Gold Complexes for Focused-Electron-Beam-Induced Deposition

Sep 16, 2014 - ABSTRACT: Four gold complexes were tested as a precursor for focused- electron-beam-induced deposition: [ClAuIIIMe2]2, ClAuI(SMe2),...
1 downloads 0 Views 8MB Size
Article pubs.acs.org/Langmuir

Gold Complexes for Focused-Electron-Beam-Induced Deposition W. F. van Dorp,*,† X. Wu,‡ J. J. L. Mulders,§ S. Harder,‡ P. Rudolf,† and J. T. M. De Hosson† †

Zernike Institute for Advanced Materials, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands Inorganic and Organometallic Chemistry, Friedrich-Alexander Universität Erlangen-Nürnberg, Egerlandstr. 1, 91058 Erlangen, Germany § FEI Electron Optics, Achtseweg Noord 5, 5600 KA Eindhoven, The Netherlands ‡

S Supporting Information *

ABSTRACT: Four gold complexes were tested as a precursor for focusedelectron-beam-induced deposition: [ClAu III Me 2 ] 2 , ClAu I (SMe 2 ), ClAu I (PMe 3 ), and MeAu I (PMe 3 ). Complexes [ClAu III Me 2 ] 2 and MeAuI(PMe3) are volatile, have sufficient vapor pressure at room temperature for deposition experiments, and were found to yield deposits that contain gold (29−41 and 19−25 atom %, respectively). Electrons easily remove the Cl ligand from [ClAuIIIMe2]2, and predominantly both methyl ligands are incorporated into the deposit. Electrons remove at least one methyl group from MeAuI(PMe3). Complexes ClAuI(SMe2) and ClAuI(PMe3) are not suitable as a precursor. They dissociate in vacuum, and the only volatile components are Cl, SMe2, and PMe3, respectively.



and 2−3 atom % for dimethylgold hexafluoroacetylacetonate20 (Figure 1c). High-purity gold nanostructures are essential to the directed self-assembly of functional organic molecules,21 as seeds for the growth of nanorods and nanotubes,22,23 and for plasmonics.24 For the latter application, complex 3D shapes are interesting, for which FEBIP is a particularly suitable fabrication technique.25,26 The low purity obtained from the current (CVD) gold precursors causes poor electrical conductivity, and as-written deposits are not plasmonically active.26 From numerous experiments over the past few decades it is clear that electrons are very inefficient at removing the ligands that are used to make CVD precursors volatile.27 The reason for this inefficiency is twofold: (1) The reactions of neutrals (such as in CVD) are fundamentally different from the reactions of ionized molecules (such as in FEBIP). (2) The CVD precursors have never been designed to be successful for FEBIP. For instance, recent results strongly suggest that dissociative electron attachment is an important reaction mechanism in the electron-induced dissociation of FEBIP precursors. In dissociative electron attachment, an electron is absorbed in a molecular orbital to form an excited transient anion.28 The excited transient anion decays, and one of the possible outcomes is the dissociation of the molecule. Because the CVD precursors have not been tuned to this type of reaction, the reaction paths are not optimal, leading to poor product quality.

INTRODUCTION Sample modification with electron beams plays an important role in advanced materials engineering.1 When focusedelectron-beam-induced processing (FEBIP) is used, is it possible to add or remove material locally without affecting the sample outside the area of interest. In a sense, it is a local variant of chemical vapor deposition (CVD) or atomic layer deposition. In FEBIP, a beam of electrons is focused onto a target in the presence of a gaseous precursor. The precursor molecules adsorb onto the sample surface and react under the impact of electrons. Local etching occurs when the reaction products form a gaseous product with the target material. Local deposition occurs when the reaction product is a solid residue. FEBIP is currently the only technique with which masks for (extreme) ultraviolet lithography can be repaired.2,3 And among others, FEBIP enables the fabrication of nanowires4 or scanning probes,5 the contacting of nanotubes,6 and local functionalization using directed self-assembly.7,8 Historically, the precursors for FEBIP are adopted from CVD. For instance, popular precursors for W, Pt, and Au are tungsten hexacarbonyl,9 trimethylplatinum methylcyclopentadienyl,10 and dimethylgold acetylacetonate,11,12 respectively. Three CVD Au precursors that are used for FEBIP are shown in Figure 1a−c.13,14 Although indeed such precursors work well for CVD and yield pure W, Pt, or Au,9−11 the precursors yield material of insufficient quality when processed with FEBIP.12−14 The deposits consist mainly of an amorphous carbon matrix in which metal crystallites (a few nanometers in size) are embedded. Reported gold contents are 8−20 atom % for dimethylgold acetylacetonate15,16 (Figure 1a), 10−40 atom % for dimethylgold trifluoroacetylacetonate17−19 (Figure 1b), © 2014 American Chemical Society

Received: July 3, 2014 Revised: September 5, 2014 Published: September 16, 2014 12097

dx.doi.org/10.1021/la502618t | Langmuir 2014, 30, 12097−12105

Langmuir

Article

Figure 1. Gold complexes used for FEBIP. (a−c) Commercially available gold complexes that have been adopted from CVD and are used for FEBIP. In the absence of further processing, they yield deposits of insufficient quality (i.e., low metal content). (d, e) Gold complexes with small ligands that have been tested for FEBIP. While yielding pure gold deposits, these complexes are thermally unstable or unstable in vacuum. (f−i) Complexes studied in this work.

Table 1. Gold Compounds Tested in this Work name

abbreviation

CAS number

reference

figure

dimethylgold(III) chloride chloro(dimethylsulfide)gold(I) chloro(trimethylphosphine)gold(I) methyl(trimethylphosphine)gold(I)

[ClAuIIIMe2]2 ClAuI(SMe2) ClAuI(PMe3) MeAuI(PMe3)

30676-27-8 29892-37-3 15278-97-4 32407-79-7

37 38 39 39

1f 1g 1h 1i

insights. Our ultimate goal is to develop rules for the rational design of new precursors by screening well-chosen sets of complexes. Our experiments are therefore aimed at identifying potential candidates rather than characterizing the deposition behavior of each individual compound in extensive detail. Taking ClAuI(PF3) and ClAuI(CO) as a starting point, we selected the four compounds listed in Table 1 for electroninduced dissociation. They all have relatively small ligands, are thermally stable at room temperature, and are likely to have sufficient vapor pressure. ClAuI(SMe2) is commercially available, the synthesis of the other three complexes is described in the Experimental Section. In the past, the use of MeAuI(PMe3) for (laser-assisted) CVD has been suggested.35,36 We inspected the solid compounds with scanning electron microscopy (SEM) and transmission electron microscopy (TEM), monitored the vapor with mass spectrometry, and examined the condensed volatile components using Xray photoelectron spectroscopy (XPS). Finally, we tested the performance of the complexes as precursors for focusedelectron-beam-induced deposition.

Although strategies are being developed to purify FEBIP deposits during or after processing,29−31 precursors optimized for FEBIP would be preferable. Such dedicated precursors do not require additional (and sometimes) complex processing and have the potential to open entirely new avenues of research and applications. However, the many additional requirements for FEBIP precursors make it a significant challenge to find such a compound. The vapor pressure needs to be high enough to give sufficient molecule flux at the sample. Because harmful compounds are banned more and more from the workplace, toxic precursors should be prevented. An ideal precursor should not be aggressive to the equipment and should preferably be inexpensive. In addition, a long lifetime both on the shelf and after loading onto the machine is essential for practical suitability. Two innovative reports have been published where gold compounds have been selected and tested specifically for FEBIP: chloro(trifluorophosphine)gold(I) (ClAuI(PF3),32 Figure 1d) and chloro(carbonyl)gold(I) (ClAuI(CO),33 Figure 1e). Indeed these precursors yield pure Au, demonstrating that the Cl, PF3, and CO ligands are fully removed by the electrons. However, the lifetime of ClAuI(PF3) and ClAuI(CO) is short, as the compounds decompose either thermally or in vacuum. In the case of ClAuI(CO), this appears to occur through the release of the weakly bound CO ligand: 33 crystalline ClAuI(CO) already decomposes at room temperature, and it is advisible to store this complex cold under a CO atmosphere.34 Although the two compounds are a logical and promising starting point for developing dedicated FEBIP precursors, a more rational design of FEBIP precursors would strongly benefit from additional data on electron-induced dissociation. At the moment, it is not clear which ligand types are successfully removed by the electrons, and data that allow us to correlate the molecule architecture with product quality are missing. Ideally, trends in how sensitive ligand−metal bonds are to bombardment with electrons should be identified. In this article, we make the first attempt to obtain such data and



EXPERIMENTAL SECTION

Synthesis. ClAuI(SMe2) was ordered from Aldrich and used as received. Complexes [ClAuMe 2 ] 2 , 4 0 ClAu I (PMe 3 ), 3 9 and MeAuI(PMe3)39 were synthesized according to literature procedures. The synthesis was performed under a nitrogen atmosphere using predried solvents and standard Schlenk techniques. The starting material for all synthesis routes was H[AuCl4]·3H2O, which was obtained in the form of orange crystals by dissolving gold metal in aqua regia, evaporating all liquids and, after the addition of concentrated HCl, evaporating all liquids again. The products were analyzed by C and H elemental analysis and showed satisfactory values (except for the C value of ClAuI(PMe3) which was slightly higher than calculated, probably as a result of slight contamination with Me 3 PO). [ClAuMe 2 ] 2 , calcd (%) for C4H12Cl2Au2: C 9.15, H 2.30; found (%) C 9.75, H 2.40. ClAuI(PMe3), calcd (%) for C3H9PClAu: C 11.68, H 2.94; found 12098

dx.doi.org/10.1021/la502618t | Langmuir 2014, 30, 12097−12105

Langmuir

Article

Figure 2. SEM characterization of crystals of the compounds. (a) [ClAuIIIMe2]2 crystals. (b) The same [ClAuIIIMe2]2 crystals after 41 min in vacuum. (c) The EDX spectrum of [ClAuIIIMe2]2 crystals, showing a composition of 6−7 atom % Cl, 5−6 atom % Au, about 69−73 atom % C, 7−9 atom % Si, and 7−10 atom % O. (d) ClAuI(SMe2) crystals. (e) ClAuI(SMe2) at higher magnification, showing a porous morphology. (f) The EDX spectrum of ClAuI(SMe2) crystals, showing a composition of 22−29 atom % C, 44−48 atom % Au, 18−23 atom % S, and 8−12 atom % Cl. (g) ClAuI(PMe3) crystals. (h) ClAuI(PMe3) crystals at higher magnification, showing a needlelike morphology. (i) The EDX spectrum of ClAuI(PMe3), showing a composition of 11 atom % Cl, 15 atom % Au, 15 atom % P, and 60 atom % C. (j) MeAuI(PMe3) crystals. (k) The same MeAuI(PMe3) crystals after 22 min in vacuum. (l) The EDX spectrum of MeAuI(PMe3), showing a composition of 4−9 atom % Au, 4−8 atom % P, 70 atom % C, 9−15 atom % Si, and 5−8 atom % O. base pressure of 8 × 10−10 mbar or better, with a Scienta R4000 highenergy-resolution spectrometer equipped with a monochromatic Al Kα source (hν = 1486.6 eV). The photoelectrons were collected at normal emission. The energy resolution was 0.4 eV. The XPS spectra were analyzed using least-squares curve-fitting program Winspec.41 Binding energies are reported ±0.05 eV and referenced to the Ag 3d5/2 photoemission peak originating from the substrate, centered at a binding energy of 368.3 eV. Deconvolution of the spectra included Shirley baseline subtraction and fitting with a minimum number of peaks consistent with the structure of the molecules on a surface, taking into account the experimental resolution. The profile of the peaks was taken as 60% Gaussian/40% Lorentzian functions. The precursor was contained at room temperature in a stainless steel reservoir and introduced into the UHV chamber through a leak valve. The precursor and/or the volatile components were condensed onto a cooled Ag foil (T ≤ 163 K) by backfilling the UHV chamber to 6 × 10−8 mbar for 30 or 35 min. Prior to the precursor being dosed onto the Ag foil, the Ag foil was cleaned by Ar+ sputtering at 1 kV and the cleanliness was verified by XPS. The gas-phase analysis is done with an MKS Vac Check mass spectrometer with a maximum detection range of m/z = 100. Focused-Electron-Induced Deposition. The FEBIP experiments were performed on an FEI Helios Nanolab dual-beam system. All deposition experiments were carried out with an acceleration voltage of 5 kV. In all cases, the substrate was a Si wafer. No special

(%) C 13.13, H 3.29. MeAuI(PMe3), calcd (%) for C4H12PAu: C 16.68, H 4.20; found (%) C 17.15, H 4.26. Samples were also analyzed by 1H and 31P NMR spectroscopy. [ClAuMe2]2, 1H NMR (300 MHz, C6D6, 25 °C): δ = 0.98 (s, CH3). ClAuI(PMe3), 1H NMR (400 MHz, C6D6, 25 °C): δ = 0.41 (d, 2JHP = 11.1 Hz, CH3); 31P NMR (162 MHz, C6D6, 25 °C): δ = −10.8 (m, 2 JHP = 11.1 Hz, PMe3). MeAuI(PMe3): 1H NMR (400 MHz, C6D6, 25 °C): δ = 1.20 (d, 3JHP = 8.5 Hz, 3 H, AuCH3), 0.60 (d, 2JHP = 8.7 Hz, 6 H, PCH3); 31P NMR (162 MHz, C6D6, 25 °C): δ = 11.4 (m, 2JHP = 8.7 Hz, PMe3). These values correspond to values published earlier.40,39 The compounds were stored at 243 K in a dry N2 atmosphere and loaded into vacuum reservoirs, either a stainless steel reservoir or the Al crucible of an FEI gas injection system (GIS) in an Ar or N2 atmosphere. Electron Microscopy. Crystals of the gold complexes were studied using a Philips XL30 environmental scanning electron microscope (SEM) equipped with a field emission gun and an EDAX EDX detector for energy-dispersive X-ray spectroscopy (EDX). ClAuI(SMe2) crystals were also studied in a JEOL 2010F transmission electron microscope (TEM) with a field-emission gun. The TEM was operated at 200 kV and has a Bruker EDX detector and a Gatan system for electron energy loss spectroscopy (EELS). Surface and Gas-Phase Analysis. X-ray photoelectron spectroscopy experiments were performed in ultrahigh vacuum (UHV) at a 12099

dx.doi.org/10.1021/la502618t | Langmuir 2014, 30, 12097−12105

Langmuir

Article

Figure 3. Mass spectra of (a) [ClAuIIIMe2]2, (b) ClAuI(SMe2), (c) ClAuI(PMe3), and (d) MeAuI(PMe3). The NIST reference spectra are given for SMe2 and PMe3 in b−d, respectively. effort was taken to clean the Si wafer prior to the deposition experiments. The complexes were loaded in a gas injection system (GIS) at room temperature and in a dry N2 atmosphere. The chamber was pumped for ≥1 h after mounting the GIS, and the GIS itself was pumped for at least 30 min before attempting focused-electron-beaminduced deposition. The EDX measurements of the deposits were performed in situ, directly following the deposition.



(Figure 2g,h). The composition was 11 atom % Cl, 15 atom % Au, 15 atom % P, and 60 atom % C (Figure 2i), i.e., slightly more C and less Cl then expected stoichiometrically (17 atom % Cl, 17 atom % Au, 17 atom % P, and 50 atom % C). This is in agreement with the elemental analysis that was also slightly too high for carbon (Experimental Section). In contrast, MeAuI(PMe3) was found to be volatile, with large crystals subliming within half an hour (Figure 2j,k). The EDX measurements showed that, similar to [ClAuIIIMe2]2, the crystals contained a significant amount of SiOxCy, which is likely silicone grease originating from the synthesis (Figure 2l). The composition was 4−9 atom % Au, 4−8 atom % P, 70 atom % C, 9−15 atom % Si, and 5−8 atom % O. Figure 3 shows the analysis of the volatile components with mass spectrometry. Because the maximum range of the mass spectrometer was m/z = 100, Au could not be detected directly but the presence of all ligands could be verified. [ClAuIIIMe2]2 gave no detectable signal in the mass spectrometer; the mass spectrum for [ClAuIIIMe2]2 in Figure 3a shows no significant peaks, other than the standard background from water and nitrogen. The mass spectrum for ClAuI(SMe2) (Figure 3b, solid black line) showed mainly the signal for SMe2 (gray dots, NIST reference data). We attribute the peak at m/z = 35 to the formation of a sulfonium ion (ionization fragment of SMe2)42 rather than to Cl because the isotope peak for Cl at m/z = 37 is absent. For ClAuI(PMe3) (Figure 3c) only a very weak signal was detected for the PMe3 ligand (gray triangles, NIST reference data). There are two peaks at 36 and 38. These peaks are consistent with the reference signal for HCl (gray dots, NIST data). That HCl is detected at m/z = 36 and 38, rather than Cl at m/z = 35 and 37, suggests that the complex releases Cl instead of reaching the vapor phase intact. The mass spectrum for MeAuI(PMe3) (Figure 3d) shows the signal for the methyl ligand at m/z = 15. The signal for the

RESULTS

Figure 2 shows SEM images from crystals of the four gold complexes. The crystals were inspected in the SEM to determine the stability and to determine the composition with EDX. For all EDX measurements, the compositions were determined directly after vacuum insertion while keeping the electron exposure due to imaging to a minimum. For [ClAuIIIMe2]2, the composition determined from the EDX measurements (Figure 2c) was 6−7 atom % Cl, 5−6 atom % Au, about 69−73 atom % C, 7−9 atom % Si, and 7−10 atom % O. This suggests that, apart from [ClAuIIIMe2]2, the crystals contain a significant amount of SiOxCy that may be related to the use of silicone grease in the synthesis. From the SEM inspection it is clear that [ClAuIIIMe2]2 is a volatile complex; the crystals sublimed within about 40 min, leaving only a SiOxCy residue. (Figure 2a,b). ClAuI(SMe2) crystals did not sublime in vacuum. For larger crystals the morphology appeared to be stable (Figure 2d,e). An EDX spectrum is shown in Figure 2f. The EDX measurements suggest that Cl is easily removed, showing no more than 8−12 atom % in larger particles. The SMe2 ligand is also easily removed with the electron beam. Although in larger particles the stoichiometric amount of S was detected (18−23 atom %), small ClAuI(SMe2) crystals were very sensitive to the electron beam and could easily be reduced to pure Au upon electron irradiation (Supporting Information). These results suggest that ClAuI(SMe2) is very sensitive to electrons. ClAuI(PMe3) crystals also did not sublime. The needles and platelets were stable over time and under the electron beam 12100

dx.doi.org/10.1021/la502618t | Langmuir 2014, 30, 12097−12105

Langmuir

Article

Figure 4. XPS spectra collected after condensation of the four Au complexes on Ag foil. The Au complexes are contained in a metal reservoir and are introduced into an ultrahigh vacuum chamber. The volatile components are condensed onto a Ag foil that is cooled to ≤163 K. The XPS spectrum of the sputtered Ag foil is given as a reference. The fits to the spectra are plotted with an offset to favor readability.

Figure 5. Testing the gold complexes as a precursor for focused-electron-beam-induced deposition. (a) A deposit written using [ClAuIIIMe2]2 with a beam current of 0.1 nA and a growth time of 20 min. (b) The same as in panel a, but now with 0.4 and 15 min of growth time. (c) EDX spectrum of a deposit from [ClAuIIIMe2]2. (d) A rectangular area scanned with the e-beam in the presence of ClAuI(SMe2) vapors using a beam current of 1.6 nA and a crucible temperature of 25 °C. (e) The same as in panel d, but now with the crucible at 95 °C. (f) EDX spectrum of deposits from ClAuI(SMe2) with the crucible at 25, 40, 60, and 95 °C. (g) Deposit written using MeAuI(PMe3) with a beam current of 0.1 nA and 15 min of growth time. (h) The same as in panel g, but with a beam current of 1.6 nA and 5 min of growth time. (i) EDX spectrum of a deposit using MeAuI(PMe3) as the precursor.

eV, close to the expected value.44 There is only a single component in the C 1s peak, centered at 284.8 eV, consistent with the identical bonding state of the two methyl ligands. The composition of the condensed vapor can be estimated from the XPS spectra using sensitivity factors. For [ClAuIIIMe2]2 this is 24.9 atom % Au, 27.4 atom % Cl, and 47.7 atom % C. This is close to the expected values (25 atom % Au, 25 atom % Cl, and 50 atom % C). For ClAuI(SMe2) only Cl, C, and S were detected in XPS; Au was not detected. Although the signal-to-noise ratio was poor (the cooling capacity of the manipulator was suboptimal for these experiments and the sample temperature was not low enough to condense multilayers of SMe2), it is clear that there was no Au on the surface.

PMe3 ligand at higher m/z corresponds well to the reference data (gray dots, NIST reference data). This suggests that MeAuI(PMe3) is volatile and reaches the mass spectrometer intact. Because Au cannot be detected using our mass spectrometer, we condense all volatile components and analyze these using XPS. Figure 4 presents the Au 4f, Cl 2p, C 1s, P 2p, and S 2p core-level regions of the XPS spectra measured after condensation on the Ag foil; the signal for the clean Ag foil is shown as a reference at the bottom. For [ClAuIIIMe2]2 the core-level photoemission lines of Au, Cl, and C were detected. The Au 4f binding energy of 86.9 eV is consistent with Au in a high oxidation state (AuIII), as expected.43 The Cl 2p line peaks at a binding energy of 199.5 12101

dx.doi.org/10.1021/la502618t | Langmuir 2014, 30, 12097−12105

Langmuir

Article

Similar results were observed for ClAuI(PMe3). There was no signal for Au, and only Cl, C, and P were detected. As seen in the panel with the Cl 2p core-level regions in Figure 4, two species of Cl are present (contributions peaked at 198.2 and 199.0 eV, respectively). The binding energies are fairly consistent with those for AgCl45 and HCl, the latter of which is known to adsorb molecularly on Ag films at the temperatures used in this work.46 There is a single contribution to the C 1s line, indicative of the three methyl groups that are all bound to the P. For MeAuI(PMe3) all elements in the complex are detected. A minor amount of oxygen is also detected (not shown). The Au 4f spectrum shows several oxidation states. We attribute this to the fact that the Ag foil is reactive toward MeAuI(PMe3) because the sample temperature is not low enough, causing MeAuI(PMe3) to decompose partially (details in Supporting Information). The binding energy of the main peak for Au 4f in MeAuI(PMe3) (85.8 eV) is lower than for [ClAuIIIMe2]2, which is consistent with its lower oxidation state and with literature values.47 The C 1s spectrum shows two components, consistent with the two nonequivalent types of C in the molecule. The intensity ratio between these components reasonably agrees: 79% of the total C intensity comes from the three methyl groups bound to P, and 21% comes from the methyl group bound to Au. The spectrum of the P 2p core region shows two components. The binding energy of the main component, 132.3 eV, is fairly consistent with those of related compounds.48 The origin of the smaller component at 133.1 eV is unclear; it is possibly related to the oxygen on the surface. On the basis of the spectral intensities, the deposition is 12 atom % Au, 72 atom % C, and 16 atom % P. This is approximately consistent with the expected values (17 atom % Au, 67 atom % C, and 17 atom % P). Therefore, the XPS results suggest that MeAuI(PMe3) sublimes and reaches the gas phase. Complexes [ClAuIIIMe2]2, ClAuI(SMe2), and MeAuI(PMe3) were tested as precursors for Au deposition. Figure 5a,b shows two deposits made with [ClAuIIIMe2]2. The deposit in Figure 5a was written with a beam current of 0.1 nA, and the deposit in Figure 5b, with 5 kV and a beam current of 0.4 nA. The GIS was kept at room temperature during the experiment. A grainy structure was observed on the surface, similar to what has been observed for ClAuI(PF3)32 and ClAuI(CO).33 The EDX spectrum in Figure 5c shows C and Au from the precursor, with only a minor amount of Cl. The Si signal is from the supporting wafer, the Al signal is from the sample holder. The composition was determined to be 29−41 atom % Au, 2−6 atom % Cl, and 53−68 atom % C. Complex ClAuI(SMe2) did not yield any deposit when the GIS was kept at 25 °C. Instead, the Si wafer was etched slightly (Figure 5d). The beam current was 1.6 nA, and the writing time was 1 min. Although we did not investigate this effect in detail, it strongly resembles etching with Cl gas. We observed the same effect for GIS temperatures of 40 and 65 °C. At a GIS temperature of 95 °C the pressure increased significantly upon opening the valve, and a substantial deposit was observed (Figure 5e). At GIS temperatures of 25, 40, and 65 °C the only two peaks in the EDX spectrum are from Si (strong) and C (very weak) signals. At a GIS temperature of 95 °C this is reversed: the deposit in Figure 5d appeared to consist of pure C, and only a small Si signal was detected. These results strongly suggest that the precursor decomposes thermally between 65 and 95 °C.

Finally, in Figure 5 panels g and h show deposits written using MeAuI(PMe3) with beam currents of 0.1 and 1.6 nA, respectively. The surface of the deposits looked smooth, and the grainy structure of the [ClAuIIIMe2]2 deposits was not observed. According to the EDX spectrum in Figure 5i the deposits consisted of 19−25 atom % Au, 54−62 atom % C, 12−16 atom % P, and 2−7 atom % O. In these exploratory experiments the composition did not appear to depend on the beam current for beam currents of between 0.1 and 13 nA.



DISCUSSION From the SEM measurements it is clear that [ClAuIIIMe2]2 is volatile, with the crystals disappearing in about 40 min (Figure 2a,b). Although its signature is not observed in the mass spectrometer (Figure 3a), the XPS experiments (Figure 4) suggest that [ClAuIIIMe2]2 sublimes and reaches the vapor phase intact. Consequently, [ClAuIIIMe2]2 can be used as a precursor for Au. As shown in Figure 5c the EDX measurements indicate that the deposits from [ClAuIIIMe2]2 contain 29−41 atom % Au, 2−6 atom % Cl, and 53−68 atom % C. This is, in the absence of optimizing the writing conditions, already on par with the highest purity for dimethylgold trifluoroacetylacetonate.19 The EDX results suggest that Cl is removed almost completely from the molecule. Only 2−6 atom % is detected by EDX, whereas 25 atom % is expected stoichiometrically. This is consistent with earlier results for ClAuI(PF3) and ClAuI(CO), where Cl was also found to be removed easily by the electrons.32,33 In contrast, the electrons do not appear to desorb much carbon from [ClAuIIIMe2]2. If both methyl ligands were incorporated into the deposit, then one expects a carbon content of 66 atom %, which compares well with the experimental range of 53−68 atom % C. This suggests that indeed both methyl ligands are predominantly incorporated into the deposit and only occasionally a Me ligand is removed. The deposits had a grainy structure (Figure 5a,b), similar to what has been observed for FEBIP precursors such as ClAuI(PF3)32 and ClAuI(CO).33 The results for ClAuI(SMe2) show consistently that this is not a suitable precursor for focused-electron-beam-induced deposition. Although small, micrometer-sized ClAuI(SMe2) crystals appeared to be very sensitive to electrons (Supporting Information), the SEM experiments showed that large crystals retain their morphology over time. This suggests that the complex is not volatile. In the mass spectrometer only the SMe2 ligand was detected. In the XPS spectra only Cl, C, and S were identified, and no signal for Au was observed. Together, these results suggest that ClAuI(SMe2) does not sublime as a molecule but instead releases only SMe2 and Cl. The deposition experiments confirmed this. No Au deposition was observed at any GIS temperature. Instead, a slight etching of the Si wafer (most likely from the released Cl) and significant carbon deposition were found for a GIS temperature of 95 °C, presumably as a result of the thermal decomposition of the complex. ClAuI(PMe3) is not a suitable precursor either. The crystals did not disappear over time and retained their morphology in the SEM, which means that the complex is not very volatile. In the mass spectrum, predominantly HCl was detected. This strongly suggests that the complex releases the Cl ligand in vacuum, which reacts with water on the walls of the vacuum system to form HCl. The XPS measurements showed that the complex also releases the PMe3 ligand (detection of C and P). 12102

dx.doi.org/10.1021/la502618t | Langmuir 2014, 30, 12097−12105

Langmuir

Article

instability of Cl-containing AuI compounds is the proximity of neighboring molecules in the solid phase. Tentatively, we suggest designing a gold compound using PF3 and/or CO ligands, in combination with a single methyl ligand. However, whereas undesired etching of the sample would be avoided, the thermal stability is a key issue. CF3AuICO is known to be unstable and highly reactive. For instance MeAuI(CO) is expected to be substantially less stable.52 Although instability is a desired property during electron-induced decomposition, it should not interfere with practicality. This underlines the challenge of synthesizing novel precursors based on the formulated design rules. However, we want to emphasize that there is still very little data on the dissociation chemistry of organometallic complexes. More work is required to determine to what extent our design rules apply to FEBIP precursors in general.

However, no Au was observed. On the basis of these results, we have not attempted to use ClAuI(PMe3) as a precursor. The experiments demonstrate that MeAuI(PMe3) is a volatile complex and reaches the gas phase intact. The SEM images (Figure 2j,k) testify to the sublimation of the complex, the expected ionization pattern was observed in the mass spectrum (Figure 3d), and all elements were detected in the stoichiometric ratio in XPS (Figure 4). Using the precursor MeAuI(PMe3) resulted in deposits with a smooth surface and a composition of 19−25 atom % Au, 54−62 atom % C, 12−16 atom % P, and 2−7 atom % O. The stoichiometric Au and C contents in MeAuI(PMe3) are 17 and 66 atom %, respectively. If a single methyl ligand was removed from the molecule, then the Au and C contents should be 20 and 60 atom %, respectively. The experimentally observed concentration of Au and C in the deposit was more consistent with the latter than with the former. We therefore conclude that at least one methyl group is removed from the molecule during deposition. Previous reports have shown that typically methyl ligands bound directly to the metal desorb, rather than any of the methyl groups from the ligand.27 This suggests that the PMe3 ligand is not a suitable ligand for FEBIP precursors, but more detailed experiments are necessary to confirm this. The oxygen that was detected by XPS (in the condensed volatile components) and by EDX (in the deposit) is most likely due to oxidized PMe3 ligands. PMe3 is very sensitive to air,49 and because O2 cannot be fully excluded from high-vacuum systems, a minor amount of OPMe3 can be formed. On the basis of these and previously reported experiments32,33 it is clear that complexes ClAuI(PF3), ClAuI(CO), ClAuI(SMe2), and ClAuI(PMe3) are unstable in vacuum. The reason for this instability is currently unknown. In this respect, the difference between ClAuI(PMe3) and MeAuI(PMe3) offers an opportunity to study this aspect in more detail. Whereas ClAuI(PMe3) is unstable in vacuum and releases only its ligands, MeAuI(PMe3) appears to be stable and reaches the gas phase intact. This may be due to the chemical difference between Cl and Me ligands: whereas a Cl ligand is strongly electron-withdrawing, the Me ligand is electron-releasing. Other differences in their physical and chemical properties might be related to a difference in dipole moments or aurophilicity. Aurophilicity refers to the tendency of Au complexes to form short intermolecular Au···Au contacts (< 3 Å).50 These metal−metal contacts represent weak secondorder metal−metal bonding, the bond energy of which can be on the order of 7−12 or 29−50 kJ/mol.51 From its crystal structure it is clear that this interaction is observed for ClAuI(PMe3).50 The crystal structure for MeAuI(PMe3) is unknown. Further work is necessary to clarify this issue. Preliminary design rules can be deduced on the basis of the results from this and earlier published papers. PF3, CO, and Cl are successfully desorbed upon electron irradiation.32,33 C is generally difficult to remove and should most likely be avoided in ligands.27 A notable expection is methyl ligands, which are known to desorbed upon irradiation, albeit under rather specific conditions. Of the methyl ligands in a molecule, only those bound to the metal center desorb.27 And if there are two metalbound methyl ligands (e.g., in [ClAuMe2]2), then only one appears to be removed. PMe3 appears to be unsuitable as a ligand because neither P nor its methyl ligands desorb when exposed to electrons. A limitation for using Cl as a ligand is that it causes (undesired) etching of the sample when it is released during electron exposure. We speculate that the source of the



CONCLUSIONS We tested four gold complexes as precursors for focusedelectron-beam-induced deposition: [ClAu I I I Me 2 ] 2 , ClAuI(SMe2), ClAuI(PMe3), and MeAuI(PMe3). Our experiments show that [ClAuIIIMe2]2 and MeAuI(PMe3) are suitable precursors for Au deposition. They are volatile, have a sufficient vapor pressure at room temperature for deposition experiments, and produce deposits that contain gold. [ClAuIIIMe2]2 yields a gold content of 29−41 atom %, which is (in the absence of any process optimization) already on par with the best reported values for commercial Au precursors. The Cl ligand is easily removed by the electrons, and both methyl ligands are predominantly incorporated into the deposit, with the occasional loss of a single methyl ligand. MeAuI(PMe3) yields a deposit with a Au content of 19−25 atom %. At least one methyl group is removed from the molecule during the deposition. Complexes ClAuI(SMe2) and ClAuI(PMe3) are not suitable as precursors. They dissociate in vacuum, and the only volatile components are Cl, SMe2, and PMe3, respectively. Finally, complexes ClAuI(PMe3) and MeAuI(PMe3) offer an opportunity to understand the origin of the instability of Au complexes in vacuum. Although ClAuI(PMe3) is unstable in vacuum and partially releases its ligands, MeAuI(PMe3) appears to be stable and reaches the gas phase intact. This difference is potentially related to the chemical difference between Cl and Me ligands or differences in dipole moments or in aurophilicity. Further work is necessary to clarify this issue.



ASSOCIATED CONTENT

S Supporting Information *

Reactivity of Ag foil to MeAu I (PMe 3). Sensitivity of ClAuI(SMe2). This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This research was carried out within the Top Research School program of the Zernike Institute for Advanced Materials under the Bonus Incentive Scheme (BIS) of The Netherlands’ Ministry of Education, Science, and Culture. 12103

dx.doi.org/10.1021/la502618t | Langmuir 2014, 30, 12097−12105

Langmuir



Article

Nanowires on Three-Dimensional Microstructures by Using a Minimally Invasive Catalyst Templating Method. Nano Lett. 2011, 11, 4213−4217. (20) Folch, A.; Servat, J.; Esteve, J. High-vacuum versus “environmental” electron beam deposition. J. Vac Sci. Technol., B 1996, 14, 2609−2614. (21) Vericat, C.; Vela, M. E.; Benitez, G.; Carro, P.; Salvarezza, R. C. Self-assembled monolayers of thiols and dithiols on gold: new challenges for a well-known system. Chem. Soc. Rev. 2010, 39, 1805− 1834. (22) Wagner, R. S.; Ellis, W. C. Vapor-liquid-solid mechanism of single crystal growth. Appl. Phys. Lett. 1964, 4, 89. (23) Schmidt, V.; Senz, S.; Gosele, U. Diameter-Dependent Growth Direction of Epitaxial Silicon Nanowires. Nano Lett. 2005, 5, 931−935. (24) Ozbay, E. Plasmonics: Merging Photonics and Electronics at Nanoscale Dimensions. Science 2006, 311, 189−193. (25) Hoflich, K.; Yang, R. B.; Berger, A.; Leuchs, G.; Christiansen, S. The Direct Writing of Plasmonic Gold Nanostructures by ElectronBeam-Induced Deposition. Adv. Mater. 2011, 23, 2657−2661. (26) Graells, S.; Acimovic, S.; Volpe, G.; Quidant, R. Direct Growth of Optical Antennas Using E-Beam-Induced Gold Deposition. Plasmonics 2010, 5, 135−139. (27) Wnuk, J.D.; Rosenberg, S.G.; Gorham, J.M.; Van Dorp, W.F.; Hagen, C.W.; Fairbrother, D.H. Electron Beam Deposition for Nanofabrication: Insights from Surface Science. Surf. Sci. 2011, 605, 257−266. (28) Arumainayagam, C.R.; Lee, H.L.; Nelson, R.B.; Haines, D.R.; Gunawardane, R.P. Low-Energy Electron-Induced Reactions in Condensed Matter. Surf. Sci. Rep. 2010, 65, 1−44. (29) Plank, H.; Noh, J.H.; Fowlkes, J.D.; Lester, K.; Lewis, B.B.; Rack, P.D. Electron-Beam-Assisted Oxygen Purification at Low Temperatures for Electron-Beam-Induced Pt Deposits: Towards Pure and High-Fidelity Nanostructures. ACS Appl. Mater. Interfaces 2014, 6, 1018−1024. (30) Elbadawi, C.; Toth, M.; Lobo, C.B. Pure Platinum Nanostructures Grown by Electron Beam Induced Deposition. ACS Appl. Mater. Interfaces 2013, 5, 9372−9376. (31) Roberts, N.A.; Gonzalez, C.M.; Fowlkes, J.D.; Rack, P.D. Enhanced By-Product Desorption Via Laser Assisted Electron Beam Induced Deposition of W(CO)6 With Improved Conductivity and Resolution. Nanotechnology 2013, 24, 415301−1-6. (32) Utke, I.; Hoffmann, P.; Dwir, B.; Leifer, K.; Kapon, E.; Doppelt, P. Focused Electron Beam Induced Deposition of Gold. J. Vac. Sci. Technol., B 2000, 18, 3168−3171. (33) Mulders, J. J. L.; Veerhoek, J. M.; Bosch, E. G. T.; Trompenaars, P. H. F. Fabrication of Pure Gold Nanostructures by Electron Beam Induced Deposition With Au(CO)Cl Precursor: Deposition Characteristics and Primary Beam Scattering Effects. J. Phys. D 2012, 45, 475301−1-7. (34) ClAuI(CO) product sheet; Strem Chemicals, Inc., 2014. (35) Puddephatt, R. J.; Treurnicht, I. Volatile Organogold Compounds [AuR(CNR′)]:Their Potential For Chemical Vapour Deposition of Gold. J. Organomet. Chem. 1987, 319, 129−137. (36) Messelhäuser, J.; Flint, E. B.; Suhr, H. Laser Induced CVD of Gold Using New Precursors. Appl. Surf. Sci. 1992, 54, 64−68. (37) Tobias, R. S.; Scovell, W. M.; Stocco, G. C. Dimethylgold(III) Halides and Pseudohalides. Reactions, Raman, Infrared, and Proton Magnetic Resonance Spectra, and Structure. Inorg. Chem. 1970, 9, 2682−2688. (38) Jones, P. G.; Lautner, J. Chloro(dimethyl sulfide)gold(l). Acta Crystallogr., C 1988, 44, 2089−2091. (39) Schmidbaur, H.; Shiotani, A. Organogold-Chemie, VI Darstellung komplexer Organogold-Verbindungen durch LigandenSubstitutionsreaktionen. Chem. Ber. 1971, 104, 2821−2830. (40) Paul, M.; Schmidbaur, H. A New Synthesis of Dimethylgold(III) Chloride Using Tetramethyltin. Z. Naturforsch., B 1994, 49, 647−649. (41) LISE laboratory of the Facultés Universitaires Notre-Dame de la Paix, Namur, Belgium.

REFERENCES

(1) Utke, I.; Golzhauser, A. Small, Minimally Invasive, Direct: Electrons Induce Local Reactions of Adsorbed Functional Molecules on the Nanoscale. Angew. Chem., Int. Ed. 2010, 49, 9328−9330. (2) Lassiter, M. G.; Liang, T.; Rack, P. D. Inhibiting Spontaneous Etching of Nanoscale Electron Beam Induced Etching Features: Solutions for Nanoscale Repair of Extreme Ultraviolet Lithography Masks. J. Vac. Sci. Technol., B 2008, 26, 963−967. (3) Gonzalez, C. M.; Timilsina, R.; Li, G. L.; Duscher, G.; Rack, P. D.; Slingenbergh, W.; Van Dorp, W. F.; De Hosson, J. T. M.; Klein, K. L.; Wu, H. M. M.; Stern, L. A. Focused Helium and Neon Ion Beam Induced Etching for Advanced Extreme Ultraviolet Lithography Mask Repair. J. Vac. Sci. Technol., B 2014, 32, 021602−1-9. (4) Fernandez-Pacheco, A.; Serrano-Ramon, L.; Michalik, J. M.; Ibarra, M. R.; De Teresa, J. M.; O’Brien, L.; Petit, D.; Lee, J.; Cowburn, R. P. Three Dimensional Magnetic Nanowires Grown by Focused Electron-Beam Induced Deposition. Sci. Rep. 2013, 3, 1492. (5) Brown, J.; Kocher, P.; Ramanujan, C. S.; Sharp, D. N.; Torimitsu, K.; Ryan, J. F. Electrically Conducting, Ultra-Sharp, High Aspect-Ratio Probes for AFM Fabricated by Electron-Beam-Induced Deposition of Platinum. Ultramicroscopy 2013, 133, 62−66. (6) Mackus, A. J. M.; Dielissen, S. A. F.; Mulders, J. J. L.; Kessels, W. M. M. Nanopatterning by Direct-Write Atomic Layer Deposition. Nanoscale 2012, 4, 4477−4480. (7) Djenizian, T.; Balaur, E.; Schmuki, P. Direct Immobilization of DNA On Diamond-Like Carbon Nanodots. Nanotechnology 2006, 17, 2004−2007. (8) Slingenbergh, W.; De Boer, S. K.; Cordes, T.; Browne, W. R.; Feringa, B. L.; Hoogenboom, J. P.; De Hosson, J.; Th, M.; Van Dorp, W. F. Selective Functionalization of Tailored Nanostructures. ACS Nano 2012, 6, 9214−9220. (9) Kaplan, L. H.; d’Heurle, F. M. The Deposition of Molybdenum and Tungsten Films From Vapor Decomposition of Carbonyls. J. Electrochem. Soc. 1970, 117, 693−700. (10) Xue, Z.; Strouse, M. J.; Shuh, D. K.; Knobler, C. B.; Kaesz, H. D.; Hicks, R. F.; Williams, R. S. Characterization of (Methylcyclopentadienyl)trimethylplatinum and Low-Temperature Organometallic Chemical Vapor Deposition of Platinum Metal. J. Am. Chem. Soc. 1989, 111, 8779−8784. (11) Feurer, E.; Suhr, H. Preparation of Gold Films by Plasma-CVD. Appl. Phys. A: Mater. Sci. Process. 1987, 44, 171−175. (12) Wnuk, J. D.; Gorham, J. M.; Rosenberg, S. G.; Van Dorp, W. F.; Madey, T. E.; Hagen, C. W.; Fairbrother, D. H. Electron Beam Irradiation of Dimethyl-(acetylacetonate) Gold(III) Adsorbed Onto Solid Substrates. J. Appl. Phys. 2010, 107, 054301−1-11. (13) Jenke, M. G.; Lerose, D.; Niederberger, C.; Michler, J.; Christiansen, S.; Utke, I. Toward Local Growth of Individual Nanowires on Three-Dimensional Microstructures by Using a Minimally Invasive Catalyst Templating Method. Nano Lett. 2011, 11, 4213−4217. (14) Folch, A.; Servat, J.; Esteve, J. High-Vacuum Versus “Environmental” Electron Beam Deposition. J. Vac. Sci. Technol.. B 1996, 14, 2609−2614. (15) Botman, A.; Mulders, J. J. L.; Weemaes, R.; Mentink, S. Purification of platinum and gold structures after electron-beaminduced deposition. Nanotechnology 2006, 17, 3779−3785. (16) Mulders, J. J. L.; Belova, L. M.; Riazanova, A. Electron beam induced deposition at elevated temperatures: compositional changes and purity improvement. Nanotechnology 2011, 22, 055302. (17) Koops, H. W. P.; Kretz, J.; Rudolph, M.; Weber, M.; Dahm, G.; Lee, K. L. Characterization and application of materials grown by electron beam induced deposition. Jpn. J. Appl. Phys. 1994, 33, 7099− 7107. (18) Utke, I.; Jenke, M. G.; Roling, C.; Thiesen, P. H.; Iakovlev, V.; Sirbu, A.; Mereuta, A.; Caliman, A.; Kapon, E. Polarisation stabilisation of vertical cavity surface emitting lasers by minimally invasive focused electron beam triggered chemistry. Nanoscale 2011, 3, 2718−2722. (19) Jenke, M. G.; Lerose, D.; Niederberger, C.; Michler, J.; Christiansen, S.; Utke, I. Toward Local Growth of Individual 12104

dx.doi.org/10.1021/la502618t | Langmuir 2014, 30, 12097−12105

Langmuir

Article

(42) Olah, G. A. Onium Ions; John Wiley and Sons: New York, 1998; p 170. (43) McNeillie, A.; Brown, D. H.; Smith, W. E.; Gibson, M.; Watson, L. X-ray Photoelectron Spectra of Some Gold Compounds. J. Chem. Soc., Dalton Trans. 1980, 5, 767−770. (44) Van Attekum, J. M; Th, M.; Van der Velden, J. W. A.; Trooster, J. M. X-ray Photoelectron Spectroscopy Study of Gold Cluster and Gold(I) Phosphine Compounds. Inorg. Chem. 1980, 19, 701−704. (45) Kaushik, V. K. XPS Core Level Spectra and Auger Parameters for Some Silver Compounds. J. Electron Spectrosc. Relat. Phenom. 1991, 56, 273−277. (46) Dadiza, Y. M.; Saleh, J. M. Interaction of Hydrogen Chloride With Evaporated Metal Films. Part 1.Chlorination of Iron, Nickel, Palladium, Silver and Lead. J. Chem. Soc., Faraday Trans. 1 1972, 68, 269−279. Bowker, M.; Woflindale, B.; King, D. A.; Lamble, G. The Coadsorption of K and Cl on Ag(100): Electronic, Kinetic and Thermodynamic Surface Modification by Promotion. Surf. Sci. 1987, 192, 95−106. (47) Moulder, J.; Stickle, W. F.; Sobol, P. E.; Bomben, K. D. Handbook of X-ray Photoelectron Spectroscopy, 2nd ed.; Physical Electronics: Eden Prairie, MN, 1995. (48) Shul’ga, Y.M.; Bulatov, A.V.; Gould, R. A. T.; Konze, W.V.; Pignolet, L.H. X-ray Photoelectron Spectroscopy of a Series of Heterometallic Gold-Platinum Phosphine Cluster Compounds. Inorg. Chem. 1992, 31, 4704. (49) e-EROS: Encyclopedia of Reagents for Organic Synthesis; John Wiley and Sons: New York, 2014. (50) Angermaier, K.; Zeller, E.; Schmidbaur, H. Crystal Structures of Chloro(trimethylphosphine) Gold(I), Chloro(tri-i propylphosphine) Gold(I) and Bis(trimethylphosphine) Gold(I) Chloride. J. Organomet. Chem. 1994, 472, 371−376. (51) Schmidbaur, H. The aurophilicity Phenomenon: A Decade of Experimental Findings, Theoretical Concepts and Emerging Applications. Gold Bull. 2000, 33, 3−10. (52) Martinez-Salvador, S.; Fornies, J.; Martin, A.; Menjon, B. [Au(CF3)(CO)]: A Gold Carbonyl Compound Stabilized by a Trifluoromethyl Group. Angew. Chem., Int. Ed. 2011, 50, 6571−6574.

12105

dx.doi.org/10.1021/la502618t | Langmuir 2014, 30, 12097−12105