Characterization of gold nanoparticle films: Rutherford backscattering spectroscopy, scanning electron microscopy with image analysis, and atomic force microscopy

Gold nanoparticle films are of interest in several branches of science and technology, and accurate sample characterization is needed but technically demanding. We prepared such films by DC magnetron sputtering and recorded their mass thickness by Rutherford backscattering spectroscopy. The geometric thickness dg—from the substrate to the tops of the nanoparticles—was obtained by scanning electron microscopy (SEM) combined with image analysis as well as by atomic force microscopy (AFM). The various techniques yielded an internally consistent characterization of the films. In particular, very similar results for dg were obtained by SEM with image analysis and by AFM.


I. INTRODUCTION
Gold nanoparticles (AuNPs) have numerous applications, especially in green nanotechnology, 1,2 and can be used for catalysis 3 and in plasmonically enhanced devices such as photovoltaic cells, 4,5 light emitting diodes, 6 photocatalytic reactors, 7,8 and gas sensors. 9,10in films comprised of AuNPs have a geometric thickness d g -between the substrate and the tops of the nanoparticles-that clearly is larger than the mass thickness d m for a hypothetical, uniform layer containing the same number of atoms.Both of these thicknesses are of interest for analyzing device performance.This paper reports on AuNP films prepared by sputter deposition onto glass and analyzed by Rutherford backscattering spectroscopy (RBS), scanning electron microscopy (SEM) combined with image analysis of the AuNP distribution on the substrate, and atomic force microscopy (AFM).X-ray reflectivity measurements could have served as an alternative to the RBS data, as shown recently by Kossoy et al. 11 A major result of our investigation is that very similar results for d g were obtained by SEM with image analysis and by AFM.

II. SAMPLE PREPARATION
Gold was deposited onto glass by DC magnetron sputtering.The target was a 5-cmdiameter plate of 99.99% pure Au placed 13 cm above the substrate holder.The system was first evacuated to ~1.6 × 10 -5 Pa, and sputtering was then performed at 50 W in ~0.8 Pa of 99.98% pure Ar.The substrate holder was rotated to ensure even deposits.Samples were prepared without deliberate substrate heating and also at a substrate temperature of 140 ± 10 ºC, as determined by a thermocouple.Deposits with 4.5 ≤ d g ≤ 10.4 nm and 1.7 ≤ d m ≤ 5.1 nm are reported on below.
3][14] Furthermore, detailed knowledge of the substrate is required for RBS analysis, especially for determining accurate values of d m , since the signal from the substrate is crucial for normalization of the number of atomic species incident onto the RBS detector.Specifically, we used 1-mm-thick plates of glass (standard microscope slides, supplied by Thermo Scientific, UK).According to RBS their composition was, in at.%, 59.4 O, 23.1 Si, 2.5 Ca, 0.5 K, 11.0 Na, 3.2 Mg, and 0.3 Al; this composition is consistent with information by the glass supplier.

III. SAMPLE CHARACTERIZATION: TECHNIQUES AND DATA
A. Rutherford backscattering spectroscopy RBS data were taken at Uppsala University's Tandem Laboratory, by use of 4 He + ions backscattered at an angle of 170º, and were simulated with the SIMNRA code. 15 where M Au is the molar mass of Au, N A is Avogadro's constant, and ρ Au is the density of Au and taken to be 19.31g/cm 3 , i.e., the bulk value.The spectrum in Fig. 1 was recorded on Sample D in Table I; the same table also contains data for three other samples that were analyzed by RBS in the same way.

B. Scanning electron microscopy and image analysis
A LEO 1550 FEG instrument with in-lens detection was used to obtain SEM images.The samples were oriented perpendicular to the electron beam in order to obtain a picture of the AuNP distribution on the substrate.The acceleration voltage was kept as low as 2-5 keV to avoid charging effects, and the distance between lens and sample was 2-3 mm. Figure 2 shows data for Samples A-D and verifies that all of them are comprised of nanoparticles.
The SEM pictures were analyzed by a procedure based on an image processing tool supplied with MATLAB's Toolbox. 16Each image was converted to n a pixels of equal size, and a number n p accounted for the pixels representing particles.The area fraction f SEM for the substrate coverage by AuNPs was then obtained from and we finally derived the geometric thickness of the AuNP film by which assumes that the particles can be described as objects with top surfaces parallel to the substrate.The image analysis also gave an average width l p of the particles.4, where the straight line signifies equality between d AFM and d g .

IV. CONCLUDING REMARKS
The characterization of nanoparticle films is important in many applications, for example in environmentally benign green nanotechnology, 1,2 and is notoriously difficult.We reported here on a comprehensive study of gold nanoparticles and applied RBS, SEM combined with image analysis, and AFM to a set of AuNP films in order to determine their mass thicknesses, geometrical thicknesses, and particle widths.The particle widths were typically twice the geometrical thickness and about four times the mass thickness.A particularly interesting result was that RBS and SEM with image analysis, and AFM provided almost identical data on particle heights.
Figure 1 is an example of a RBS spectrum and shows a well-defined peak at high energies, caused by Au, and onsets of scattering at lower energies due to the various constituents of the substrate.The number N Au of Au atoms per area unit was determined via the simulation, and d m was then derived from Figure 1

Figure 2 Figure 4
Figure 2 Table I reports values of f SEM , l p and d g ; it shows that 0.38 ≤ f SEM ≤ 0.49 and that l p is roughly twice as large as d g .
Data on d AFM are given in TableIfor the various samples.It is interesting to note that d AFM and d g are in very good agreement.The relationship between these parameters is highlighted in Fig.

Table TABLE I
. Data for films of AuNPs sputter deposited onto glass at the shown substrate temperature τ s .Values are given on mass thickness d m determined from Rutherford backscattering spectroscopy; area fraction of particles f SEM , particle width l p , and geometric thickness d g determined from scanning electron microscopy combined with image analysis; and particle height d AFM determined from atomic force microscopy.Experimental and simulated RBS data for a film of AuNPs (sample data are given in TableI).The various features are associated with the shown elements.The deviation between the two types of data for energies lower than ~400 keV is due to multiple scattering effects and inaccurate stopping values.FIG.2.SEM images for films of AuNPs.Sample data are given in TableI.FIG.3.AFM data for a film of AuNPs (sample data are given in TableI).Upper and lower panels show a three-dimensional rendition of sample roughness and a histogram of relative heights, respectively.The peak in the histogram defines d AFM .FIG.4.Geometric thickness d g determined from SEM combined with image analysis, and average particle height d AFM determined from AFM, for films of AuNPs (sample data are given in TableI).The line represents equality between the two parameters.