. However, quantifying the influence of these parameters enables a much more precise Au(0) determination for specific systems. As an example, the reduction process of the well-known Turkevich method was monitored and the Au(0) concentration was determined with a deviation of less than 5%. Hence, a simple, fast, easy, and cheap in situ method for Au(0) determination is demonstrated that has in the presence of other gold species such as Au(III) an unprecedented accuracy.
M
From SAXS measurements of GNPs, the volume fraction measured in absolute scale can be converted into a molar concentration of Au(0).20,32 However, this evaluation is only suitable for noninteracting nanoparticles and for common SAXS setups for which nanoparticle size should not exceed 20 nm in radius. X-ray absorption near-edge structure spectroscopy (XANES) delivers information about absorption edge positions and thus oxidation states. It can be used to calculate the ratio of Au(0) and Au(III) in solution. From this ratio, the evolution of molar Au(0) concentration during a GNP synthesis can be investigated if the initial concentration of the gold precursor is known and a complete reduction is assured.25,33 The drawback of XANES is its relatively long measurement time, ranging from several minutes to hours. A further approach for X-ray absorption spectroscopy (XAFS) with time resolutions in the millisecond range is to apply a time-resolved energy dispersive XAFS technique (DXAFS).34,35 For DXAFS experiments, X-rays are polychromatized with a bent crystal, leading to an angle-dependent primary beam. X-rays in the required energy range are monitored simultaneously by a positionsensitive detector.36 Nevertheless, X-ray absorption spectroscopy demands synchrotron radiation. Voltammetric methods have been applied to determine the Au(0) concentration in GNP solutions with very low detection limits and relatively high robustness against organic molecules on the GNPs.37−40
etal nanoparticles have become an excelling research field with a broad variety of applications.1,2 Among them, gold nanoparticles (GNPs) play an outstanding role, representing one of the most synthesized and intensely studied materials.3,4 Synthetic pathways are manifold, resulting in GNPs of different sizes, shapes and surface functionalities5−7 that are applied in photonics,8 surface-enhanced Raman spectroscopy,9 catalysis,10,11 and biomedicine.12 The morphology and size distribution of GNPs have been studied using a variety of analytical techniques, such as electron microscopy,13,14 scanning probe microscopy,15,16 dynamic light scattering (DLS),17 mass spectroscopy,18,19 small-angle X-ray scattering (SAXS), 20,21 and X-ray diffraction (XRD). 22 However, an exact determination of the Au(0) concentration in a GNP solution remains a major issue. The Au(0) concentration represents important information of a GNP solution. Its exact determination is crucial for biological applications and their related toxicological discussions.23,24 Moreover, for investigations of GNP formation mechanisms or various GNP surface modifications and particle assembling procedures, accurate Au(0) concentration values are required.25,26 The optimal determination of Au(0) concentrations would be a fast and easy procedure that requires only standard laboratory equipment and can be carried out in situ. Different analytical techniques have been applied to determine the Au(0) concentration in GNP solutions.27,28 Very low detection limits can be achieved with inductively coupled plasma mass spectroscopy (ICP-MS) analysis,29 for which GNP solutions can be directly injected into the torch30 or dissolved previously in aqua regia.31 © 2014 American Chemical Society
Received: June 2, 2014 Accepted: October 7, 2014 Published: October 7, 2014 11115
dx.doi.org/10.1021/ac502053s | Anal. Chem. 2014, 86, 11115−11124
Analytical Chemistry
Article
that in these publications, the GNPs have different sizes, shapes, and surface modifications. This paper evaluates the determination of the absolute Au(0) concentration in aqueous GNP solution applying the Abs400. At first, the general applicability of that method was studied. Subsequently, the influences of significant characteristics of colloidal GNP solutions on the Abs400 were investigated. This includes particle size, surface modifications, and oxidation state. For size-dependent investigations, two common GNP syntheses were used to obtain a wide size range. GNPs with mean radii of 1−2.5 nm were synthesized by reduction of HAuCl4 with NaBH4 (GNPBH4).21 The classical Turkevich method (HAuCl4 reduction with sodium citrate at elevated temperature) with and without subsequent seeded growth results in GNPs with mean radii between 5 and 28 nm (GNPcit).14,63 Surface modification was carried out with common stabilizing agents [i.e., poly(vinylpyrrolidone) (PVP) and Pluronic F-127] and thiol ligands [i.e., mercaptosuccinic acid (MSA) and 1-dodecanethiol (DT) in toluene]. To validate the selectivity of the Abs400 for Au(0) concentrations, mixtures of GNP with gold precursor HAuCl4 [i.e., Au(III)] were investigated. The interpretation of the resulting spectra demands an analysis of the pH-dependent HAuCl4 absorption and the precursor interaction with GNPs.
From atomic spectroscopy such as atomic absorption spectroscopy (AAS)23,41 or optical emission spectroscopy (OES),42−44 elemental gold concentrations can be obtained. Typically, GNP solutions are dissolved in strong acids before being analyzed. A powerful technique for the elemental analysis in GNP solutions is neutron activation analysis (NAA).41 The crucial drawback is the complex instrumentation, since a research reactor as neutron source is needed for results with high accuracy. In contrast, UV−vis spectroscopy represents a relatively cheap and easy characterization technique that is commonly applied in research and industrial laboratories. GNP solutions exhibit a characteristic UV−vis extinction spectra due to the presence of a localized surface plasmon resonance (LSPR) signal in the visible part of the spectrum.26,45 It was reported in literature that the size of GNPs and the molar concentration of Au(0) can be extracted directly from UV−vis spectra. The approaches use either the position of the LSPR as well as the extinction at this wavelength46,47 or the ratio of extinctions at the wavelength of the LSPR and at 450 nm (ALSPR/A450).48 In our opinion, the wavelength and the extinction at the LSPR should only be used for comparison of size and concentration between morphologically and chemically similar GNP solutions, because the position and absorbance of the LSPR are strongly influenced by further parameters, such as shape and size distribution,49,50 the dielectric surroundings, or the particle surface chemistry.49,51 Measurement of UV−vis absorbance at a wavelength of 400 nm (Abs400) could enable an accurate determination of Au(0) concentration, since this wavelength is located between two decisive spectral regions. Below 400 nm, the absorbance is increasingly influenced by organic substances. Common stabilizers for GNPs are colorless and thus do not show an absorption at 400 nm. In contrast, the Abs400 of dyefunctionalized GNPs52 might not be suitable for a gold concentration determination. Above 400 nm, the absorbance is increasingly influenced by the LSPR of the GNPs. The dipolar LSPR mode of GNPs located at around 520 nm might contribute to the Abs400 but only for the smallest GNPs and only to a very small extent.53,54 LSPRs with higher modes (e.g., quadrupoles) are located at around 560 nm but contribute to the spectra only for GNPs with diameters above 100 nm, which are not within the scope of this publication.55 Thus, it can be assumed that at 400 nm both organic molecules and LSPRs have only a minor effect on the absorbance of GNP solutions. The photon energy at 400 nm (3.1 eV) is in the range of interband transition energies from 5d to 6sp in bulk gold.56,57 Moreover, the band structure of bulk gold reveals that direct interband transitions at room temperature are allowed at 3.1 eV (i.e., transitions with preserved k-vector between states below and above the Fermi level).56,57 Thus, the Abs400 would scale linearly with the Au(0) concentration and could enable its direct determination if the band structure and transition probability does not change with nanoparticle size, surface chemistry, etc. Obviously, for nanoparticles or clusters this is rather speculative. Thus, the Abs400 as a measure for Au(0) concentrations in colloidal solutions needs to be validated, which to the best of our knowledge has not been described yet. However, UV−vis spectra of GNPs with equal Au(0) concentration are normalized at 400 nm occasionally and in general without further explanation.44,55,58−62 It has to be noted
■
EXPERIMENTAL DETAILS Materials. Tetrachloroauric acid (HAuCl4·3H2O, 99.9%), sodium citrate (C6H5O7Na3·2H2O, 99%), sodium borohydride (NaBH4, 99.99%), Pluronic F-127 (ethylene oxide−propylene oxide block copolymer), and DT (C12H26S, 98%) were purchased from Sigma-Aldrich. PVP was supplied by Alfa Aesar (MW 10 kDa, 40 kDa) and Acros (MW 58 kDa). MSA (C4H6O4S, 98%) was supplied by Alfa Aesar. Aqueous solutions were prepared using ultrapure water (18.2 MΩ·cm, Millipore). Characterization. UV−vis measurements were carried out using an Evolution 220 spectrophotometer (Thermo Scientific). Time-resolved UV−vis spectra of the GNPBH4 were recorded using a USB2000 spectrometer (Ocean Optics) including an ILX511B CCD detector (Sony). A DT-MINI-2GS deuterium/tungsten halogen lamp (Ocean Optics) was connected via fiber optics. The pH values were measured with an HI 2211 pH meter equipped with a glass electrode (Hanna Instruments). The particle size distribution of final colloidal solutions were investigated using a lab-scale SAXS instrument (SAXSess, Anton Paar GmbH). The final colloidal solutions were extracted from the batch mix and inserted in a flow cell. A PVP solution was added to the colloids to prevent aggregation during the measurement.64 The scattering curves of the colloidal solutions were analyzed by assuming spherical shape, a homogeneous electron density, and a Schulz−Zimm size distribution. For details about the analysis of SAXS curves, see our former publications.64,65 Colloidal Synthesis. Syntheses of GNPBH4 were adapted from a procedure reported by Polte et al.21 The GNPcit were synthesized on the basis of the procedure described by Turkevich et al.63 According to the literature, a variation of the mean radii of GNPcit was achieved.66 The preparation of larger GNPcit followed a seeded-growth approach previously reported.14 Details for all colloidal syntheses can be found in the Supporting Information. Time-Resolved Measurements. For time-resolved UV− vis measurements during the synthesis of the GNPBH4, a 11116
dx.doi.org/10.1021/ac502053s | Anal. Chem. 2014, 86, 11115−11124
Analytical Chemistry
Article
stopped-flow setup was applied. Equal flow rates of a HAuCl4 solution (0.5 mM) and freshly prepared NaBH4 solution (2 mM) were obtained by applying an inhouse-built syringe pump. The two solutions were injected into an inhouse-built Yshaped mixer (PTFE), and the homogeneous reaction flow was connected to a flow cell (d = 10 mm, 176.700-QS, Hellma). When the reaction flow was stopped, UV−vis data acquisition was started in the flow cell. The dead time between the mixer and flow cell was 280 ms. According to literature for in situ SAXS measurements of GNPcit synthesis, time-resolved UV−vis measurements were carried out at 75 °C.25,33 Aliquots were taken every 30 or 60 s and immediately cooled using an ice bath. UV−vis spectra were measured as soon as the samples reached room temperature. Modification of the GNPs. For the functionalization with steric stabilizers, stock solutions of different PVPs (10, 40, and 58 kDa) and of Pluronic F-127 were prepared with concentrations of 50, 100, and 250 mg/L. Equal volumes of GNP and stabilizer solutions were mixed rapidly. As a reference, equal volumes of water and the GNP solution were mixed. For functionalization with MSA, equal volumes of asprepared GNPBH4 and aqueous MSA solution (1 mM) were mixed rapidly. The spectrum of MSA-capped GNPs has been doubled for comparison with spectra of as-prepared GNPBH4. For the phase transfer of GNPs from water to toluene, equal volumes of an as-prepared GNPBH4 solution and a solution of DT in toluene (2.5 mM) were added to a glass vial. The mixture was vortexed for 30 s and stored overnight to ensure complete phase separation. For the water or HAuCl4 addition experiments, small portions of the respective solutions were mixed with the GNP solution in a cuvette equipped with a micro stirring bar and UV−vis measurements were carried out subsequently.
■
Figure 1. Dependence of the Abs400 on a varied precursor concentration both for the GNPBH4 (black) and the GNPcit (red).
However, comparing the Abs400 of both syntheses at equal concentrations, those of the GNPcit are constantly higher. Later in this contribution, this will be attributed to the influence of GNP radius on Abs400. To meet the criteria for a precise determination of the Abs400 of colloidal gold solutions, possible error sources of the synthetic procedures and the accuracy of lab instrumentations were checked carefully. The syntheses for GNPBH4 and GNPcit differ significantly in their preparation details. Several precautions were taken to avoid preparative mistakes resulting in a deviation of final Au(0) concentrations in solution and thus a deviation of the Abs400. In general, the synthesis of GNPBH4 occurs on a time scale of a few seconds and is performed at room temperature. Thus, solvent evaporation is negligible. GNPBH4 with radii larger than 1.5 nm were obtained by short-period heating, whereby solvent evaporation was prevented by sealing the reaction vessel. Contamination of synthesis equipment (e.g., stirrer or reaction vessel) was not observed. In comparison, the synthesis of GNPcit is a relatively slow process carried out at elevated temperatures within several minutes up to 0.5 h. Therefore, solvent evaporation becomes a relevant issue for this synthesis. To avoid evaporation during GNPcit syntheses, reflux condensers were used for all syntheses. The accuracy of our instrumentation was checked by simple dilution experiments. Small portions of water were added to final GNPBH4 and GNPcit solution and subsequently measured with UV−vis. The results are displayed in Figure S1 of the Supporting Information. The resulting plots of Abs400 show good linearity for both syntheses, with a mean deviation of less than 0.4% and a maximum deviation of 1.3%. As a result, GNP solutions can be diluted with water in a wide concentration range without any nonlinear effects. Hence, the instrumentation that was used as well as the lab operations met the necessary accuracy. Influence of Particle Radius on Abs400. The influence of the GNP radius on the Abs400 was investigated with colloidal solutions having a constant Au(0) concentration of 0.25 mM. GNPs with different radii were obtained from NaBH4 and citrate reduction as well as from a citrate-based seeded-growth procedure. 14,21,63 Mean radii and polydispersities were determined by SAXS measurements (see Table S1 in the Supporting Information). The final GNPBH4 radii are in a range from 1.5 to 2.5 nm. The citrate reduction results in GNPcit with radii between 6.5 and 11.3 nm. The mean radius of GNPcit was
RESULTS AND DISCUSSION
The principal applicability of the Abs400 to determine Au(0) concentrations is verified by comparing six final colloidal solutions from two different synthetic procedures (reduction of HAuCl 4 with NaBH 4 and with sodium citrate). The concentration of HAuCl4 was varied (0.125, 0.25, and 0.5 M) for both synthetic procedures. In each case, the concentration of the reducing agent was adjusted proportionally to ensure full precursor reduction. The syntheses result in GNP solutions with different Au(0) concentrations and particle size distributions. The mean radii determined from SAXS measurements and the corresponding Abs400 are displayed in Table 1. In the investigated concentration range, the Abs400 shows a linear behavior both for the GNPBH4 and GNPcit. The results indicate that the Abs400 can be used to determine Au(0) concentrations (Figure 1). Table 1. Abs400 and Mean Radii Determined from SAXS Data for the Syntheses of GNPBH4 and GNPcit with Different Au(0) Concentrations mean radius, nm (polydispersity, %)
Abs400 Au(0) conc., mM
GNPBH4
GNPcit
GNPBH4
GNPcit
0.125 0.25 0.5
0.268 0.546 1.091
0.284 0.573 1.192
2.0 (20) 1.5 (20) 1.6 (20)
12.7 (10) 9.9 (10) 5.3 (10) 11117
dx.doi.org/10.1021/ac502053s | Anal. Chem. 2014, 86, 11115−11124
Analytical Chemistry
Article
Figure 2. (a) Results from in situ SAXS measurements of the synthesis of GNPBH4.21 (b) UV−vis spectra of the time-resolved measurements during the NaBH4 reduction. (c) Abs400 as a function of reaction time for the synthesis of the GNPBH4. (d) Dependence of the Abs400 on the particle radius for all GNPBH4 and GNPcit. Note that the x-axis is broken for the purpose of clarity.
small clusters at 400 nm.67,68 Further discussions on the absorbance of gold clusters are included in a later section of this publication. Figure 2d summarizes the dependence of the Abs400 on the particle radius for all investigated GNPBH4 and GNPcit with a Au(0) concentration of 0.25 mM. The particles’ mean radii range from 1.1 to 28 nm. A representative selection of the UV− vis spectra can be found in Figure S2 of the Supporting Information. The Abs400 is strongly size-dependent and can be divided into four size ranges. The first range (denoted as I in Figure 2d) corresponds to gold particles with mean radii in the sub-1.5 nm regime, for which the Abs400 was obtained from time-resolved UV−vis measurements. The Abs400 decreases from around 0.6 to a constant value of around 0.525. As already stated, the increased Abs400 can be attributed to the presence of sub-1 nm gold clusters. These clusters can still be existent in solution, since the Abs400 was determined during the particle growth. The second size range (denoted as II in Figure 2d) is represented by the final GNPBH4 for which the Abs400 is slightly increased to around 0.55. This increase continues with increasing particle size, including the size ranges of the final GNPcit (denoted as III in Figure 2d) and the GNPcit from the seeded growth preparation (denoted as IV in Figure 2d). As a result, the maximum relative deviation of the Abs400 is more than 27%. To ensure that the deviation of Abs400 is mainly affected by particle size and not a result of borate or citrate stabilization, the influence of NaBH4 and sodium citrate on the Abs400 was investigated. For this purpose, NaBH4 was added to the GNPcit and citrate to the GNPBH4. Although, a complete ligand exchange cannot be assured, it can be assumed that citrate is able to exchange borate on the GNPBH4. Borate and citrate are
further increased to 28 nm using the seeded-growth approach as described by Polte et al.14 The final GNPBH4 have radii higher than 1.5 nm. Nevertheless, the Abs400 of smaller particles can be measured during their growth, since the reduction with NaBH4 assures a full conversion to Au(0) within the first 100 ms. Thus, the Abs400 values of the smallest particles (mean radii between 1 and 1.3 nm) were derived from time-resolved UV−vis measurements during the GNPBH4 synthesis, and the size was taken from previous SAXS measurements of the same synthesis.21 In a previous work, the evolution of the radius during the NaBH4 reduction was studied via an in situ SAXS setup (Figure 2a). Briefly, in the first milliseconds of the reaction small gold clusters with radii of approximately 0.8 nm are formed that subsequently grow to a final radius of 1.7 nm. The volume fraction of the GNPs reaches a constant value within the first 100 ms, indicating that the reduction of the gold precursor is already completed at this point, which was also proved by XANES measurements. Time-resolved UV−vis measurements of the GNP BH4 synthesis were conducted with a stopped-flow setup to access information on the absorbance of sub-1.5 nm GNPs existent in solution during the first seconds of the synthesis. The corresponding particle size was correlated from time-resolved SAXS measurements.21 A selection of spectra is displayed in Figure 2b. The Abs400 decreases during the reaction, while a characteristic LSPR at around 500 nm evolves. In Figure 2c, the Abs400 is shown as a function of reaction time. The absorbance increases in the beginning of the reaction and decreases to a constant value after about 3 s. The decrease of Abs400 in the first 3 s is about 10%. The higher Abs400 in the beginning of the reaction can be attributed to the absorbance of 11118
dx.doi.org/10.1021/ac502053s | Anal. Chem. 2014, 86, 11115−11124
Analytical Chemistry
Article
functionalization. Upon addition of the stabilizer, the Abs400 increases significantly, whereby PVP has a higher impact than Pluronic F-127. A relevant influence of the PVPs’ molecular weight on the Abs400 was not found. In Figure 3f, relative deviations of Abs400 after functionalization related to the as-prepared GNPs vs the particle mean radius are shown. The deviation is found to be significantly higher for smaller particles. The typical deviation for the GNPcit is in the range between 1 and 6%, whereas it increases to over 20% for the GNPBH4. For the GNPcit, the deviations both for Pluronic F-127 and the PVPs are in the same range. The particle radii remain constant upon addition of polymer, which was proven by SAXS measurements. Thus, it can be excluded that the increased absorbance is caused by aggregation or growth of the GNPs. Mean radii of all functionalized GNPs as well as the related deviations of the Abs400 can be found in Table S2 in the Supporting Information. The data include a variation of stabilizer concentration, which shows that an increased stabilizer concentration also slightly increases the deviation of the Abs400. Thiols and Transfer of GNPs to Organic Solvents. GNPs can also be stabilized by small molecules that adsorb on a gold surface through strong covalent binding. Thiols are among the most studied molecular stabilizers. They can be deprotonated to form thiolates, which bind to the nanoparticle surface by a strong Au−S bond. A large number of synthetic procedures for thiolate-stabilized GNPs is available.7,26,69 Thiolate-capped GNPs can be dissolved in various solvents depending on the chemical structure of the thiol. Water-soluble thiol-capped GNPs can be prepared by functionalization with short-chain polar thiols. Solubility in nonpolar solvents can be achieved by long-chain alkanethiols. The influence of thiol functionalization on the Abs400 for GNPs in water and organic solvents was investigated by a comparison of standard GNPBH4, MSA-functionalized GNPBH4, and DT-capped GNPs in toluene. Water-soluble thiolatecapped GNPs were prepared upon addition of an aqueous solution of MSA to as-prepared GNPBH4.44,73 GNP in organic solvent were prepared by transferring GNPBH4 to DT in toluene via simple extraction (see Experimental Details for more information). The completeness of the phase transfer with DT was checked by measuring the UV−vis spectrum of the corresponding water phase. A measurable absorbance was not detected (Figure 4, waterDT). Also, aggregation at the interface or glass vial was not observed. UV−vis spectra of as-prepared GNPBH4 and thiol-functionalized particles are displayed in Figure 4. The Abs400 values with their relative deviations from the as-prepared GNPBH4 as well as particle mean radii and polydispersities derived from SAXS analysis are displayed in Table 2. The modification of the colloidal GNP solution with thiols causes a notable change in the spectra for both aqueous and organic solutions. It can be attributed to a significant interaction of the covalently bound thiols with the electronic system of the GNPs. The overall absorbance increases and the LSPR is broadened, damped, and slightly shifted to higher wavelengths, as known from the literature.74 The Abs 400 increases significantly for both the MSA- (11.4%) and the DT-capped GNPs (16.9%), which might be a result of the LSPR broadening or the changed electron band structure. The particle mean radius after addition of MSA remains constant while being slightly higher upon phase transfer with DT (see Table 2). The thiol functionalization experiment shows that
weakly bound ligands attracted by electrostatic forces and can easily be exchanged by covalently binding phosphane or thiol ligands.69 Citrate in turn is able to exchange dicarboxy acetone on the particle surface70 and also dominates when GNPs are produced by reduction with NaBH4 in the presence of sodium citrate.48,70 According to UV−vis measurements (see Figure S3 in the Supporting Information), the citrate addition results in a slightly higher deviation (2.6%) than the borate addition (0.3%), which indicates that citrate exchanges borate as surface ligand. Nevertheless, the strong increase of the Abs400 in Figure 2d is attributed predominantly to the GNPs size, whereas citrate and borate have only minor influences. Influence of Surface Functionalization on the Abs400. Steric Stabilizers. Steric stabilizers are commonly used to increase the colloidal stability and to obtain long-time stability of nanoparticles.71,72 Their influence on the Abs400 was investigated regarding (i) the type of steric stabilizer, (ii) the molecular weight, (iii) the stabilizer concentration, and (iv) the mean radius of the functionalized GNPs. Polyvinylpyrrolidone (PVP) with different molecular weights (10, 40, and 58 kDa) and Pluronic F-127 were added in different concentrations to as-synthesized GNPs with different particle mean radii (samples r4−r7, r9, r11, r13 and r16−r18 in Table S1 in the Supporting Information). The UV−vis spectra are shown in Figure 3. Pure stabilizer solutions (without GNPs) do not show a detectable absorbance contribution in the spectral range from 300 to 800 nm (see Figure S4 in Supporting Information). For all investigated solutions, the use of steric stabilizers increases the Abs400 considerably. Moreover, the LSPR is slightly shifted to longer wavelengths, while the shape of the spectra is not altered (Figure 3a−d). Figure 3e shows the Abs400 of GNPBH4 (mean radius 1.5 nm) and GNP cit (mean radius 8.7 nm) before and after
Figure 3. UV−vis absorbance spectra of (a) GNPBH4 and (b) GNPcit before and after functionalization with steric stabilizers. (c and d) The same spectra in a smaller wavelength range. (e) Abs400 of pure semidilute GNPs and GNPs after the functionalization with steric stabilizers and (f) the relative deviation at 400 nm vs nanoparticle size. 11119
dx.doi.org/10.1021/ac502053s | Anal. Chem. 2014, 86, 11115−11124
Analytical Chemistry
Article
Figure 4. UV−vis spectra for as-prepared and thiol-functionalized GNPBH4. The water phase after extraction with DT in toluene does not show an absorption (waterDT). Inset: Photograph of the phase transfer experiment. A solution of DT in toluene (left, top) was added to a GNPBH4 solution (left, bottom). After phase transfer, the upper toluene phase contains the extracted GNPs, while the lower water phase is colorless (right).
Table 2. Abs400, the Deviation Related to As-prepared GNPBH4, and Particle Mean Radii and Polydispersities Determined from SAXS for As-prepared GNPs and GNPs after Thiol Functionalization GNPBH4 sample
Abs400
deviation, %
mean radius, nm (polydispersity, %)
as prepared MSA DT/toluene
0.559 0.622 0.653
11.4 16.9
1.5 (20) 1.5 (20) 2.7 (10)
covalently bound molecules have a strong impact on the GNPs spectra and in particular on the Abs400. Therefore, general statements for the influence of surface modifications on the Abs400 can hardly be made. Influence of HAuCl4 on the Abs400. In the previous sections, the Abs400 for colloidal solutions after the complete reduction of the gold precursor is discussed. The Abs400 was investigated as a measure for the Au(0) concentration in the absence of any ionic gold species. In this section, the Abs400 of gold complexes and its influence on the absorbance of GNPs are discussed. HAuCl4 is the most common precursor for GNP syntheses in aqueous solution. Its absorption behavior, as well as its interaction with GNPs in solution, is an important and complex aspect. It can provide information on reduction processes that are necessary to understand growth mechanisms of GNPs. An aqueous solution of HAuCl4 has a yellow color as a result of absorption of AuCl4−x(OH)x− complexes in the UV−vis region. The absorption of the different complexes includes a spectral contribution at 400 nm (see Figure 5a). Moreover, the pH-dependent equilibrium of AuCl4−x(OH)x− complexes needs to be considered, since HAuCl4 is an acid and lowers the pH value with increasing concentration. In the literature, it is welldescribed that chloride ions are gradually exchanged against hydroxy ions, with increasing pH value favoring the formation of hydroxy complexes [e.g., AuCl3(OH)−, AuCl2(OH)2−, ...]. These complexes exhibit a lower contribution to Abs400 compared to AuCl4−.75 As a rule of thumb, the contribution of gold precursor to the Abs400 decreases with increasing pH
Figure 5. (a) UV−vis absorption spectra of HAuCl4 solutions diluted with water. (b) The Abs400 related to the concentration of HAuCl4 at high concentration values and (inset) at lower concentrations. Changes in the UV−vis spectra of (c) GNPBH4 and (d) GNPcit upon the addition of a HAuCl4 solution. (e and f) The values are compared to the corresponding dilution of the particle solutions with water (linear line), and the following deviations are highlighted (g and h).
value since the AuCl4− percentage of all gold complexes decreases. This effect can be investigated by a simple dilution experiment of HAuCl4 solution with water (Figure 5a,b). At high HAuCl4 concentrations (1−10 mM), the pH value is between 2 and 3 and almost only AuCl4− is present in solution. Consequently, in this concentration range the Abs400 is linear (see Figure 5b). At concentrations typical for most syntheses (below 1 mM) the pH is above 3. Thus, other gold complexes are also present [in particular AuCl3(OH)− at this pH range], which results in a nonlinear decrease of the Abs400 with decreasing HAuCl4 concentration (see the inset in Figure 5b). Besides this nonlinear absorption effect, the dilution experiment reveals that Abs400 is much more sensitive to Au(0) than to Au(III). For all concentrations, Au(0) has a much higher absorption than the respective Au(III). The Abs400 of a GNP solution [i.e., Au(0)] with a concentration of 0.25 mM is at around 0.55, whereas a HAuCl4 solution [i.e., Au(III)] with the same gold concentration is at around 0.02. However, the influence of ionic gold complexes on the Abs400 of GNPs remains unclear. Thus, final GNPs were mixed with HAuCl4 in different percentages retaining a constant total gold 11120
dx.doi.org/10.1021/ac502053s | Anal. Chem. 2014, 86, 11115−11124
Analytical Chemistry
Article
changes for sub-20 nm particles, resulting in a size dependence of the Abs400 (see Figure 2d). Concerning the smallest GNPs with particle diameters of around 1 nm, the metal to insulator transition occurs gradually (i.e., for which small metallic clusters develop a band structure and at which size the valence electrons can be treated as electron gas) and needs to be taken into account. In this size regime of molecular-like clusters, the Abs400 can obviously not be attributed to interband transitions and can therefore most likely not be used as a measure for Au(0) concentrations. The exact cluster size for the metal to insulator transition can hardly be given for any metal as it even depends on the criterion for metallicity.80 Nevertheless, the absorbance of molecular-like metal clusters can be wellunderstood with recent contributions from Harbich and coworkers. They were able to precisely determine the absorption of size-selected noble metal clusters in a rare gas matrix at low temperatures starting from the single atom.81−83 As expected, the gold atom does not absorb in the visible range and in particular at 3.1 eV.81 The first absorption peak is around the Fermi energy at 5.5 eV (∼225 nm). However, gold clusters larger than the dimer already absorb at 3.1 eV, but the absorbance is different for each cluster size (Aun=2−9). As a consequence, absorbance in the visible spectrum of these metallic clusters cannot be used for a reliable determination of Au(0) concentrations. As a consequence, the Au(0) concentration in GNP solutions that contain substantial fractions of gold clusters (e.g., as part of the inherent size distribution) can hardly be determined using Abs400. The different electronic structures of metal clusters might be the reason for the relatively large Abs400 deviation between the smallest GNPBH4 measured during the growth (mean radius around 1 nm) and the final GNPBH4 of that synthesis (mean radius around 1.5 nm) in Figure 2c. In addition to particle size, the surface chemistry can also have an influence on band structure and transition probability and therefore on Abs400. As a result, the Abs400 of gold colloids changes due to added ligands or surface oxidation. Thus, changing the surface chemistry by adding steric stabilizers (PVP and Pluronic F-127) or thiols can cause relatively large Abs400 deviations. Indeed, the Abs400 deviation determined for the different particle sizes or ligands might also be affected by a change of the dielectric function of the surrounding medium.9,26,84 Both are known to affect the LSPR. Thus, the contribution of the LSPR to the Abs400 might not be as minimal as previously assumed. Regardless, for systems with similar particle sizes and similar chemistry (e.g., the same stabilizing agent), the Au(0) concentrations can be determined reliably via Abs400. The accuracy and detection limit of Abs400 were determined (see Figure S5 and related discussions in the Supporting Information) to allow a practical comparison with the abovementioned techniques of quantitative gold concentration determination (Table S4 in the Supporting Information).37−39,41,43,85−94 Abs400 is found to be highly accurate, while some techniques can obtain lower detection limits. However, a general extinction coefficient at 400 nm for GNP solutions cannot be stated. It rather has to be referred to certain colloidal GNPs. In Table 3, extinction coefficients are presented as a rough estimation that can be used to determine the Au(0) concentration. For the investigated solutions, the extinction coefficient ranges from 2.14 to 2.74 L·mol−1·cm−1. Extinction
concentration of 0.5 mM. In fact, GNPBH4 and GNPcit solutions were diluted with different amounts of HAuCl4 solutions. Both GNP solutions were synthesized with an initial Au(0) concentration of 0.5 mM to cover a wider concentration range upon dilution with HAuCl4. The addition of HAuCl4 causes a change of the LSPR at around 520 nm both for the GNPBH4 (Figure 5c) and the GNPcit (Figure 5d). The LSPR is broadened and shifted towards lower energies. This shift is not caused by aggregation and growth. It was previously shown by SAXS measurements that particle sizes remain constant upon addition of HAuCl4 to GNP solutions.14 The evolution of the Abs400 in dependence of HAuCl4 addition is illustrated in Figure 5e for GNPBH4 and in Figure 5f for GNPcit. The addition of HAuCl4 is represented by the molar fraction of GNPs (x-axis) calculated from the initial sample volume and the resulting volume upon HAuCl4 addition. For instance, at a molar fraction of 0.5 the investigated solution consists of 0.25 mM Au(0) (the GNPs) and 0.25 mM Au(III) (HAuCl4). The linear line in Figure 5e,f is the fitting of the Abs400 values from water addition to the respective GNPs (see Figure S1 in the Supporting Information). It serves as a comparison for Abs400 of the GNP solutions upon addition of HAuCl4 represented by the open circles in Figure 5e,f. As expected, the Abs400 decreases for both additions. However, values from HAuCl4 addition differ significantly from the simple water dilution. The relative deviations are displayed in Figure 5g,h. The deviation increases exponentially to over 26% for the GNPBH4 and 13% for the GNPcit. The deviations can mainly be attributed to the increased spectral contribution of HAuCl4 to the Abs400, which is in accordance with the HAuCl4 dilution experiment with water (see Figure 5a). This interpretation is also confirmed by the appearance of the characteristic absorption band of the AuCl4− complex at about 295 nm.75 The higher deviations for GNPBH4 compared to GNPcit are most likely caused by the stronger decrease of the pH value upon HAuCl4 addition, since citrate acts as a pH buffer. pH values as well as Abs400 of all prepared dilutions can be found in Table S3 of the Supporting Information. A further characteristic is found for the dilution of GNPBH4 with HAuCl4 at molar fractions above 0.8 (i.e., only small portions of HAuCl4 added), for which the deviation reaches negative values. It can be assumed that a red shift of the LSPR (as shown in Figure 5c) is responsible for this effect which might reduce the spectral contribution of the LSPR to the Abs400. Abs400 as a Method To Determine Au(0) Concentrations in Colloidal Gold Solutions. The Abs400 of colloidal gold solutions is located between two decisive spectral regions. Below 400 nm organic substances influence the absorbance, and above 400 nm it is influenced by the LSPR of the GNPs. The principle capability of the Abs400 to determine Au(0) concentrations is attributed to the allowed interband transitions at 3.1 eV. Interband transitions are characteristics of nanoparticles with electronic band structure (i.e., bulklike nanoparticles). Actually, only gold and copper have interband transitions with preserved k-vector in the visible spectrum, making them the only metals with specific colors for the human eye.57,76−79 In this contribution, it is shown that the Abs400 is in principle capable of determining Au(0) concentrations of colloidal gold solutions, but is essentially influenced by a variety of parameters (e.g., particle size, surface modifications). Indeed, it is not surprising that band structure and transition probability 11121
dx.doi.org/10.1021/ac502053s | Anal. Chem. 2014, 86, 11115−11124
Analytical Chemistry
Article
ongoing consumption of gold precursor (2). In the third phase, a fast particle growth to the final particle size takes place by complete reduction of the remaining precursor (3). For the UV−vis measurement, the reaction volume is relatively small (500 mL) and the reaction temperature can be controlled with good accuracy. The total duration of the reaction (thus the duration of its single reaction phases) is slightly different between the syntheses measured with UV−vis and SAXS. The longer reaction duration in the time-resolved SAXS measurement is most likely a consequence of the relatively big reaction volume (3 L). Thus, the reaction temperature might have been a few degrees lower than 75 °C. It has to be noted that such a difference in the reaction temperature can influence the duration of the single reaction phases but in principle does not influence the GNP formation mechanism. The time-resolved measurement of the Abs400 presents a fast and powerful tool to obtain reduction progress data during a GNP synthesis. The technique can be carried out with standard laboratory equipment and offers higher control of the synthetic parameters. Hence, a simple, fast, easy, and cheap in situ method for Au(0) determination is demonstrated. Moreover, in the presence of other gold species such as Au(III), the Abs400 method has to the best of our knowledge an unprecedented accuracy compared to expensive and laborious methods such as XANES.
Table 3. Ranges of the Estimated Extinction Coefficients for As-prepared GNPs and GNPs with Steric Stabilization ε400, L·mol−1·cm−1 sample/mean radii, nm
as prepared
PVP stabilized
Plu F-127 stabilized
GNPBH4/1.5−2.5 GNPcit/5−10 GNPcit/10−20 GNPcit/20−28
2.14−2.23 2.29−2.38 2.38−2.57 2.63−2.68
2.43−2.65 2.38−2.43 2.45−2.51 2.68−2.71
2.34−2.46 2.36−2.41 2.44−2.52 2.70−2.74
coefficients of all investigated GNP solutions can be found in the Supporting Information (see Table S2). As previously mentioned, only gold and copper have interband transitions in the visible spectrum. Consequently, the Abs400 could in principle be applied to determine Cu(0) concentrations for colloidal copper, since the plasmon resonance of copper is at around 550 nm.95,96 Although not within the scope of this contribution, a first experiment was conducted with copper nanoparticles synthesized by reduction of Cu(NO3)2 with NaBH4 (for details see Figure S6 and related discussions in the Supporting Information). The results reveal that for colloidal copper the Abs400 is not particularly suitable for a determination of Cu(0) concentration. In fact, absorbance below 400 nm (e.g., at 300 nm) seem to be more suitable, but at these wavelengths organic stabilizers may contribute to the absorbance. Thus, the results of this work concerning colloidal gold cannot be adapted to colloidal copper in a simple manner. Time-Resolved UV−Vis Measurements of the GNPcit Synthesis. Time-resolved measurement of the Abs400 can be used to gain precise information on reduction rates in the GNPcit synthesis. It is found that the AuCl4− complex contributes to the Abs400. Nevertheless, in the relevant HAuCl4 concentration range between 0.125 and 0.5 mM, common for the synthesis of GNPs, the Abs400 shows only minor deviations (
