Quantifying Citrate Surface Ligands on Iron Oxide Nanoparticles with TGA, CHN Analysis, NMR, and RP-HPLC with UV Detection

Citrate is widely used as a surface ligand for nanomaterials due to its biocompatibility and ease of exchange during postsynthetic surface modification. However, its quantification on nanoparticle surfaces is rarely addressed. Here, we present a multimethod analytical approach for determining citrate content on iron oxide nanoparticles (IONPs), which are broadly applied across life and material sciences. The methods evaluated include thermogravimetric analysis (TGA), elemental CHN analysis, photometry, citrate-selective reversed-phase HPLC with UV detection, and quantitative NMR spectroscopy (qNMR). Challenges arising from the strong absorbance of magnetic nanomaterials and interference from paramagnetic iron were overcome through optimized sample preparation.

Our results highlight the value of combining selective and nonselective methods to fully characterize nanomaterial surface chemistry and cross-validate measurements. We demonstrate that nonselective approaches such as TGA or CHN can overestimate citrate levels due to contributions from residual synthesis reagents or surfactants. In contrast, selective RP-HPLC provides higher sensitivity and avoids multi-step sample preparation required for qNMR. These findings underscore the potential of RP-HPLC as a powerful, underused tool for ligand quantification in nanomaterial research.

The original article

Quantifying Citrate Surface Ligands on Iron Oxide Nanoparticles with TGA, CHN Analysis, NMR, and RP-HPLC with UV Detection

Anna Matiushkina, Sarah-Luise Abram, Isabella Tavernaro, Robert Richstein, Michael R. Reithofer, Elina Andresen, Matthias Michaelis, Matthias Koch, Ute Resch-Genger*

Anal. Chem. 2025, 97, 36, 19627–19634

https://doi.org/10.1021/acs.analchem.5c03024

licensed under CC-BY 4.0

Selected sections from the article follow. Formats and hyperlinks were adapted from the original.

Engineered inorganic, organic, and hybrid nanomaterials (NMs), which commonly present core or core/shell nanostructures stabilized with covalently or coordinatively bound surface ligands, are meanwhile widely used in the material and life sciences, e.g., in nanomedicine, (1) medical diagnostics, (2) for energy conversion and storage, (3) optoelectronics, (4) and catalysis. (5) Application-relevant NM features such as optical and magnetic properties are largely determined by size, morphology, composition, and crystal phase, while the surface chemistry controls the colloidal stability, dispersibility, processability, and NM interaction with the environment and biological species, and hence exposure and potential toxicity. (6,7) This highlights the importance of methods for the controlled surface functionalization of NMs (8,9) and analytical methods to quantify ligands and functional groups (FGs) on NM surfaces (6,10,11) as well as test and reference materials for method validation. (12) The latter is also relevant for property-application and property-safety relationships to enable NM grouping, and to ease the development of the next generation of safe and sustainable by design (SSbD) NMs. (7)

Analytical methods to determine and quantify surface FGs and ligands include quantitative nuclear magnetic resonance (qNMR) techniques, (13) X-ray photoelectron spectroscopy (XPS), (14) vibrational spectroscopy, e.g., Fourier-transform infrared (FTIR) and Raman spectroscopy, (15) thermogravimetric analysis (TGA), (16) inductively coupled plasma mass spectrometry or optical emission spectroscopy (ICP-MS or ICP-OES), (11,17) and electrochemical titration techniques such as conductometry and potentiometry. (6,18) Chemo-selective methods such as solid state or solution qNMR can quantify ligands either on NMs or in solution after NM dissolution, (13,19,20) while FTIR or Raman yield only semiquantitative results. (6,11) Other methods provide less specific information such as surface coating mass loss like TGA, unless combined with mass spectrometry or FTIR, yield only the total amount of (de)protonable FGs such as conductometry, or are solely suitable for ligands containing heteroatoms, such as sulfur, like ICP-MS and ICP-OES. (6) Also, simple optical assays are frequently employed, which require a chemical reaction or an electrostatic interaction of the FGs and the signal-generating reporter. (18,21) Such assays yield the assay-specific amount of reporter-accessible FGs, which depends on reporter size and charge. (13) The applicability of all these analytical methods for NM surface analysis depends on NM composition and surface ligand(s) and the NM-ligand bonding interactions. This also determines mandatory sample preparation steps prior to ligand analysis, such as NM removal from dispersion or NM dissolution, and can affect the accuracy of the measurements. (22)

One of the most frequently utilized surface ligand for stabilizing metal, metal oxide, and lanthanide nanoparticles (NPs) in hydrophilic environments (23) is citrate due to its biocompatibility and simple replacement by other more strongly binding ligands in postsynthetic surface modification reactions. (9,24) However, the quantification of citrate on NM surfaces has rarely been addressed, although it is a frequent analyte in medical and food analysis. (25) Examples present the determination of citrate on gold, silver, and iron oxide NPs (IONPs) using TGA, elemental (CHN) analysis, and FTIR. (26) The frequent use of citrate ligands encouraged us to develop and compare different methods for citrate quantification on NMs representatively for IONPs broadly applied in the life sciences, e.g., for magnetic resonance imaging (MRI), magnetic separation, and magnetic hyperthermia, (27) with a focus on assessing the potential of versatile high-performance liquid chromatography (HPLC) methods still underexplored for quantifying surface ligands on NMs. Methods explored, varying in citrate selectivity, include photometry, TGA, CHN analysis, as well as reversed-phase high-performance liquid chromatography (RP-HPLC) with absorption (UV) detection and solution qNMR. Challenges originating from the strongly absorbing and magnetic IONPs and their constituting paramagnetic iron ions interfering with optical and NMR methods were tackled by specific sample preparation workflows. Our multimethod approach to citrate quantification, which demonstrates the applicability of RP-HPLC for NM ligand analysis and solution qNMR for magnetic NMs, highlights the importance of combining selective and unspecific methods for NM surface analysis, easing method validation by cross-comparison, and shows limitations of ligand-unspecific measurements. Our results can pave the road to a frequent usage of versatile HPLC methods in NM characterization workflows.

Experimental Section

Particle Characterization

The particle characterization was done by dynamic light scattering (DLS) yielding the number-based hydrodynamic diameter via cumulants fitting, as well as by zeta potential measurements at pH 9.2 using a Zetasizer Nano ZS (Malvern Panalytical), equipped with a 633 nm laser. The measurements were conducted at 25 °C using NP dispersions with concentrations of about 1 and 2 g/L. Transmission electron microscopy (TEM) images were obtained on a Talos F200S microscope (Thermo Fisher Scientific) and analyzed using the software ImageJ (Version 1.54g), evaluating 300 and 500 particles for IONPs-OA and IONPs-CA, respectively. The IONP iron content was determined by ICP-OES using a SPECTRO Arcos-EOP (Model: FHX, 76004553) spectrometer (SPECTRO Analytical Instruments) and iron ICP standard solution (SRM from NIST, Sigma-Aldrich).

Citrate Quantification

Photometric measurements were performed with a Cary 5000 spectrophotometer (Agilent) using 2 mm quartz cuvettes. TGA measurements of dried samples were done with a Hitachi STA 7200 setup with an AS3 Sample Charger under argon atmosphere at heating rates of 10 °C/min for IONPs-CA and 20 °C/min for IONPs-OA and sodium citrate dihydrate, and CHN analysis was performed with a PerkinElmer 2400 Series II analyzer. For the RP-HPLC studies, a 1260 Infinity system (Agilent Technologies) equipped with a diode array detector (DAD) was employed with an Eurospher II 100–5 C18 (250 × 4.6 mm) column. The mobile phase consisted of a phosphate buffer (pH 2.9 ± 0.1)-methanol mixture (97.5:2.5), the flow rate was 1 mL/min, and the detection wavelength set to 210 nm. Solution qNMR measurements done at 25 °C with a Bruker BioSpin AV III 600 spectrometer operating at 600.25 MHz for ¹H detection, using the maleic acid signal (6.57–6.65 ppm, 2H) as internal reference for quantifying the citric acid signals (3.03–3.10 ppm, 2H; 3.21–3.28 ppm, 2H).

Results and Discussion

Analytical Characterization of the IONPs

Prerequisites for our citrate quantification study, summarized in Figure 1, are IONPs with a narrow size distribution, as the dispersity in NP size and shape can lead to enhanced uncertainties for the determination of the NP number concentration and total surface area. (22) To obtain monodisperse, high-quality, crystalline IONPs with a uniform spherical shape for our multimethod citrate quantification study, we used a thermal decomposition method in apolar solvents, which provides a better monodispersity compared to aqueous syntheses. (34) The resulting OA-capped IONPs (SI, Figures S1–S3) reveal a uniform spherical shape and a narrow size distribution with size variations of 5%. Subsequently, the hydrophobic OA surface ligands were exchanged for hydrophilic CA. For NMs such as AgNPs, IONPs, quantum dots, and lanthanide NPs, this approach is frequently used for preparing high-quality water-dispersible NMs. (9,35) As confirmed by TEM, ligand exchange did not affect IONP size and monodispersity (Figure 2, panels A and B). Successful ligand exchange is also supported by the highly negative zeta potential of IONP-CA samples of −40 ± 5 mV under alkaline conditions, while DLS reveals the absence of aggregates (Figure 2, panel B, and SI, Figure S4). However, the desired quantitative exchange of the hydrophobic for the hydrophilic ligands is difficult to achieve and prove. As shown in Figure 2 (panel C), the mass concentration of the IONPs-CA sample obtained gravimetrically, along with the iron oxide (Fe₂O₃) concentration calculated from the iron content (5.50 ± 0.05 g/L) determined by ICP-OES (SI, eq 1), indicates a mass not attributable to iron oxide (CA, water residuals, etc.) of 14.6 ± 2.3 wt %. This value provides the upper limit of the citrate content in the sample. From the log-normal fit of the TEM size distribution (SI, eq 2 and 3), an average IONP surface area of 293 nm² and an average IONP volume of 472 nm³ were derived. Combining these values with the ICP-OES data and the material density taken from the literature (γ-Fe₂O₃; ρ of 4.9 g/cm³) gave a particle number concentration and total surface area of the IONPs-CA in the stock dispersion (36) of 3.40 × 10¹⁵ particles and 9.96 × 10¹⁷ nm² per mL (SI, eq 4 and 5).

Anal. Chem. 2025, 97, 36, 19627–19634: Figure 2. (A) TEM image and size distribution (inset) of the IONPs-CA sample. (B) Bar diagram displaying the size of the IONPs-OA and IONPs-CA samples obtained from TEM images and number-based DLS (error bars refer to the IONP size distribution). (C) Mass concentration of the IONPs-CA sample determined by gravimetry and ICP-OES.

TGA

The thermogravimetric (TG) curve of sodium citrate dihydrate used as a reference, shown in Figure 3 (red curve), displays four distinct steps of the thermal decomposition (SI, Table S1). The first step corresponds to dehydration (37) and the other three steps are associated with citrate degradation, in agreement with the literature. (38) The TG curve of the IONPs-CA sample (Figure 3, blue curve), however, does not show clearly distinguishable steps, but rather multiple mass loss steps. The first step with a mass loss of ∼ 3.5 wt % is ascribed to water elimination. The other steps, associated with a total mass loss of ∼ 12 wt %, are attributed to the degradation of organic species on the IONPs surface. The total mass loss of 15.5 ± 0.3 wt % agrees well with the calculation of the sample composition from a combination of gravimetric and ICP-OES measurements presented above. The observation that the mass loss ratio of IONPs-CA during the three stages of citrate decomposition does not correspond to that of sodium citrate could point to the parallel degradation of other organic compounds, see also the section on RP-HPLC measurements. A more detailed analysis of the TG and differential thermogravimetric (DTG) curves of sodium citrate dihydrate and IONPs-CA can be found in the SI.

Anal. Chem. 2025, 97, 36, 19627–19634: Figure 3. TG and DTG curves of sodium citrate dihydrate (red, right axis) and the IONPs-CA sample (blue, left axis).

Absorption Spectroscopy

Citrates show a characteristic absorption band at about 210 nm with a small molar absorption coefficient of about 202 M^–1·cm^–1, that has been exploited before for citrate quantification. (33) However, the strong absorption of iron species in the UV/vis interferes with such measurements. (31) Absorption measurements of IONPs-CA dissolved by addition of HCl (SI, Figure S5) reveal that the contribution from the citrate absorbance in the samples and standards is negligible compared to the high background absorption of the iron species in the wavelength range of 200 to 250 nm. This renders this simple photometric approach not suitable for determining citrate in our IONPs-CA samples.

RP-HPLC with UV Detection

The challenging high background imposed by the presence of iron ions can be elegantly circumvented by combining a classical separation method such as RP-HPLC, separating different species in solution based on their interaction with the mobile and stationary phases, with optical detection. This is shown in Figure 4 (panel A). CA, corresponding to the peak at about 5.2 min in the HPLC chromatograms, is well separated from chloride and iron species present in solution which pass through a RP-HPLC column without retention (peaks at around 2.5 min), see HPLC chromatograms of all standards in the SI in Figure S6. The calibration curves obtained for the CA standards with and without Fe³⁺ ions (Figure 4, panel B) can be well fitted with a linear regression in the concentration range of 0.01 mM to 0.50 mM using 7 standards per calibration and can be extended to higher CA concentrations. This correlates well with the limit of citrate quantification of 0.016 mM determined by Luo et al. (39) The calibration curves measured on three different days (SI, Figure S7, panels A-C) confirm a reproducibility with an accuracy of 4%. The presence of iron ions does not affect the citrate peak, which is clearly distinguishable from the background, and hence its quantification from the calibration curves. Fe³⁺ ions do not need to be considered for citrate quantification which was also supported by measurements of the citrate calibration standards containing even higher iron ion concentrations (SI, Figure S7, panel D). The accuracy and precision of citrate quantification with this HPLC approach was confirmed by two control samples A (0.198 mM) and B (0.092 mM) with measured values 0.198 and 0.093 mM, respectively, with measurement error 0.002 mM. Based on our HPLC measurements, the amount of citrate in the as prepared IONPs-CA samples was determined to 0.164 ± 0.002 mM. This corresponds to a citrate concentration of 3.28 ± 0.03 mM in the stock IONPs-CA dispersion and 6.7 ± 0.2 wt % considering the mass concentration of the stock dispersion as determined by drying. This value amounts to only half of the amount of citrate derived from the nonselective TGA method. Apparently, the total organic content determined by TGA cannot solely be attributed to citrate. This suggests the presence of OA ligands on the IONP surface which were not completely replaced by citrate or other organic residuals from the IONP synthesis and/or ligand exchange procedure.

Anal. Chem. 2025, 97, 36, 19627–19634: Figure 4. (A) Typical HPLC chromatogram of a dissolved IONPs-CA sample with the inset showing the peak corresponding to citrate; the detection wavelength is 210 nm. (B) Calibration curves used for quantifying the citrate concentration from the HPLC data measured in the absence and presence of Fe3+ ions; the inset shows the normalized absorption spectrum of citrate.

qNMR

Next, we explored solution qNMR for quantifying citrate ligands of dissolved IONPs-CA samples. Thus, in the sample preparation workflow involving drying, weighting, and dissolution with DCl, a step had to be included for the quantitative removal of the released molecular paramagnetic iron species interfering with the qNMR measurements. Iron ions were precipitated by addition of NaOD, forming iron deuteroxide, which was separated from the analyte solution by centrifugation. Thereby, 99.97 ± 0.01% of the iron species were removed, as confirmed by ICP-OES. The NMR spectra of the supernatant obtained for representative IONPs-CA samples are shown in Figure 5 and in the SI (Figure S8). The arbitrarily assigned protons H_A and H_B, which yield two doublet signals due to geminal coupling, were independently integrated within a narrow frequency window to exclude possible influences from impurities. The concentration of citrate in the IONPs-CA stock solution was then calculated by comparing the integrals of the citric acid protons with that of maleic acid (SI, eq 6). This yielded a citrate concentration of 2.95 ± 0.04 mM (6.0 ± 0.1 wt %) in the IONPs-CA stock dispersion. The additional peaks observed in the NMR spectra of the supernatant, e.g., at 8.44, 3.16, and 2.91 ppm, indicate the presence of organic impurities in the sample, originating from IONP synthesis including the reagents used, the ligand exchange process, and the qNMR sample preparation procedure. This finding agrees well with the observed differences between the HPLC and TGA data. Additionally performed NMR spiking experiments revealed that the peaks at 8.44 and 2.28 ppm correspond to formic and acetic acids, respectively, that could possibly be degradation products of CitAc. Based on these qNMR measurements, the concentrations of formic and acetic acids were determined to about 0.49 mM and 0.05 mM, respectively.

Anal. Chem. 2025, 97, 36, 19627–19634: Figure 5. Solution 1H NMR spectrum of the supernatant of a dissolved IONPs-CA sample after removal of the iron species, revealing the peak integrals used for citrate quantification by comparison with maleic acid utilized as internal standard (δ = 6.60 ppm). The inset shows the citric acid peaks.

Conclusion and Outlook

In summary, our multimethod approach to citrate quantification highlights the advantages of combining specific and unspecific methods for the characterization of NM surface chemistry. This is also important for toxicity studies and NM risk assessment where the potential toxicity of the NM ligand is typically separately assessed. In addition, we could demonstrate the underexplored potential of RP-HPLC with UV detection as an efficient method for quantifying citrate on NPs. Contrary to qNMR, HPLC methods do not require a multistep sample preparation in the case of IONPs and most likely for other NMs as well, which could lead to an enhanced uncertainty. Also, HPLC measurements can be performed with relatively small sample amounts, as follows from the about 20-fold smaller amount of IONPs-CA sample used for the HPLC measurements compared to qNMR. This can be crucial for some applications. We expect that in the future, versatile HPLC methods with photometric, fluorescence and mass spectrometry detection as well as application-specifically adapted workflows for NM sample preparation as derived here for IONPs will receive much more interest for the analysis and quantification of NM surface ligands.