Iron Phthalocyanine and Ferromagnetic Thin Films: Magnetic Behavior of Single and Double Interfaces.

Metal-phthalocyanines are quasi-planar heterocyclic macrocycle molecules with a highly conjugated structure. They can be engineered at the molecular scale (central atom, ligand) to tailor new properties for organic spintronics devices. In this study, we evaluated the magnetic behavior of FePc in a ∼1 nm molecular film sandwiched between two ferromagnetic films: cobalt (bottom) and nickel (top). In the single interface, FePc in contact with a Co film is magnetically coupled with the inorganic film magnetization, though the relatively small Fe(Pc) X-ray magnetic circular dichroism (XMCD) signal in remanence, with respect to that observed in applied field of 6 T, suggests that a fraction of molecules in the organometallic film have their magnetic moment not aligned or antiparallel with respect to Co. When in contact with two interfaces, Fe(Pc) XMCD doubles, indicating that part of the Fe(Pc) are now aligned with the Ni topmost layer, saturated at 1 T. We discussed the relevance of the finding in terms of understanding and developing hybrid organic/inorganic spin devices.

Introduction

A spin-valve is a layered structure consisting of two ferromagnetic (FM) electrodes of different coercivity decoupled by a spacer and displaying a resistance that depends on the spin state of electrons that pass through it (magnetoresistance). The π-conjugated organic semiconductors are interesting as spin devices due to their intrinsic lattice flexibility and long range spin coherence. Two types of organic spin valves have been designed displaying the organic layer sandwiched between two ferromagnetic (FM) electrodes either aligned vertically[1,2] or horizontally[3]. Spin polarization and transmission at distances in the range of tens of nanometers from an injection organic/FM interface have been observed in NiFe 17 nm/LiF 1.9 nm/Alq3 200 nm/TPD 50 nm/FeCo 17 nm by muon spin rotation[4] and in the Co/CuPc interface by two-photon photoemission spectroscopy.[5] These results indicate π-conjugated organic semiconductors as a promising spacer for spintronic devices. However, the understanding of how and which organic materials to incorporate in spintronic devices still requires careful examination in spite of intensive investigation, and that is the reason why several experimental and theoretical works have been focused on the characterization of single organic molecule/FM interfaces.

Paramagnetic organometallic molecules are good candidates for designing a robust organic/inorganic interface. In particular, flat molecules like metal-phthalocyanine (MPc) have the advantage of self-assembling in a well-organized structure on metal surfaces, i.e., reducing the structural defect and increasing charge mobility. MPc films exhibit a semiconductor behavior with a gap (highest occupied and lowest unoccupied molecular orbital, HOMO–LUMO gap) that ranges from 1 to 1.6 eV depending on the central metal atom.[6] Their magnetic properties depend on the crystallographic structure and the metal distance within the film lattice. In the case of CoPc, some studies revealed an antiferromagnetic coupling between Co atoms of different CoPc chains within α and β crystallographic phases, as well as in template thin film grown on flexible polymeric substrates[7] or Pb islands.[8]

Differently, a iron phthalocyanine (FePc) film grown flat on gold surfaces is paramagnetic (ferromagnetic) above (below) ∼4.5 K.[9] These examples point out that MPc thin films can be thought of as a semiconductor endowed with quantum spin chains and consequently functional for designing organic spintronics device.[7] Up to now, Co/MPc interface was identified as a promising spin-polarized current source at room temperature (RT).[10] The robust magnetism of the Co/MnPc interface has been observed to favor the manganese phthalocyanine (MnPc) antiferromagnetic ordering within subsequent layers at room temperature and the induction of exchange bias at low temperature.[11]

Magnetoresistance transport experiments have been performed in molecular magnetic tunnel junctions based on CoPc showing an inverse tunnel magnetoresistance in Co/CoPc/Co and a unidirectional anisotropy at the interface due to asymmetric coupling between nominally equivalent Co/CoPc interfaces.[12] Due to the complexity of the structure and the magnetic phase for MPc, further studies are required to understand the feasibility of integrating MPc film as a component of vertical hybrid organic spin-valve.

The formation and characterization of ferromagnetic films based on organic interfaces are being widely studied and different attempts to improve the response of spin organic devices are done by using alkali metals in the molecular matrix[13] and organic ferromagnetic layers instead of metallic,[14] among others.

In a recent research paper,[12] the authors discuss as promising the fabrication of nanodevices in a vertical structure consisting of Co/CoPc/Co and its magneto-transport properties. On the other hand, in other work,[15] the Co film formation on top of an Alq3 film gives rise to an ill-defined metal film with a direct influence on the magneto-transport properties of the prototypical device.

The aim of our paper is to investigate the formation of a metallic FM layer on top of an organic thin film (FePc) by measuring the magnetic response of the system by means of element-specific experimental techniques, with particular focus on the role of molecular thin film as a spacer between two FM inorganic films.

Thin cobalt and nickel films are ferromagnetic at room temperature, displaying coercivity of 0.05 and 0.025 T, respectively; the latter depending on thin film thickness.[16] Co spin polarization has been reported to be 20%,[17] whereas Ni has a maximum spin polarization of >46%.[18] We intend to disclose the following relevant issues: (i) the chemical interaction between the organic molecular film (FePc) and the atop FM layers; (ii) the role of the bottom FM film in the FePc film growth and its magnetic response; and (iii) the properties of FePc film when in contact with two inorganic ferromagnetic films.

The investigation has been carried out by means of X-ray spectroscopies: photoemission (XPS), absorption (XAS), and magnetic circular dichroism (XMCD). Because of the element specificity of these experimental techniques, we could follow the evolution of core-level line shape of all the elements present in the grown heterostructure from the top Ni film through the organic FePc semiconductor film to the bottom Co layer, as well as the absorption edges features modifications in the presence of an external magnetic field or in remanence after the magnetization saturation. Particular emphasis was given on the FePc magnetic moment alignment with the bottom and top FM films.

Results and Discussion

Figure a,b shows the C 1s and N 1s photoemission line shape evolution of FePc (3 nm)/Co (1.4 nm)/Cu(100), hereafter FePcCo, upon Ni deposition. The quality of the Ni/FePc interface is probed as a function of nickel film thickness: 2, 3, 4 nm. In the C 1s core-level signal of the pristine FePcCo film, two components are identified with intensity ratio and energy position corresponding to benzene CB and pyrrole CP, being the signature of the FePc molecule. The N 1s spectrum presents a single feature and its satellite corresponding to the molecular N sites not resolved in the spectra. After the first nickel deposition, there is an upshift of the C 1s and N 1s spectra of ∼0.6 eV and a change of their spectral shape with the appearance of a shoulder at lower binding energies (BE). The latter corresponds to C and N atoms bound with nickel atoms, C–Ni and N–Ni (Figure a,b). The spectral shift after metal deposition can be attributed to the formation of an interface dipole[19] or an incomplete core–hole screening of small metal clusters.[20] To have further insight into N and C environments, we performed XPS measurements of Ni (4 nm)/FePc (3 nm)/Co (1.4 nm)/Cu(100), NiFePcCo, at grazing and normal emission as reported in the bottom panels of Figure d,e to probe the sample at different depths. C 1s and N 1s are normalized to the most intense feature. The C–Ni and N–Ni peaks decrease going from normal to grazing emission geometry, indicating that the Ni-derived features are detected at different depths with respect to the Ni film top surface. This behavior can be interpreted in terms of both interface roughness or Ni diffusion in the molecular matrix (Supporting Information) of the order of ∼5 Å. Figure f illustrates a cartoon picture of the formed heterostructure.

Figure 1

C 1s (a) and N 1s (b) XPS spectra measured at 480 eV vs Ni thickness. (c) Normal emissions θ = 0 and 45° take off geometries of XPS experiment. (d, e) C 1s and N 1s of Ni (4 nm)/FePc (3 nm)/Co (1.4 nm)/Cu(100) at normal emission and θ =45° take off. (f) The schematic illustration of the two interfaces profile, Co/FePc and FePc/Ni, desumed by X-ray linear dichroism (Supporting Information) and angle-dependent XPS measurements. (g) Valence band of Co(1.4 nm)/Cu(100), FePc (3 nm)/Co (1.4 nm)/Cu(100), and Ni (4 nm)/FePc (3 nm)/Co (1.4 nm)/Cu(100) films measured at 480 eV.

The magnetic properties of NiFePcCo layers are strictly connected to the electronic distribution close to the Fermi edge (EF). Figure g shows the valence band of Co, FePcCo, and NiFePcCo films measured at 480 eV. The valence bands of the bottom and top films are characterized by 3d–4s band close to the Fermi edge; in the panel of Co film, the residual contribution of the Cu substrate is highlighted by gray shadow area. The valence band of the FePc film is dominated by the highest occupied molecular state, HOMO, at BE of 0.7 eV and other molecular states at higher BE of 3.3 and 5.1 eV. All molecular features disappear and Ni density of states is predominant in the vicinity of the Fermi level after Ni deposition. This is due to difference in both Ni coverage and photoelectron cross sections (σNi ≫ σN, σC, see the Supporting Information).

Figure a,b shows the N 1s-π* and Fe 2p–3d transition of the FePcCo sample before and after Ni deposition measured at 45°, which shows similar trend as a function of Ni thickness as acquired at 0° (not shown). The null X-ray linear dichroism of the thick molecular film suggests that molecules of the upper layers are randomly oriented with respect to the substrate surface plane (the opposite is expected for the monolayer or sub-monolayer coverage, details in the Supporting Information).

Figure 2

N K XAS (a) and Fe L3 XAS (b) spectra before (gray shadow) and after (light cyan curve) Ni deposition. (c) Ni L2,3 XAS spectra vs Ni thickness. Hysteresis loops related to Co (d) and Ni films (e) obtained by recording the L3 intensity as a function of magnetic field applied along the [1̅10] direction of Cu(100) substrate. All measurements were performed at room temperature.

The XAS spectra of the FePcCo film present the characteristic features of two N sites, N1 and N2, at 396.7 and 397.1 eV, respectively, for the N K-edge (gray shadow curve in Figure a) and 3d anisotropic electronic distribution of FePc molecule with a prominent peak at 703 eV for the Fe L3 (shadow gray curve in Figure b), in agreement with the already published results.[21] Both N K and Fe L3 XAS spectral line shape broadens after Ni deposition, the intensity ratio of Fe 3d peaks modifies, and the multiplet structure is almost lost. The N K absorption edge is not sharp anymore, but the electronic states are spread across the edge energy. The relative intensity modifications within the spectral feature is an indicator of the established interactions with the top layer; in particular, the reduction of N2 component points to the partial filling of the corresponding π* molecular orbital and, therefore, an involvement of this nitrogen state in the interaction with Ni.

The Ni XAS spectra taken at the Ni L2,3 edge show L3/L2 ratio of approximately 1.8 and a feature of 6 eV above the L3 main component that are consistent with the formation of a metal-like Ni film.[22]

The Co and Ni L3 XMCD hystereses are obtained by sweeping the magnetic field along the [1̅10] direction of the underlying Cu(100) from −0.05 to 0.05 T, with the photon helicity vector projection aligned parallel or antiparallel to the magnetization axis at room temperature (experimental geometry in the Supporting Information).

A null Ni dichroic signal was observed up to 4 nm; at this thickness, a signal as low as 0.1% appears along with a magnetization curve that does not show a clear hysteresis (Figure d). On the other hand, a rectangle-like MH curve was observed for Co film (Figure e), indicating that a field of 0.05 T is enough to magnetize the epitaxial Co film at room temperature but not for inducing the magnetization of the Ni film grown atop the molecular layer. No XMCD signal has been observed at the Fe L2,3 edges.[15]

To explore in detail the magnetic behavior of the FePc film in the heterostructure, we performed XMCD experiments on similar specimens exposed to high magnetic field (6 T) and at low temperature (<20 K). To this end, test multilayered structures were prepared such as FePc (1 nm)/Co (0.5 nm)/Cu(111), indicated hereafter as FePcCo1, and Ni (0.6 nm)/FePc (1 nm)/Co (0.5 nm)/Cu(111), labeled as NiFePcCo1. In Figure c, we report Co, Fe XAS and XMCD spectra measured on CoFePc1. The XMCD signal is normalized to the L3 XAS intensity of average spectra. Co presents a XMCD signal of ∼65% at the L3 edge. The spectral line shape shows a typical fine structure of unperturbed FePc thin film, the XMCD signal is slightly anisotropic (changes as a function of the incident angle) and has a maximum extent of ∼33%.

Figure 3

(a) Experimental geometry for XMCD experiment; θ is the angle between the sample normal and the X-ray incident photon beam. The magnetic field is aligned along the impinging photons direction; FePc (1 nm)/Co (0.5 nm)/Cu(111): (b, c) X-ray absorption and XMCD spectra acquired at the Co, Fe L3 edges as a function of θ. A background was subtracted from all XAS spectra. (d) Magnetization curve of Fe and Co before Ni deposition.

Figure d illustrates the Co and Fe(Pc) magnetization curves before Ni deposition. A Co hysteresis curve appears with the usual rectangular shape and a coercive field (Hc) of 0.05 T, when the X-ray beam and magnetic field are at θ = 60° from the surface normal.

The FePc hysteresis exhibits a small intensity and an opposite sign with respect to one of the underlying Co films. It is worth noting that for the applied magnetic field lower than 0.1 T, almost no Fe(Pc) paramagnetic moment are aligned at either at room or low temperature. The magnetization curves of the bottom ferromagnetic film and the molecular film vs H has been observed and reported by Gruber et al. to be opposite in sign,[3] with an average manganese magnetic moment in the topmost layers of MnPc that is aligned antiferromagnetically with the bottom FM film.

In the FePcCo1 heterostructure, the Fe magnetization curve is opposite to the Co one in the low magnetic field region, whereas for higher magnetic field, it shows a paramagnetic behavior. It is worth remembering that in the case of 1 nm FePc grown on Co film, we did not characterize the magnetic behavior of FePc at the interface, thoroughly studied in ref (23), but only the magnetic behavior of the whole film interacting with the FM substrate. Independent of the preparation conditions, FePc in the first layer in contact with Co film shows ferromagnetic coupling.[23] Therefore, the 1 nm FePc thick film grown on Co justifies the simultaneous presence of either a ferromagnetic contribution, associated with the first ordered layer in contact with the Co film, or antiparallel aligned or paramagnetic molecules, associated with the topmost layers as in ref (3), the latter contribution has a predominant role in the magnetization curve response in the system considered as a whole.

We note that FePc behavior in FePcCo1 is at variance with respect to the FePc film grown on metal substrate; in the latter, an intrinsic FM behavior has been observed in similar (T, H) conditions.[9]

X-ray absorption spectra, measured at the Fe and Ni L2,3 edges for NiFePcCo1 multistructure at θ = 0° (top) and XMCD as a function of θ (bottom), are displayed in Figure a,b. The Ni XAS spectra are fully consistent with metallic film formation[24] and Ni XMCD signal is of ∼31% at L3 edge. The broad Fe L2,3 XAS spectra reflect a band formation, consistent with the molecular states hybridized with Ni atoms of the top layer and a ∼65% XMCD at the L3 edge. Co, Fe, and Ni located in different layers of the sample show no change in XMCD spectra measured at 6 T as a function of θ, and therefore absence of anisotropy. The relative alignment of Fe(Pc) with respect to the two FM films is important for understanding the magnetic response of the heterostructure. The Fe XMCD spectra before (Figure c) and after (Figure a) Ni deposition show an attenuation of Fe L3 XAS due to the Ni top layer that is counterbalanced by an increase of the Fe XMCD signal (from 30 to 65%) due to an additional part of the molecular film aligning its magnetic moment to the magnetization of the top layer.

Figure 4

Ni (0.6 nm)/FePc (1 nm)/Co (0.5 nm)/Cu(111): (a, b) XAS at θ = 0° incidence (top panels) and XMCD at θ = 60, 30, and 0° incidence (bottom panels) spectra are measured at L2,3 edge for Fe and Ni. The XMCDs have been acquired at ±6 T for both photon helicities. (c, d) Individually normalized magnetization of Co, Fe, and Ni as a function of the applied magnetic field at 20 K. To acquire the magnetization curve, we measure the XMCD intensity at a single energy, L3 edge for Co, Fe, and Ni at grazing (θ = 60°) and normal (θ = 0°) incidence.

To evaluate the influence of two FM layers on the FePc film, we measured the magnetization curve of Co, Fe(Pc), and Ni by measuring the helicity-dependent absorption at the L3 edge up to a field of ±6 T at θ = 60 and 0° (Figure c,d). Element-specific hysteresis loops of Co, Fe(Pc), and Ni layers are rectangular at θ = 60° and an almost zero remanent field at θ = 0°. In all layers, the magnetization reaches its saturation value at H = 1 T. Ni and Fe are fully magnetically coupled with a Hc of 0.035 T, whereas Co shows a larger coercive field of 0.055 T.

To disentangle the contribution of bottom and top metallic layers in determining the magnetic properties of FePc thin film, we measured the XMCD spectra of different test interfaces in remanence, and the results are shown in Figure : (1) FePc (0.4 nm)/Co (1.1 nm),[23] (2) Co (0.4 nm)/FePc (1.5 nm), and (3) NiFePcCo1. (1) was measured in remanence and at RT after magnetization with a field of 0.05 T, whereas (2) and (3) were acquired after applying a magnetic field of 6 T along the substrate surface and at 20 K. In the interfaces (1) and (2), FePc is in contact with only one Co film either at the bottom or at the top of molecular film. In (2), the FePc thin film’s thickness is large enough to guarantee negligible electronic interaction with the underlying Cu substrate, which is also nonmagnetic and optimal for evaluating the organic film behavior as a function of the top magnetic layer. Since it is expected for FePc to have paramagnetic behavior in the absence of magnetic coupling, the Fe XMCD signal observed in the remanence in sample (2) demonstrates that part of the FePc molecules aligns with respect to the magnetized Co top layer. The proportion of Fe atoms aligned to the magnetization of Co film differs in (1) and (2) due to the number of Fe magnetically active in the two cases. It has been shown that for thin FePc film grown on Co (1.1 nm)/Cu(001), the Fe(Pc) magnetic signal originates entirely from the FePc molecules at the interface with Co (first layer), with no contributions given by molecules present in the layers far from the interface.[23]

Figure 5

X-ray absorption spectra are acquired at Fe L2,3 edges with right- and left-circularly polarized X-ray beam at remanence with magnetization aligned along the sample surface under the applied magnetic field of 0.05 T (a) and 6 T (b, c). The XMCD results are the difference in XAS measurements normalized at L3 intensity. The results are displayed for each interface: (a) FePc (0.4 nm)/Co (1.1 nm)/Cu(111);[23] (b) Co (0.4 nm)/FePc (1.5 nm)/Cu(111), and (c) Ni (0.6 nm)/FePc (1 nm)/Co (0.5 nm)/Cu(111). The specific experimental geometry in remanent measurement is illustrated in each panel.

As a consequence, we can compare the XMCD results on the FePc in the double interface, where the molecular film is sandwiched between Co and Ni layer. The reduction of FePc XMCD signal to approximately one half in the measurements at remanence with respect to those under applied magnetic field allows us to estimate the amount of FePc molecules initially in the paramagnetic phase or possibly antiferromagnetically coupled with Co film and subsequently aligned with the top layer.

As far as the Ni/FePc/Co double interface is concerned, going along the sample from bottom to top interface, the strong magnetic alignment of FePc moment with respect to the bottom Co FM film is blurred by the presence of the molecules in the uppermost layer that is aligned either antiferromagnetically with the other molecular layers or is not polarized, i.e., in paramagnetic phase. When the top interface is formed, i.e., Ni/FePc, Fe(Pc) magnetic response increases in terms of XMCD signal as a consequence of the interaction between the organic topmost layers and the Ni-magnetized film. The extent of Fe XMCD signal, observed in Figure b,c, indicates a substantial independence from the type (Co, Ni) of the overlayer ferromagnetic film. The interesting point is that in the presence of two interfaces, the magnetization curve of Fe and Ni are overlapping, indicating how much the magnetic response of the FePc molecules is localized to the region where they are in contact with Ni atoms (inset of Figure c). The contact region definition at the interface between the organic and FM film must play a role in the extent of in-plane magnetization of Ni thin film (<1 nm) at room temperature. Our findings are in agreement with the results obtained for the magnetic coupling of Co film grown on top of Alq3 molecular film, where the broad definition of the interface affects both the top layer and molecular film response.[15]

Conclusions

We prepared a multilayer vertical heterostructure in which FePc thin film is sandwiched between two metallic films: Co and Ni. The in situ characterization by means of synchrotron-radiation-based X-ray spectroscopies pointed out that bottom (FePc/Co) and top (Ni/FePc) interfaces differ in terms of the contact region definition and their magnetic behavior. In particular, when the metallic (Ni) film is deposited on top of the organic one, a chemical shift is observed on the molecular core level and XAS spectra as a fingerprint of the ongoing interaction. The so-formed interface requires a thickness of approximately 4 nm for the top film to magnetically couple with the organics at RT. The single FePC/Co interface has been shown to have a robust coupling between the organic layers and the metallic film and stable at room temperature already with a magnetic field intensity of ∼0.05 T. In the case of metallic films grown on top of the organic one low temperature (<20 K) and high magnetic field (1 T) are required to fully polarize the organometallic layers. We evaluated the magnetic response of each layer by using the XMCD spectra at (Co, Fe, Ni) L2,3 edges for the heterostructure. A careful comparison of the XMCD measurements carried out on the molecular film at the Fe L2,3 edges in the presence of applied magnetic field or in remanence after saturating the sample magnetization in plane allowed us to estimate that a fraction of molecules (most reasonably the closest to the interface) interact with the topmost metallic layer (Co and Ni).

Experimental Details

The experiments were performed using the equipments available at DEIMOS (SOLEIL) and APE-IOM-CNR (ELETTRA) synchrotron beamlines. Cu(111) and Cu(100) single crystals have been used as substrate for the deposition of Co metal at DEIMOS and APE, respectively, and Cu(110) for the evaporation of metal–organic films at DEIMOS. We adopted a similar preparation procedure to clean the crystals surface, i.e., several cycles of Ar+ sputtering and annealing at 700 K, which guaranteed the quality of surface as verified by low-energy electron diffraction, and XPS (APE). FePc powders (Sigma-Aldrich) were outgassed in ultra-high vacuum and sublimated in a base pressure <10–7 Pa, with a deposition rate of 0.1 nm/min (as monitored in situ by a quartz microbalance). Co and Ni films were deposited using e-beam evaporators at a pressure lower than 5 × 10–7 Pa. At APE, core-level photoemission spectra were collected using a photon energy of 480 eV with the polarization vector oriented in the scattering plane including the sample surface normal and the analyzer entrance slit plane. The photoemitted electrons were collected by a hemispherical Omicron EA-125 analyzer at normal emission with an overall energy resolution (light source + analyzer) better than 200 meV. X-ray absorption (XAS) spectra were recorded by collecting sample current at the N K edge and Fe L2,3 edges at different angles between the linearly polarized light and the sample normal.[25] The XMCD spectra at Co, Fe, and Ni absorption edges have been measured in remanence by reversing the photon helicity. In the setup, an in situ electromagnet applies a pulsed magnetic field of ±0.05 T along one direction in the sample surface plane. At DEIMOS, XMCD measurements have been carried out in total electron yield mode on the Co, Fe, and Ni L2,3 absorption edges at ∼4 K with a magnetic field H of 6 T applied along the X-ray beam direction. We varied the angle between the sample normal and the X-ray beam (0, 30, and 60°).[26,27]