Supporting Information: Analysis of protein coatings on gold nanoparticles by XPS and liquid-based particle sizing techniques

Transcription

1 Supporting Information: Analysis of protein coatings on gold nanoparticles by XPS and liquid-based particle sizing techniques Natalie A. Belsey, a) Alex G. Shard a) and Caterina Minelli a),b) National Physical Laboratory, Analytical Science, Hampton road, Teddington, Middlesex TW11 LW, UK a) American Vacuum Society member. b) Electronic mail: SI. DCS of an aggregated nanoparticle (NP) sample The use of the LSPR shift for the computation of the protein shell refractive index relies on the challenging condition that NPs are not aggregated. DLS accuracy and interpretation of data also depend on the knowledge of the aggregation state of the NPs. Among the available techniques to measure NP size distribution, DCS was the only one with sufficient size resolution to allow monitoring of the NP aggregation. Figure S1 shows an example of an aggregated NP sample. Samples of this type were discarded from the data presented in this work, but this one is shown here for clarity. The sample is made of 2 nm NPs coated with IgG protein. The size distribution of the citrate-coated 2 nm NPs is also shown for comparison. The main mode of the size distribution relates to the population of NPs which are monodisperse. The size distribution exhibits a tale at higher sizes, where other modes are identifiable, which relate to the NP dimers, trimers and aggregates of higher order. The size distribution is in fact so detailed to allow quantitative data analysis, which shows that the fraction of NP population in the monodisperse state is ~18%. For this analysis, the size distribution was modelled as the sum of pseudo-voigt monomodal distributions representative of the NPs in the monomeric, dimeric, trimeric, etc. states. 1

2 Figure S1. Example of a DCS size distribution where aggregation among IgG-coated 2 nm NPs is clearly visible. For comparison, the size distribution of monodisperse citratecoated 2 nm NPs is also shown. SII. Calculation of organic shell thickness and composition on gold nanoparticles by XPS. From the work of Shard 1 we obtain the following equations: A = I 1I 2 I 2 I 1 [1] Where I i is the measured XPS intensity and I i is the measured or calculated intensity for the pure material of unique photoelectrons from the shell (overlayer), i = 1, and the core (substrate), i = 2, respectively. B = L 1,a L 2,a C = L 1,a L 1,b [2] [3] Where L i,j is the attenuation length of photoelectrons arising from material i travelling through material j, where j = a represents the shell and j = b represents the core. L 1,a. The core radius of the particle, R, and shell thickness, T, are expressed in units of 2

3 The following equation and ancillaries may be used to obtain T NP to a precision of better than 4%. T NP = T R~1+βT 1+β [4] T = R [(ABC + 1) 1 3 1] [4a] T R~1 = T R R R+α α = 1.8 A.1 B.5 C.4 [4b] [4c] β =.13α2.5 R 1.5 [4d] T R =.74A3.6 ln(a)b AB.41 A [4e] Equations 4a to 4e apply to data where the core and shell intensities are measured and normalised to the intensities from the respective pure materials, as encapsulated in A. 1 As with any measurement of overlayer thickness by XPS, this implies knowledge of or an ability to estimate the pure material intensities using the same XPS instrument and settings. In this case, the pure material intensities cannot be obtained and therefore an estimate must be made. We assume that XPS relative sensitivity factors may be employed to describe the intensity arising from different elements in the same material and therefore the intensity of element k in the pure organic overlayer material can be expressed as: I k X k S k [5] Where X k is the mole fraction of the element in the pure organic overlayer material and S k is the sensitivity factor of the XPS signal from that element used in the analysis. One may be tempted to use an expression such as that shown in equation [6bad] to obtain A, but this cannot be recommended because the sensitivity factors only relate to signals arising from the same material and, because attenuation lengths, densities and intrinsic loss processes in the two materials may differ significantly, serious errors can result. 3

4 A k = I ki Au I Au I I ks Au k I Au S k X k [6bad] NB: Do not use this; it only works if the core and shell are similar materials. A better approach is to compensate for the attenuation lengths and densities, where these are known, but this still does not adequately address intrinsic loss processes which, for all but a small handful of materials, can only be obtained experimentally. The recommended method is to analyse flat samples of the two pure materials consecutively with the sample to obtain the pure material intensities. However, obtaining pure materials without contaminants is not always possible. In the following, we make the quite reasonable assumption that attenuation lengths, densities and intrinsic losses are negligibly different for organic materials. To simplify and generalise the analysis of organic materials on gold substrates and nanoparticles, we also employ the sensitivity factor of gold, but recognise that an additional factor, f, must be used to compensate for the different attenuation length, density and intrinsic loss processes between the two materials: the gold core and organic shell (Au and Org). Thus, for each element in the shell we have a distinct value of A: A k = I ki Au I Au I = f I ks Au = f k I Au S k X k [k] [6] [Au]X k In which [k] and [Au] are the mole fractions or atomic% of those elements determined by XPS analysis in the standard manner. The first task is to find the value of f (which we estimate to be ~.4 from consideration of the relative electron attenuation lengths and densities of gold and organic materials). We have measured XPS intensities from a pure organic material and from sputtercleaned gold under the same experimental conditions on the same day. Firstly, separate samples of a spin cast PEG-biotin thiol and gold provided the results in table 1. The value of f is simply given by the ratio of the normalised and summed intensities for each material. 4

5 Table S1. XPS line I (cps ev) S I/S (cps ev) Sum for material O 1s C 1s N 1s S 2p Au 4f f In a second experiment a vacuum evaporated layer of ~5 nm Irganox 11 on gold was analysed by XPS and then the organic layer removed by argon cluster beam sputtering. The cleaned gold was analysed under the same conditions with the sample in the same position. Table S2. XPS line I, cps ev S I/S, cps ev Sum for material O 1s C 1s Au 4f f Therefore a value of f =.56 appears both useful and consistent for these two different types of organic material and should be generally applicable. This value is somewhat different to the ~.4 expected from a consideration of material densities and electron attenuation lengths, but accounts for attenuation length errors, intrinsic loss processes and the systematic errors in background subtraction during data analysis and therefore is used here. The values of L i,j may be obtained from the equation S4 in the paper from Seah. 2 The following table lists the relevant values: Table S3. XPS line L (Org), nm L (Au), nm B C O 1s ±.4 C 1s N 1s S 2p Au 4f NA 5

6 If the elements in the shell are homogenously distributed, then application of equation 4 should provide the same shell thickness (the product L k,org.t NP,k ) for all elements, providing that the values of X k are correct in equation 6. We use the constraint that the sum of all X k is equal to 1, along with an iterative calculation using the Solver function in Excel where X k are varied iteratively from an initial set of trial values to obtain the same shell thickness for all elements from the data. The result provides the shell thickness, T NP and the composition of the shell, X k. The described model for XPS data analysis was applied to XPS measurements of protein coated NPs such as those shown below in Figure S2. The figure shows typical survey and C 1s narrow scans measured for 2 nm NPs coated with citrate and with the protein molecules. The narrow scan illustrates our choice of PTFE as a substrate: the CF 2 peak emanating from the PTFE exhibits a large energy shift with respect to the signal of the other carbon bonding states. For this reason, the substrate signal contribution can be easily separated from that of the NPs. Furthermore, extremely low levels of carbon contamination were measured on clean PTFE surfaces prior to particle deposition. 6

7 1 A F 1s 2 nm IgG NPs 8 7 B 2 nm IgG NPs 8 6 O 1s N 1s C 1s Data Fit components Sum fit C 125 E 125 G Raw data Background fit Au 4d Na 1s Raw data Background fit F 1s Raw data Background fit Na 1s S 2p Au 4f 2 nm BSA NPs O 1s N 1s Au 4d Raw data Background fit F 1s C 1s 25 Au 4f 2 nm Peptide NPs O 1s N 1s C 1s Au 4f F 1s Au 4d S 2p 2 nm citrate NPs C 1s O 1s Au 4d Au 4f D F H CF Data Fit components Sum fit CF 2 2 nm BSA NPs CF 2 2 nm Peptide NPs Data Fit components Sum fit nm citrate NPs Data Sum fit Binding energy 28 Figure S2. XPS survey (A, C, E and G) and C 1s narrow spectra (B, D, F and H) of 2 nm gold nanoparticles coated with IgG (A and B), BSA (C and D), peptide (E and F) and citrates (G and H) deposited on PTFE substrates. 7

8 SIII. Other experimental results: Table S4. LSPR shifts, thicknesses measured by DLS and XPS, shell refractive index and areal mass densities measured by techniques in liquid (Opt) and by XPS for different NP sizes and different protein coatings. Sample Δλ max (nm) T DLS (nm) T XPS (nm) n s Γ s (Opt) (mg/m 2 ) 1 peptide ± ± peptide ± ± peptide ± ± peptide ± ± peptide ± ± BSA ± ± BSA ± ± BSA ± ± BSA ± ± BSA ± ± IgG ± ± IgG ± ± IgG ± ± IgG ± ± IgG ± ± Γ s (XPS) (mg/m 2 ) SIV. ζ-potential measurements of coated NPs: ζ-potential measurements were performed to confirm changes in the NP surface chemistry upon coating. ζ-potential measures the potential difference between the double layer at the NP slipping plane and that of the bulk solution. With increasing protein shell thickness, it is expected to become less negative, since the slipping plane is physically further away from the charged NP gold surface, but clearly the charge of molecules at the NP interface also affects its value. The results of these measurements are shown in Figure S3 below. The ζ-potential of the citrate-coated NPs was found to decrease with increasing NP size, from -23 mv to -41 mv. Upon acquisition of a BSA or an IgG shell, the ζ- potential of the NPs becomes less negative, which is consistent with previous work. 3, 4 The dependency of ζ-potential on NP size was less pronounced for BSA coated NPs and was lost for the IgG coated NPs. This is consistent with the IgG shells being thicker than the BSA ones. In particular, ζ-potential values of IgG-coated NPs are close to mv, 8

9 which implies an increased likelihood of particle aggregation as repulsion between individual particles is reduced. This result supports our choice to characterise the IgG coated NPs in liquid before centrifugation. On the contrary, for the NPs coated with CAG4 peptide, the measured ζ -potential is the same as the citrate-stabilised NPs for NPs of size 1 nm and 2 nm, and becomes more negative for NPs with larger size. The reasons for this behaviour may be twofold: first of all the thickness of the peptide layer is lower than that of BSA and IgG coatings. Secondly, CAG4 peptides at ph 9 are negatively charged and form a self-assembled monolayer at the surface of the particles. It is expected that this peptide layer is more ordered for larger particles, which would imply a higher peptide packing density, translating in a higher surface density of negative charges. Figure S3. ζ-potential measured for protein and citrate coated NPs. SV. DCS measurement of NP effective density and shell thickness: Equation 2 in the manuscript provides a means to plot the effective densities of the NPs as a function of particle size and Equation 3 in the manuscript can be used to fit the data and estimate the average shell thickness of the NPs, as shown in Figure S4. 9

10 Figure S4. NP effective density measured from DLS and DCS data ( ) and relative fit according to Equation 3 of the manuscript. The average thickness calculated from the fit is indicated. The values of the NP effective density, ρ eff, resulting from the fits tended to be slightly underestimated for NP sizes below 3 nm, and slightly overestimated for sizes above 3 nm. This was confirmed by observing that fit residuals tended to decrease with increasing NP size. As the fits were performed in the assumption of a uniform density of each type of coating across NPs of different sizes, one possible explanation for this effect is that molecular species are more densely distributed on surfaces with higher radius of curvature. A similar conclusion may be reached by observing that the 8 nm NPs always exhibited the lower refractive indices (Table S4). Surface molecular adsorption is a complex kinetic process and no simple model exists that can usefully be applied to this 1

11 data. However, a decrease in NP surface curvature may be responsible for enhancing the steric barrier to further molecular adsorption by increasing the density of loops and chains on the outside of the protein shell. This would decrease the rate of interaction and exchange with protein in solution and prevent larger particles attain equilibrium coverage of protein. 4 SVI. Example of SEM image of citrate coated gold NPs: In all of the characterisation techniques used in this work, it is assumed that NPs are spherical. Deviation from this assumption introduces inaccuracy in the measured size of the NPs. Figure S5 shows that under SEM investigation, the NPs appeared in a range of shapes. Currently there are no validated methods for correcting our data for deviations from sphericity, but reasonable methods exist for estimating shape factors 5 in sedimentation and diffusion which suggest that errors in measured diameters will be much less than 5% for even the least spherical particles shown here. All results will therefore need to be interpreted as relating to the equivalent average sphere. For example, DLS results should then be interpreted as the hydrodynamic size of the equivalent sphere having the same diffusion coefficient as the NP. Figure S5. SEM image of citrate coated gold nanoparticles with nominal size of 8.3 nm. The image exemplifies the range of shapes and sizes contained in the NP sample. 11

12 SVII. Relative protein content in protein shells. The abundance of nitrogen in the protein shell as measured by XPS can be used to quantify the relative amount of protein in the NP shells. Figure S6 shows that the protein mass relative to the total organic matter remains constant around 85% for the IgG shell and decreases almost monotonically from 7% to 4% in the case of BSA. For the CAG4 peptide, the peptide content varied between 8% and 4%, with a minimum for 4 nm NPs. Figure S6. Relative protein content in protein shells as measured by XPS and normalised to the expected content. SVIII. Supporting Information references: 1. A. G. Shard, Journal of Physical Chemistry C 116 (31), (212). 2. M. P. Seah, Surface and Interface Analysis 44 (1), (212). 3. S. H. Brewer, W. R. Glomm, M. C. Johnson, M. K. Knag and S. Franzen, Langmuir 21 (2), (25). 4. N. C. Bell, C. Minelli and A. G. Shard, Analytical Methods 5, (213). 5. A. J. Weinheimer, Journal of the Atmospheric Sciences 44, (1987). 12