Francesco
Pineider
a,
Matteo
Mannini
ab,
Chiara
Danieli
c,
Lidia
Armelao
d,
Federica M.
Piras
e,
Agnese
Magnani
e,
Andrea
Cornia
*c and
Roberta
Sessoli
*a
aLa.M.M.-Department of Chemistry and INSTM research unit, University of Florence, Via della Lastruccia 3, 50019, Sesto Fiorentino (FI), Italy. E-mail: roberta.sessoli@unifi.it; Fax: +39-0554573372; Tel: +39-0554573268
bISTM-CNR, URT Firenze, Via della Lastruccia 3, 50019, Sesto Fiorentino (FI), Italy
cDepartment of Chemistry and INSTM research unit, University of Modena and Reggio Emilia, Via G. Campi 183, 41100, Modena, Italy. E-mail: acornia@unimore.it
dISTM-CNR and INSTM research unit, Department of Chemistry, University of Padova, Via Marzolo 1, 35131, Padova, Italy
eDepartment of Chemical and Biosystems Science and Technology, University of Siena, Via A. Moro 2, 53100, Siena, Italy
First published on 9th November 2009
The ability of a tetranuclear iron(III) single-molecule magnet functionalized with thioacetyl-terminated ligands to form monolayers on gold has been investigated by a multitechnique approach based on STM, XPS and ToF-SIMS. We discuss in detail several aspects, which are relevant to the reported observation of a memory effect on the monolayer prepared from dichloromethane (Mannini et al., Nat. Mater., 2009, 8, 194). In particular we show that the adsorbate comprises intact surface-bound Fe4 clusters as opposed to microcrystals or multilayers. The influence of the solvent used for the self-assembly process on the morphology, composition and structure of the adsorbate is also studied for four different solvents (dichloromethane, n-hexane, toluene, 1,4-dioxane).
A different class of iron(III)-based SMMs, namely Fe4 complexes, has been recently shown to exhibit the chemical and structural robustness required for deposition on surfaces. Their propeller-like structure17–19 (see Fig. 1) stabilizes a S = 5 ground spin state and generates an anisotropy barrier as high as 17 K.20 Furthermore, they can be functionalized at will to promote interaction with different substrates,21 such as gold surfaces22 and silicon substrates23 or carbon nanotubes.24,25
![]() | ||
Fig. 1 Top: molecular structure of the Fe4C9SAc complex with iron atoms drawn as large grey spheres, sulfur atoms as pale grey spheres, carbon as light grey sticks and oxygen as black sticks. Bottom: scheme of the sulfur-functionalized tripodal ligand H3L used in this work. |
Experiments at large-scale facilities using low-temperature XMCD22 have demonstrated that the electronic structure and magnetic properties of Fe4 molecules deposited on gold from dichloromethane (DCM) solution closely resemble those of bulk phases (antiferromagnetic intramolecular interactions, high-spin ground state, hysteresis loop). This important result confirms the intactness of Fe4 complexes at surfaces and, at the same time, provides a crucial proof-of-concept, i.e. that SMMs can be linked to metallic surfaces while not altering their characteristic functionality.
The Fe4 derivative specifically designed for grafting to gold surfaces, hereafter denoted as Fe4C9SAc, has formula Fe4(L)2(dpm)6, where Hdpm is dipivaloylmethane and H3L is the sulfur-functionalised triol ligand 11-(acetylthio)-2,2-bis(hydroxymethyl)undecan-1-ol (Fig. 1).21 Such ligand design is motivated by the strong affinity of sulfur toward gold; notice that acetyl-protected sulfur atoms are used instead of free thiols, because of the redox instability of iron(III) ions in the presence of –SH groups; notice also that acetyl-protected thiols have been used without deprotection by an exogenous base to generate self-assembled monolayers (SAMs).26–28 However it is still debated whether the acetyl group is retained when the Au–S bond is formed.27,28 The long alkyl chain was chosen as an insulating spacer, in order to minimise the interaction of free electrons of gold with the magnetic Fe4 core.
In this work we present a complete morphological, chemical and structural investigation of Fe4C9SAc adsorbates on gold. Four different solvents including DCM have been employed, as strong solvent effects have been already observed in monolayers of Mn12 complexes29 and simpler molecules.30,31 The solvents examined in this study are DCM (low viscosity, moderately polar), n-hexane (low-viscosity, apolar), toluene (moderately viscous, apolar and aromatic) and 1,4-dioxane (very viscous, polar). Information on the morphology of the surface, as revealed by STM imaging, has been complemented by semi-quantitative chemical composition data obtained using XPS spectroscopy and by structural information provided by ToF-SIMS. Monolayers obtained from these solvents, called Fe4C9SAc-exDCM, -exHex, -exTol and -exDiox, respectively, were characterised using the above-mentioned techniques, which demonstrated the intactness of Fe4 units in all samples but Fe4C9SAc-exDiox. Information was gained on the nature of the deposits, which are shown to largely consist of an array of surface-bound complexes rather than aggregates, microcrystals or multilayers. In addition, the arrangement of Fe4 units on the gold surface and the fate of acetyl groups were inferred.
![]() | ||
Fig. 2 STM topography images of typical Fe4C9SAc-exDCM (a), -exHex (c) and -exTol (d) monolayers (tunneling conditions: U = 0.35 V, I = 9 pA). Diameter statistics extracted from four images of Fe4C9SAc-exDCM samples are shown in (b). |
This is probably due to the fact that the tip, while not being able to actually remove the molecules from the surface, as seen instead on Mn12 adsorbates from DCM,29 scans through them producing noisy and unresolved images.
The absence of any long-range ordering in the adsorbate is an expected consequence of the bulky cluster core, which prevents lateral interactions between aliphatic chains and thus disfavours the closely-packed arrangement typical of alkanethiol SAMs.35 Very similar morphologies, i.e. a densely-packed layer of molecules, were found in samples Fe4C9SAc-exHex and -exTol (see Fig. 2c and d); on the other hand, STM imaging of Fe4C9SAc-exDiox samples failed to detect molecular adsorbates at the surface.
STM imaging is an important tool to detect monolayer formation and to optimize deposition protocols, although topographic information alone can be deceptive and can lead to erroneous conclusions. A clearer view on the compositional and structural properties of the adsorbates was provided by XPS and ToF-SIMS data, which are presented in the following sections.
Fe4C9SAc samples | Fe | S | C | O |
---|---|---|---|---|
Calcd for Fe4O20C96S2 | 3.3 | 1.6 | 78.7 | 16.4 |
Drop cast | 2.9 | 1.5 | 78.5 | 17.1 |
BE | 711.1 | 163.9 | 284.8 | 531.2 |
-exDCM Monolayer | 3.7 | 2.0 | 71.5 | 22.8 |
BE | 711.2 | 162.7 | 284.8 | 532.2 |
-exHex Monolayer | 4.6 | 2.4 | 69.5 | 23.5 |
BE | 710.9 | 162.1, 163.6 | 284.8 | 531.9 |
-exTol Monolayer | 3.7 | 2.3 | 73.3 | 20.7 |
BE | 711.2 | 162.2 | 284.8 | 531.9 |
![]() | ||
Fig. 3 Detailed XPS spectra of S2p (a) and C1s (b) peaks carried out on a drop cast and monolayer samples of Fe4C9SAc (labelling scheme in the legend of panel b). In panel a) SI, SII, and SIII indicate the BE of surface-bound, surface-free, and oxidized sulphur components, respectively. In panels (c) and (d) we show the deconvolution for the C1s peaks of the drop cast and the -exDCM absorbate, respectively. In (e), the grey scale coding on the ligands describes the various components used to deconvolute the C1s signal. Fe2p and O1s peaks are reported in ESI.† |
A deeper analysis of the C1s signal was attempted by peak deconvolution, and three components were found (see Fig. 3c): a dominant one at 284.8 eV that can be ascribed to aliphatic and adventitious carbon; a second component at higher BE (286.3 eV) that is attributed to oxygen-bound carbon in the trimethyolol group and in dpm ligands (see Fig. 3e); a very weak component at even higher BE (288.4 eV) that can be coherently assigned to the electron-depleted thioacetyl carbon. Relative integral values of the three components, which are reported in Table 2 along with their BEs and FWHMs, are in close agreement with the expected values. The O1s peak could not be resolved into the three components originating from trimethylol, dpm and acetyl groups. Indeed, a limited peak shift is expected for these types of oxygen atoms.32,36 The sulfur S2p photopeak is centered at 163.9 eV, i.e. close to the BE exhibited by unbound thioacetyls in acetyl-protected dithiols27 or in derivatives of calix[4]arene37 on gold (163.4–163.6 eV for S2p(3/2)).
XPS studies on monolayer samples Fe4C9SAc-exDCM, -exHex and -exTol required very long acquisition times to improve the signal-to-noise ratio. A survey scan, taken in the energy range of 0–1300 eV (see ESI†), is dominated by the strong peaks originating from gold photoelectrons and shows additional signals from iron, sulfur, carbon and oxygen. Detailed scans of the S2p and C1s peaks are shown in Fig. 3a and b respectively, while Fe2p and O1s peaks are reported in ESI.† The spectral features are very similar to those observed in the drop-cast sample, all photopeaks showing relatively low broadening and the expected BEs. No significant charging effects are evident, indicating that the neutrality of the single layer of molecules is easily restored through the conducting gold surface.
The atomic percentages obtained by peak integration (Table 1) are in reasonable agreement with the calculated ones. In particular, the Fe:S ratio is in the range 1.6–1.9 (vs. 1.9 in the drop-cast sample and an expected value of 2.0), suggesting that the clusters are largely intact at the gold surface. On the other hand, carbon and oxygen integrals do not perfectly agree with the expected values, the former being less than expected, and the latter in slight excess.
The analysis of the S2p photopeaks provides important information on the arrangement of the Fe4 units on the gold surface. According to the molecular structure revealed by X-ray diffraction, the alkyl chains are long enough to allow the simultaneous binding of the two sulfur-based “alligator clips” to the gold surface, giving rise to a surface-parallel (or lying down) phase, using classical SAM terminology.27
Alternatively, Fe4 units could interact asymmetrically with the surface via only one sulfur atom, affording an upright (or standing up) phase as schematized in Fig. 4.
![]() | ||
Fig. 4 Schematic view of the standing-up and lying-down grafting modes of the Fe4C9SAc SMM, with the relative orientation of the easy axis of magnetization indicated by the black arrow. The sulfur atom not bound to gold of the standing-up configuration is depicted as a dark sphere. |
Indeed, the S2p peak in Fe4C9SAc-exDCM and Fe4C9SAc-exTol samples is observed at 162.7–162.2 eV vs. 163.9 eV in the bulk reference (Fig. 3). The shift toward lower BE is taken as evidence of an interaction with gold27,36–38 (a similar shift is observed in the S2s photopeaks reported in ESI†). Since the peak is slightly broader as compared with bulk Fe4C9SAc, we cannot completely rule out the presence of a fraction of surface-free S atoms in these samples. The contribution from unbound sulfur is clearly detectable in the Fe4C9SAc-exHex sample, whose S2p photopeak can be fitted with two components centered at 162.1 (surface-bound sulfur, component SI in Fig. 3a) and 163.6 eV (surface-free sulfur, component SII). From the integrated areas, the two sulfur species occur in comparable amounts in the -exHex sample (see ESI†). We attribute these surface-free sulfur atoms to molecules wired to the surface via a single alligator clip or, less likely, forming an overlayer. The simultaneous presence of both surface-parallel and upright configurations of the SMM, with a prevalence of the first, would be consistent with the powder-like magnetic response of the Fe4C9SAc-exDCM sample observed within the resolution of XMCD measurements.22 In fact, the lying-down and standing-up molecules are expected to have their easy axes of magnetization in the plane of the substrate and roughly orthogonal to it, respectively (see Fig. 4). A 2:1 proportion of surface-perpendicular and parallel configurations would closely mimic a random distribution, while the ratio between surface-bound and surface-free sulfur atoms (5:1) would still be consistent with XPS data.
Deconvolution of the C1s signal39 (Fig. 3d) afforded the results reported in Table 2 for the Fe4C9SAc-exDCM sample. Similar results were obtained for Fe4C9SAc-exHex and Fe4C9SAc-exTol, the C1s spectra being superimposable. The excellent agreement with the calculated percentages provides a strong indication that acetyl groups are still present in the monolayers.40 From XPS data alone, however, it is not possible to ascertain whether acetyl groups are still bound to sulfur or remain quantitatively entrapped in the monolayer.28 On the other hand, Fe4C9SAc-exDiox samples showed peculiar features, with no detectable XPS signal from sulfur, a little amount of carbon with a shoulder at higher BE and anomalous amounts of iron and oxygen (see Fig. 3a,b and ESI† for details). This finding is coherent with the results of STM imaging (see previous section), that was not able to evidence morphological features comparable to those observed in the other samples.
First, positive and negative ToF-SIMS spectra of Fe4C9SAc drop cast on Si were acquired as a bulk reference (Fig. 5).
![]() | ||
Fig. 5 Positive (a) and negative (b) ToF-SIMS spectra of Fe4C9SAc drop cast on Si (* signals from silicon grease contamination). |
The characteristic peaks detected and their assignments are listed in Table 3. In the high mass region (m/z 800–2200) of the positive ion spectrum (Fig. 5a), peaks are detected which are assigned to ion fragments generated by loss of one, two, and three dpm moieties from the molecular ion. Similar ion fragments have been previously observed on thin films of non-functionalized Fe4 derivatives obtained by thermal evaporation.41
Drop cast | Monolayers on Au | Fragment | |||
---|---|---|---|---|---|
-exDCM | -exHex | -exTol | -exDiox | ||
Positive ions | |||||
56 | 56 | Fe+ | |||
57 | 57 | 57Fe+/FeH+ | |||
— | 197 | Au+ | |||
239 | 239 | [Fe + dpm]+ | |||
422 | 422 | [Fe + 2dpm]+ | |||
733 | 733 | [3Fe + 3dpm + O]+ | |||
1036 | 1036 | [3Fe + 3dpm + L + O]+ | |||
1380 | very weak (vw) | [M − 3dpm]+ | |||
— | 1488 | [M − 2dpm − SAc]+ | |||
— | 1520 | vw | 1520 | — | [M − 2dpm − Ac]+ |
1563 | 1563 | vw | 1563 | — | [M − 2dpm]+ |
— | — | — | — | 1568 | [M* − dpm]+ |
weak | 1671 | [M − dpm − SAc]+ | |||
— | — | — | 1686 | — | [M − dpm − SAc − H + O]+ |
— | 1702 | 1702 | 1702 | — | [M − dpm − Ac]+ |
1746 | 1746 | [M − dpm]+ | |||
— | 1900 | [M − dpm − Ac + Au]+ | |||
— | 2051 | vw | 2051 | — | [M − SAc + Au]+ |
— | — | — | 2066 | — | [M − SAc − H + O + Au]+ |
— | 2083 | vw | 2083 | — | [M − Ac + Au]+ |
— | 2126 | [M + Au]+ | |||
Negative ions | |||||
32 | 32 | S− | |||
75 | 75 | [SAc]− | |||
183 | 183 | [dpm]− | |||
— | 197 | Au− | |||
902 | — | [2Fe + dpm + 2L]− | |||
1085 | — | [2Fe + 2dpm + 2L]− | |||
1507 | — | [M − Fe − 2dpm]− | |||
1929 | — | 1929 | — | 1929 | [M]− |
— | 1934 | [M*]− ≡ [M − Ac + 3O]− | |||
— | vw | vw | vw | 2131 | [M* + Au]− |
The intense peak at m/z 1036 is attributed to the ion fragment with formula [3Fe + 3dpm + L + O]+ originated by rearrangement of the Fe4 complex after loss of one Fe atom, three dpm groups and one tripodal ligand. In the low mass region (m/z 0–800, Fig. 5a), Fe+ and other characteristic ion peaks are observed. The high mass region of the negative ion spectrum (m/z 800–2200, Fig. 5b) is characterised by the molecular ion peak at m/z 1929 ([M]−), and other signals assigned to molecular ion fragments generated by loss of Fe atoms and/or dpm moieties.
Positive and negative ToF-SIMS spectra were then recorded on Fe4C9SAc monolayers prepared by dissolving the Fe4 complex in the four solvents. The recorded spectra present a rich variety of molecular secondary ions and show peaks assigned to molecular ions ([M]−), to key molecular ion clusters ([M + Au]+) and fragments (e.g. [M − dpm]+, [M − 2dpm]+, [M − 3dpm]+), demonstrating the presence of Fe4C9SAc complexes on Au.
In Fig. 6, we gather the high mass regions (m/z 800–2200) of the positive ion spectra recorded on the four samples. Besides the distinctive peaks of the Fe4C9SAc complex detected in the drop-cast sample, new characteristic signals are observed in the spectra of monolayers (Table 3). The Fe4C9SAc-exDCM sample (Fig. 6a) shows peaks at m/z 1702 ([M − dpm − Ac]+) and m/z 1520 ([M − 2dpm − Ac]+) assigned to molecular ion fragments generated by loss of the acetyl group, as well as ion–Au and molecular fragments–Au clusters at m/z 1900 ([M − dpm − Ac + Au]+), m/z 2083 ([M − Ac + Au]+), and m/z 2126 ([M + Au]+). These last peaks are not observed in the drop-cast sample, thus excluding that they are originated by the combination with the Au+ of the beam. The positive ion spectra acquired on Fe4C9SAc-exHex, -exTol, and -exDiox show some differences and similarities in the fragmentation patterns.
![]() | ||
Fig. 6 Positive ToF-SIMS spectra of Fe4C9SAc-exDCM (a) and (b), -exHex (c), -exTol (d), and -exDiox (e) SAMs on Au. The mass regions m/z 800–2200 (a) and m/z 1500–2200 (b–e) are shown. |
The -exDCM and -exHex (Fig. 6c) spectra are identical, except for differences in the peak relative intensities, whereas the -exTol (Fig. 6d) spectrum shows additional features. Finally, the -exDiox spectrum (Fig. 6e) differs from the previous three spectra by the absence of the peaks at m/z 1900, 1702, and 1520 and the presence of a new peak at m/z 1568 tentatively attributed to an oxidized molecular ion fragment [M* − dpm]+ (Table 3), as discussed later.
The high mass regions (m/z 800–2200) of the negative ion spectra are characterised by very intense signals from the Au substrate (Fig. 7). All monolayer spectra show an intense peak at m/z 1934, slightly above the molecular ion. A more detailed investigation of the isotopic distribution (see ESI†) allows us to assign it to a species [M*]− with formula [Fe4(LL*)(dpm)6]−, where L* is an oxidized tripodal ligand containing a terminal sulfonate group (R–SO3−). Such a partially-oxidized species is absent in the drop-cast sample and is accompanied by the molecular ion peak in -exHex and -exDiox samples.
![]() | ||
Fig. 7 Negative ToF-SIMS spectra of Fe4C9SAc drop cast (a) -exDCM (b), -exHex (c), -exTol (d), and -exDiox (e) SAMs on Au. |
The oxidation of sulfur to sulfonate has been previously observed in non-tightly packed layers of thiolated molecules upon exposure to atmospheric oxygen.42–44 The extent of oxidation was closely related to the surface coverage and to the layer thickness, which determine the permeability of the layer to atmospheric oxidants.45
A similar situation is expected to hold in the monolayers under study, where a tight-packing arrangement of alkylsulfanyl chains cannot be achieved due to the bulky cluster core. Notice that a hint of a S2p component around 168–169 eV (SIII component in Fig. 3a), attributable to oxidised sulfur,27,37 is only visible in the XPS spectra of the -exHex and -exDiox samples. The failure to clearly resolve such oxidized sulfur species in XPS spectra may be related to the different sensitivities and detection mechanisms of the two techniques. In fact, whereas XPS provides a quantitative detection of all S-containing species, the ToF-SIMS spectra are influenced by the interaction strength of the analyte with the substrate, as well as by ionization yields. In this respect, sulfonates are expected to be very easily detected by the ToF-SIMS technique because of their weak interaction with the surface and their negative charge. As a confirmation of our hypothesis, the ozone-induced oxidation of alkanethiol monolayers on gold was reported to become detectable at a much earlier stage by ToF-SIMS than by XPS.46
The absence of the [M*]− peak in the bulk reference as well as other peculiarities observed in the fragmentation patterns of -exDCM, -exHex and -exTol monolayers, like the extensive clustering with Au, support the occurrence in these samples of a large fraction of intact Fe4 complexes interacting with the Au surface via their terminal alligator clips. In addition, fragments are detected in all monolayer samples which still contain one ([M − 2dpm − SAc]+, [M − dpm − SAc]+, [M − dpm − Ac]+, [M − dpm − Ac + Au]+, [M*]−, [M* + Au]−) or two ([M + Au]+, [M − dpm]+, [M − 3dpm]+) acetyl groups. Combined with the XPS signature of an extensive S–Au interaction (especially in sample Fe4C9SAc-exDCM), the ToF-SIMS data suggest that absorption can be largely (though not exclusively) promoted by unprotected thioacetyls.
Deposition from n-hexane afforded similar results, though with a broader S2p photopeak, taken as evidence of comparable amounts of surface-bound (162.1 eV) and surface-free (163.6 eV) sulfur. By contrast, use of 1,4-dioxane failed to give adsorbates showing the expected morphology and XPS signatures.
The work presented here not only addressed the questions concerning the specific system under study, but also suggests some general methodological guidelines for the study and characterisation of monolayers of complex systems. The deposition of intact SMM layers is directly proved by their XMCD-detected magnetism.22 However, laboratory-scale instrumentation can be employed with success to individuate the most effective deposition protocols, provided that several techniques are used in parallel. In this respect, combined STM, XPS and ToF-SIMS studies contributed to gain considerable insight into the nature of SMM adsorbates and must be regarded as a valuable tool for the investigation of nanostructured molecular materials.
Footnote |
† Electronic supplementary information (ESI) available: Details of ligand and samples preparation, 2H-NMR stability experiments, XPS and ToF-SIMS additional spectra. See DOI: 10.1039/b916895h |
This journal is © The Royal Society of Chemistry 2010 |