Thermal desorption and stability of cobalt phthalocyanine on Ag(100)

Applied Surface Science 435 (2018) 894–902

Contents lists available at ScienceDirect

Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc

Full Length Article

Thermal desorption and stability of cobalt phthalocyanine on Ag(100) ˙ Agata Sabik ∗ , Franciszek Gołek, Grazyna Antczak Institute of Experimental Physics, Faculty of Physics and Astronomy, University of Wrocław, 50-204, Poland

a r t i c l e

i n f o

Article history: Received 30 August 2017 Received in revised form 14 November 2017 Accepted 17 November 2017 Available online 21 November 2017 Keywords: Work function Thermal desorption Dissociation Phthalocyanine Silver Low energy electron diffraction

a b s t r a c t By performing work function change (WF) measurements, we characterized thermal stability and desorption of cobalt phthalocyanine (CoPc) molecules on the Ag(100) surface from sub-monolayer to multilayer coverages. Based on the temperature dependence of the WF we were able to determine the desorption temperature from multilayer. Obtained dependences of WF and a low-energy electron reflectivity (R) for sub- and monolayer reveal that layers with contact with Ag(100) have higher thermal stability and their desorption is accompanied by decomposition of CoPc molecule. Exploring the time evolutions of the WF at various temperatures allowed us to establish effective activation energies and effective frequency prefactors for processes occurring at various temperatures. The effective activation energies remain almost the same, from sub-monolayer to multilayer (2.97 eV – 2.62 eV), whereas the frequency prefactors vary from 1013 s−1 (monolayer) to 1024 s−1 (multilayer). For multilayer only desorption occurs, whereas for layers in contact with reactive Ag(100) surface (monolayer) the decomposition occurs at the same temperature range as desorption. Low-energy electron diffraction was used to describe CoPc molecular arrangements. To the best of our knowledge, we are the first who have observed (5 × 5)R ± 37◦ structure for CoPc on Ag(100). © 2017 Elsevier B.V. All rights reserved.

1. Introduction Among the variety of organic molecules, metal phthalocyanines (MPcs), i.e. macrocyclic coordinative compounds, exhibit a remarkable chemical and thermal stability simultaneously with the interesting (opto-)electronic properties that could be tuned e.g. by changing the central metal atom [1]. For these reasons, MPcs are, on the one hand, the promising candidates for molecular electronic devices, on the other hand, model systems for both experimental and theoretical investigations. In the case of cobalt phthalocyanine (CoPc), application studies were performed with respect to spintronics [2,3] or gas sensors [4]. Interestingly, MPcs form ordered layers on various metal surfaces e.g. aluminium, gold or silver. Recently, a lot of efforts have been made to understand their adsorption and thin film growth on the monocrystalline substrates [5,6]. The electronic and structural properties of CoPc-Ag interface have been investigated by various surface science techniques, such as photoelectron spectroscopy [7–11], scanning tunnelling microscopy (STM) [11–13] or low-energy electron diffraction (LEED) [9,10,13]. The CoPc adsorption on Ag(100) or Ag(111) results in non-trivial charge transfer. The charge donation

∗ Corresponding author. E-mail address: [email protected] (A. Sabik). https://doi.org/10.1016/j.apsusc.2017.11.149 0169-4332/© 2017 Elsevier B.V. All rights reserved.

to the molecule is localized and involves mainly molecular orbitals associated with the central cobalt atom [7–12]. An additional feature that occurs is charge back-donation which is associated with the rehybridization of  electrons delocalized over the molecular ligand with silver substrate [11]. Thermal desorption is a central process in understanding thermal stability of functional materials on the surfaces. Most studies and procedures concerning thermal desorption concentrated on the desorption of small species. The complex nature of molecules frequently complicates the description of the process, e.g. by the activation of various degrees of freedom during annealing. The thermal desorption of the molecular layer often competes with the dissociation of the molecules [14]. Despite the technological importance of macrocyclic organic molecules, their thermal desorption was rarely examined. Detailed temperature-programmed desorption (TPD) studies combined with comprehensive molecular dynamics analyses were carried out for 1.0 ML of linear hydrocarbons on various surfaces [15–24]. The authors found that the activation energies and frequency prefactors increase with chain length. Similar correlation between activation energy and chain length was reported for polyethers on graphite [25]. Additionally it was shown that the presence of terminal double bonds in hydrocarbons increases the desorption activation energy [18]. The correlation of the multilayer desorption activation energies and frequency prefactors with number of carbon atoms in the

A. Sabik et al. / Applied Surface Science 435 (2018) 894–902 Table 1 Parameters of thermal desorption for polyaromatic hydrocarbons [33]. E- activation energy, ␯- frequency prefactor. System

Benzene/HOPG Naphthalene/HOPG Coronene/HOPG Ovalene/HOPG

Experimental Technique TPD(0.7 K/s) TPD(1.0 K/s) TPD(2.0 K/s) TPD(2.0 K/s)

1 ML E, eV(T, K)

, s-1

0.50 ± 0.08(151) 0.8 ± 0.1(235) 1.3 ± 0.2(390) 2.2 ± 0.2(490)

1 × 1016 ± 3 5 × 1016 ± 2 2 × 1016 ± 2 5 × 1021 ± 3

organic molecule was observed also for cyclic molecules (e.g. paraquaterphenyl, para-sexiphenyl etc.) [14,26,27]. The monolayer desorption of cyclic organic molecules is frequently accompanied by dehydrogenation/cyclodimerization of the molecules [14]. For multilayer desorption of hexaazatriphenylene-hexacarbonitrile (HATCN) from gold [28] and silver [29] (111) surfaces authors do not detect dependence of desorption parameters on the type of the surface. For desorption of monolayer of rubicene, indigo and quinacridone molecules from SiO2 and carbon covered SiO2 surfaces [14,30–32] authors found that molecules decompose during the desorption from those reactive surfaces [14]. From those studies there are clear that both desorption activation energies and frequency prefactors depend on the number of carbon atoms in organic molecule and their contact with reactive surface frequently cause decomposition of the molecules. Dependence of monolayer desorption parameters on the size of the polyaromatic hydrocarbons adsorbed on unreactive HOPG is presented in Table 1 [33]. The dependence of the desorption activation energy (from 0.5 eV for benzene to 2.2 eV for ovalene) with number of carbon atoms in the molecule is clear from Table 1. The frequency prefactor also increase from 1014 s−1 for benzene to 1021 s−1 for ovalene [33]. The CoPc molecule, investigated in this paper, is composed of 32 carbon, (similar as ovalene), one cobalt, eight nitrogen and 16 hydrogen atoms. For such size of molecule, based on study reported in Ref [33] we expect prefactor, at least, in the range of 1021 s−1 and activation energy in the range of 2–3 eV. Additionally, CoPc in our studies is adsorbed on more reactive Ag(100) surface then HOPG, what makes the decomposition of the molecule very likely to happen. In Table 2 we present desorption parameters available for large organic molecules, such as 3,4,9,10-perylene-tetracarboxylicdianhydride (PTCDA) [34], MPcs [35–41] or oligopyridine (2,4 -BTP) [42]. For multilayer desorption the values of activation energy are around 2.3 eV. There is no clear dependence of activation energy on the type of surface from which molecules are desorbed. In most available studies the frequency prefactor was assumed to be in the range of 1018 s−1 . Such assumed frequency prefactor slightly influences determined value of an activation energy. The multilayer desorption proceeds at temperature range 480–600 K. Determined temperature depends on the heating rate used. The first layer, owing to the interaction with the substrate, is bonded more strongly than further overlayers of the material. For desorption of monolayer the type of surface is an important factor, since frequently, large molecules in contact with reactive surfaces undergo decomposition in the same temperature range as desorption [35,36,38]. After desorption of CuPc molecules from monolayer adsorbed on Cu(100), the surface does not come back to clean state, and some adsorbed fragments of CuPc remain on the surface [35]. Snezhkova et al.[36] reported a partial desorption of 1.0 ML of FePc from Cu(111) above 593 K and provided evidence for the fragmentation of FePc molecules during the desorption [36]. Annealing leads to partial dehydrogenation, polymerization of the benzene groups, and creation of dendrite-like chains from FePc on Cu(111) [36] and CuPc on Ag(111) [37]. Manandbar et al. [37] observed dendrite-like chains from CuPc after partial desorption of monolayer at 780 K from Ag(111). Thussing and Jakob

895

[38] reported partial dissociation of CuPc molecules during the desorption of 1.0 ML from Ag(111) surface with Cu atoms staying on the surface and the rest of molecule desorbed in segments of ¼ Pc [38]. The temperature-programmed X-ray photoemission spectroscopy (TP-XPS) was applied to investigate the influence of the central metal atom in MPc (M = Co, Cu, Fe) upon the molecular adsorption and thermal stability on Ir(111) covered with graphene (graphene/Ir(111)). In the case of FePc, the study was extended to highly oriented pyrolytic graphite (HOPG) and Au(110) substrates [39,40]. The authors showed that, on the surface of graphene/Ir(111), all MPcs exhibit layer-by-layer growth of flat lying molecules. For CuPc desorbed from graphene/Ir(111) their data are consistent with desorption of intact molecules, whereas for CoPc and FePc they observe thermal desorption above 820 K, and decomposition of the molecule above 900 K. Clearly Au(110), HOPG and graphene/Ir(111) are less reactive surfaces than Cu(100), Cu(111) or Ag(111), which is visible in tendency of decomposition of Pc molecules during desorption. The desorption activation energies of Pc monolayer from non-reactive surfaces vary from 2.6 eV for CuPc desorbed from graphene/Ir(111) [39] to >3.2 eV for Co-and FePc desorbed from Au(110) [41]. Desorption activation energies clearly depend on the type of central atom of Pc molecule. For reactive surfaces competition between different types of decomposition (dehydrogenation, dissociation, etc.) and thermal desorption is observed. So far we explored the influence of size of the molecule and possibility of decomposition on the desorption characteristics. Another factors which can influence desorption parameters are the molecule-molecule interactions and the initial molecular state (e.g. mobile/immobile, flat-laying/upright standing) [42]. Roos et al. studied the thermal desorption of 2,4 -BTP from HOPG by TPD and STM [42]. The molecules lie flat on the surface and are rotationally/translationally mobile in the sub-monolayer, but remain immobile and upright standing in the multilayer [42]. The activation energy for the desorption varies from 2.69 eV to 2.48 eV, from sub-monolayer to multilayer, respectively. The frequency prefactors varies from 1015 s−1 for sub-monolayers to 1024 s−1 for multilayers. According to authors, such prefactors result from different initial molecular states for desorption [42]. Recently, we have shown that work function change (WF) measurements by means of the retarding field diode method, known also as Anderson method, are suitable for studying MPcmetal interfaces and provide insight into the low-energy electron reflectivity (R) of the sample [43,44]. In this work, we have concentrated on the study of thermal stability and the desorption of CoPcs on the Ag(100) surface from the sub-monolayer to the multilayer regime. LEED investigations were carried out to probe structural properties of CoPc layer before and after the thermal treatment of the sample.

2. Experiment The experiments were carried out in an ultra–high vacuum system with a base pressure in the range 1 × 10−10 mbar. We probed the WF and R, with accuracy ±0.01 eV and ± 1%, respectively, implementing the retarding field diode method with the e-gun setting described in [43]. In the e-gun all electrodes have been at positive potentials with respect to the ground. The electrons can escape from the e-gun only when the Ag(100) sample is in front of e-gun (around 2 cm) at high enough positive potential. The beam diameter used is around 1.5 mm. The total current needed to determine reflectivity is measured with voltage of +180 V applied to the sample. For such voltage it was assumed that all electrons from the e-gun reach the sample. Further increase of voltage does not change the current detected from the sample. For WF and R the shift

896

A. Sabik et al. / Applied Surface Science 435 (2018) 894–902

Table 2 Parameters of thermal desorption for large organic molecules. E- activation energy, ␯- frequency prefactor. System

CoPc/graphene/ Ir(111) CuPc/graphene/ Ir(111) FePc/graphene/ Ir(111) CoPc/Au(110) CuPc/Au(110) FePc/Au(110) FePc/Au(110) FePc/HOPG CuPc/Ag(111) CuPc/Cu(100) PTCDA/Cu(111) 2,4’-BTP/HOPG CoPc/Ag(100)

Experimental Technique(dT/dt; m) XP-TPD(0.3 K/s; N 1s) XP-TPD(0.3 K/s; N 1s) XP-TPD(0.3 K/s; N 1s) TPD(0.3 K/s; 128 u) TPD(0.3 K/s; 128 u) TPD(0.3 K/s; 128 u) TPD(0.3 K/s) TPD(0.3 K/s) TPD(1 K/s) TPD(5 K/s) TPD(1 K/s) TPD(1 K/s) WF

1 ML

>1 ML

Ref.

E, eV(T, K)

, s−1

E, eV(T, K)

, s−1

3.2 ± 0.1(820 ± 15) 2.6 ± 0.1(670 ± 15) 3.2 ± 0.1(820 ± 15) >3.20 ± 0.05(>820 ± 15) 2.74 ± 0.05(700 ± 15) >3.20 ± 0.05(>820 ± 15) 3.2 ± 0.1(823 ± 15) 2.8 ± 0.1(723 ± 15) Dissociation(600-700) Dissociation of 0.5 ML Not desorb until 900 K 2.69(∼700) Effective values 2.97 ± 0.27 (780–860) 2.62 ± 0.29

ass. 1018 ass. 1018 ass. 1018 ass. 1018 ass. 1018 ass. 1018 ass. 1018 ass. 1018

2.3 ± 0.1(590 ± 15) 2.2 ± 0.1(570 ± 15) 2.4 ± 0.1(600 ± 15) 2.27 ± 0.05(576 ± 15) 2.27 ± 0.05(576 ± 15) 2.27 ± 0.05(576 ± 15) 2.3 ± 0.1(578 ± 15) 2.3 ± 0.1(583 ± 15) 2.20 ± 0.05(480-530) 1.73 2.35 ± 0.20 2.48(∼500)

ass. 1018 ass. 1018 ass. 1018 ass. 1018 ass. 1018 ass. 1018 ass. 1018 ass. 1018 1.5 × 1020 (2.6 and 5 ML) – 4 × 1019 1024

[40]

2.70 ± 0.35 (475 – 550)

1024.3 ± 3.5

This work

1015 (<1ML) 1015.6 ± 1.6 (<1 ML) 1013.7 ± 1.7

[41]

[39] [38] [35] [34] [42]

(after normalization) and maximum current from I–V curves during adsorption is measured, respectively. The influence of electron beam with energy up to 2 eV on the molecular layer was tested, and the electron stimulated destruction of molecules was not detected [44]. The Ag(100) sample was mounted on the manipulator and heated resistivily. The crystal temperature was measured with a c-type thermocouple (tungsten 5% rhenium vs tungsten 26% rhenium), with an accuracy ±4.0 K, spot welded to a tantalum disk on which the Ag sample was mounted. The apparatus was equipped with an Auger electron spectrometer (AES) and a LEED optics from OCI Vacuum Microengineering Inc.. The Ag(100) sample was cleaned by subsequent cycles of argon ions sputtering with energy of 1.2 keV and ions current of 2.5 ␮A for 40 min and annealing at 910 K for six minutes. The cleanliness of the crystal was checked by AES and LEED. The CoPc molecules (Aldrich, 97%) were deposited onto the Ag(100), held at room temperature (RT), from home-made Knudsen cell kept at 700 K. The deposition rate was 0.1 ML/min. Before the experiments the molecular source had been carefully outgassed at 600 K for 24 h. The molecular coverage was determined from the WF induced by the adsorption of CoPc combined with LEED investigations [44]. The WF is determined with respect to the work function (WF) of the clean Ag(100) surface, otherwise stated. The monolayer was assumed to be at the minimum of WF, as described in 44 and consisted of flat-lying CoPc molecules fully covering the (100) silver surface in an ordered 5 × 5 arrangements. The LEED patterns presented were recorded at 80 K (apart from the LEED pattern for 2D gas, which was recorded at RT) and were the same (somewhat sharper) as the LEED patterns obtained at RT.

3. Results and discussion 3.1. Temperature evolutions of work function and low-energy electron reflectivity To determine the thermal stability of multi-, mono-, and sublayer we studied the temperature evolutions of WF and R for 0.3 ML, 1.0 ML, 2.8 ML and 4.0 ML of CoPc deposited onto Ag(100). The dependences shown in Fig. 1 are typical for the coverages under study. The experiment proceeded as follows: the sample was annealed at a fixed temperature for one minute, then quenched to RT for the I–V curve measurements, and subsequently subjected to a next higher annealing temperature. Fig. 1a shows the temperature evolutions of the WF, whereas Fig. 1b the temperature depen-

Fig. 1. CoPc on Ag(100) temperature evolutions of: (a) WF for initial coverages of 0.3 ML, 1.0 ML, 2.8 ML and 4.0 ML; (b) R for initial coverages: 0.3 ML, 1.0 ML, 2.8 ML. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

dences of R. The changes are discussed together with the molecular arrangements found through LEED investigations. The deposition of 0.3 ML of CoPc at RT leads to a decrease in both the WF by 0.15 eV and R by 10%. The R for clean Ag(100) is around 55% [44]. For this coverage the LEED pattern looks similar to that obtained for the clean Ag(100) surface. For higher submonolayer coverages, around 0.6 ML, in addition to the reflexes which come from Ag(100), a ring-like LEED pattern has been

A. Sabik et al. / Applied Surface Science 435 (2018) 894–902

Fig. 2. The ring-like superstructure associated with the presence of 2D molecular gas phase visible in LEED pattern for 0.6 ML of CoPc on Ag(100).

recorded and is presented in Fig. 2. Such LEED feature is associated with an isotropic distribution of scatters on the surface, similar like in 2D liquid phase [45,46]. It was shown by the STM studies that in sub-monolayer regime above 80 K, the Pc molecules move faster than STM scanning tip, creating a diffusive noise [51]. Additionally, there is no nucleation of molecular islands [51]. The radius of the diffusive ring is associated with certain moleculemolecule distance typical for given coverage. Similar LEED images for MPcs in sub-monolayer coverages on silver surfaces were found e.g. for FePc on the Ag(100), H2 Pc, FePc and CuPc on Ag(111) [47–50]. Annealing 0.3 ML of CoPcs to 700 K leads to a decrease in R from about 45% to 35%, with WF remaining constant (green circles in Fig. 1a and b). Between 700 K and 750 K the changes are observed in both WF and R. WF increases and WF reaches value 0.11 eV and R goes down to 31%. Such behaviour of R may be connected with the creation of CoPc-Ag islands in the temperature range between RT and 750 K or creation of chains due to dehydrogenation of CoPc molecule. The CuPc-Ag networks at RT were reported for CuPc on Ag(100) system [52]. The silver adatoms (which come from fluctuating steps of Ag(100)) assemble with CoPc molecules. The dehydroganated Pcs were reported to form covalently bounded organic chains [36,37]. Cirera et al [53] observed creation of Pc tapes from 2,3,7,8,12,13,17,18-octaethyl-5,10,15,20tetraaza-porphirin (OETAP) already at around 500 K. Limiting step of this surface reaction is dehydrogenation of the molecule. Both networks and dehydrogenation may lower the R of the sample and areas influenced by them are too small to be detect in LEED patterns. In case of network, a higher temperature generates a higher number of adatoms available for stabilization of the molecules, which means that the number or size of CoPc-Ag islands increases. In case of dehydrogenation, molecules are kept together by covalent C C bounding. As a consequence, in both cases, fewer molecules are mobile and so the adsorbate-free surface is more exposed during I–V curve acquisition. The growth of CoPc-Ag or dehydrogenated islands may also cause slight increase in the WF, visible between 700 K–750 K, in Fig. 1a. In the temperature range from 750 K to 800 K, R remains constant, with WF increasing. In this temperature region, we observe overlap of different processes such as: the creation of CoPc-Ag networks, dehydrogenation of CoPc molecules and creation of covalent molecular bounding, the desorption and the dissociation of CoPc. Between 800 K and 830 K the recurrence of both WF and R is observed. Around 830 K, WF saturates and reaches levels slightly higher than those for the clean Ag(100) surface. At the same time, R reaches only 37%, i.e. 18% lower than that for clean Ag(100) surface. The increase of both WF and R suggests the removing the CoPc molecules from Ag(100) and thereby surface comes

897

back to state before adsorption. However, lower final value of R (in comparison to value for clean Ag(100)) is an indication that the thermal treatment of the sub-monolayer of CoPc on Ag(100) leads to the decomposition of the molecules. Since R is probing the state of the surface, that is an evidence that some molecular parts remain on the surface. For coverages close to 1.0 ML, after the adsorption of CoPcs, we observed by LEED two ordered commensurate structures: (5 × 5)R0 and (5 × 5)R ± 37◦ , as shown in Fig. 3a and b. The structures coexist on the surface. Using LEED we observe (5 × 5)R0 in one place, then changing the studied sample place, by changing coordinates using manipulator, we observe the (5 × 5)R ± 37◦ structure. We also are able to observe overlap of both structures, see Fig. 3e. The areas covered by single type of structure is quite big, since we are allowed to register LEEDs of (5 × 5)R0 and (5 × 5)R ± 37◦ separately. However, searching various places on the sample we more frequently observed (5 × 5)R0 than (5 × 5)R ± 37◦ structure. In Fig. 3c and d, we present the models of each superstructure. The molecular lateral arrangement used for models was based on previous studies by density functional theory and STM of this system [12,13,54]. The CoPcs are assumed to be adsorbed with cobalt atoms at the hollow sites with ∼30◦ rotation of molecular axis, with respect to the silver [110] direction (see Fig. 3c and d). The structure (5 × 5)R0 is commonly observed for this system by LEED and STM [9,10,13,44]. To the best of our knowledge, we are the first who have observed (5 × 5)R ± 37◦ structure for CoPc on Ag(100). This arrangement was reported for CuPc on Ag(100) by LEED and STM [55]. This phase consists of two domains, the model for one of them is shown in Fig. 3d, whereas the other one is its mirror reflection in relation to the [110] silver direction. Fig. 1a and b (blue triangles) show the temperature evolutions of WF and R occurring for 1.0 ML during thermal treatment. After adsorption at RT, the values of WF and R are 0.31 eV and 19%, respectively. Annealing the sample results in an emergence of additional LEED pattern, shown in Fig. 4a. We believe that this LEED pattern is associated with the appearance of a new incommensurate phase with 5 × 5 superstructure and around ±7◦ rotation of the unit cell in relation to the [110] silver lattice direction. The corresponding LEED scheme for (5 × 5)R ± 7◦ structure is shown in Fig. 4b. This structure after thermal treatment coexists on the sample with structure (5 × 5) R0 and (5 × 5)R ± 37◦ . In Fig. 4c we present overlap of LEED patterns from (5 × 5)R ± 7◦ and (5 × 5)R ± 37◦ . For 1.0 ML, both WF and R are constant up to 680 K, indicating no changes in the state of the sample. From about 680 K it is visible an increase of both WF and R. It is worth to emphasise that for sub- and monolayer the increase in WF are observed in similar temperature range, i.e. above 680 K. This temperature is comparable with values of desorption temperatures available in the literature, for similar systems. Using TPD, for desorption of 1.0 ML of CuPc from Ag(111), the value of 600 K – 700 K was estimated [38]. For desorption of 1.0 ML of CoPc from graphene/Ir(111) the value of 820 ± 15 K was determined [40]. For desorption of CuPc from Au(110) surface, the TPD desorption peak at 700 K was observed, whereas for FePc and CoPc no desorption peaks up to 820 K were detected. The study are illustrating importance of the central atom of Pc for thermal desorption [41]. We associate changes in WF and R above 680 K with a desorption of 1.0 ML of CoPc from Ag(100). After the last annealing (at 910 K), WF was 0.07 eV higher, and R was 16% lower than for clean Ag(100), which is connected with the decomposition of CoPc during desorption. Similar as was mentioned for 0.3 ML CoPc some molecule parts stay on the surface, influencing WF and R. The decomposition of CuPc during desorption from Ag(111) [38], as well as for FePc from Cu(111) [36], was found from TPD and XPS studies. Authors reported that during desorption of CuPc molecules, which are in contact with substrate, additionally to intact molecules they desorb parts of one-fourth of

898

A. Sabik et al. / Applied Surface Science 435 (2018) 894–902

Fig. 3. LEED patterns recorded after the deposition of 1.0 ML of CoPc on Ag(100): (a) (5 × 5)R0 structure; (b) (5 × 5)R ± 37◦ structure. The proposed corresponding model of unit cell of (c) (5 × 5)R0 structure; (d) (5 × 5)R-37◦ , the clockwise rotation of unit cell is shown (e) coexistence of (5 × 5)R0 and (5 × 5)R ± 37◦ structure.

CuPc molecule [38]. It is likely that similar situation happened in our case. For multilayer coverages we have studied (from above 1.0 ML to 4.0 ML) R is about 21% and does not depend on molecular coverage, whereas WF increases for adsorption of coverages between 1.0 ML and 3.0 ML, see [43]. We have not observed LEED patterns which would allow us to describe the structure of the multilayer. Fig. 1a and b (red squares) illustrate temperature evolutions of WF and R for 2.8 ML. After adsorption at RT the R is 21% and WF is 0.11 eV. During thermal treatment, R remains constant for temperatures up to 680 K. Situation differs for WF. Here, from 475 K to 550 K the WF decreases and finally reaches the level characteristic of 1.0 ML of CoPcs, i.e. WF reaches value around 0.30 eV. Furthermore, the LEED patterns recorded after annealing to 550 K are the same as shown for 1.0 ML after thermal treatment, i.e. (5 × 5)R0, (5 × 5)R ± 37◦ and (5 × 5)R ± 7◦ . It is consistent with observations reported in [10] where the authors after the desorption of CoPc multilayers from Ag(100), observed rotated 5 × 5 superstructures [10]. One can conclude that the decrease in WF from 475 K to 550 K is associated with the desorption of multilayer. Further changes in WF and R, after multilayer desorption, are the same as previously discussed for 1.0 ML. What means that after the desorption of a mul-

tilayer, the CoPc monolayer stays on the surface. The temperature evolution of WF (up to 650 K) for 4.0 ML is indicated by orange polygons in Fig. 1a. Note that during adsorption at RT, the WF saturates after the deposition of 3.0 ML [44], that is why the coverage of 4.0 ML was estimated based on the deposition time, not by changes in WF. Temperature evolution of WF for 4.0 ML looks similar to one for 2.8 ML. The initial WF is 0.05 eV and we observe slight increase in WF (from 0.05 to 0.01 eV) due to reordering between RT and 475 K. Then, we observed a significant decrease in WF in the temperature range of 475 K to 550 K, beyond which WF reaches levels corresponding to 1.0 ML of CoPc on Ag(100). It appears that for coverages between 1.0 ML and 4.0 ML, the multilayers desorption occurs in comparable temperature range. For 4.0 ML the changes in WF and R, for temperatures higher than 680 K, look similar as for 2.8 ML and for 1.0 ML (not shown in Fig. 1). The desorption temperature range we determined for the multilayer is in agreement with the data reported for CuPc on Ag(111) by TPD, with the heating rate 1 K/s [38]. The desorption of 2.0 ML occurs at 500 K −600 K, whereas the desorption of layers higher than 2.0 ML at 480 K−530 K [38]. In the case of the desorption of CoPc thin films (5 nm thick) from graphene/Ir(111) the desorption temperature was established as 590 ± 15 K by TP-XPS, with the heating rate 0.3 K/s [40]. For des-

A. Sabik et al. / Applied Surface Science 435 (2018) 894–902

899

Fig. 4. LEED pattern (a) and corresponding scheme (b) for incommensurate (5 × 5)R ± 7◦ structure obtained after the thermal treatment of 1.0 ML of CoPc on Ag(100) (c) coexistence of (5 × 5)R ± 7◦ and (5 × 5)R ± 37◦ structure.

orption from Au(110) the value of (576 ± 15) K was determined for FePc, CuPc and CoPc with heating rate 0.3 K/s [41].

3.2. Time evolutions of work function changes To get an insight into the energetics of the processes occurring on the surface when the most rapid changes in WF and R are observed, we performed experiments at a fixed temperature for three different initial adsorbate states: 0.3 ML (2D molecular gas phase), 1.0 ML (5 × 5 ordered structure) and 2.0 ML (multilayer). The temperature evolutions described in the Section A allowed us to find temperature range where the multilayer desorption is a dominant process (i. e. from 475 K to 550 K). For sub- and monolayer we expect that decomposition of CoPc and desorption occur above 680 K. For determination of time evolutions of WF and R, after the deposition of a particular CoPc coverage, the sample was annealed at a constant temperature between 485 K and 535 K for 2.0 ML and 750 K−850 K for 0.3 ML and 1.0 ML. The annealing was interrupted and the sample was quenched to RT for the WF and the R measurements. In these experiments we concentrated on the time evolution of the apparent work function change (AWF) [43,44]. AWF is determined from the shift of I-V curve (without normalization to the saturation values as described in 44 in our case i.e. at 10% of height of the I–V curve for clean Ag(100). In Section A, we have shown that R remains constant during the desorption of multilayers. The situation is different for 0.3 ML and 1.0 ML, where both WF and R change during subsequent steps of annealing above 680 K. However, the WF and R are comparable for every thermally-induced stage at every annealing temperature. That means that AWF allows us to probe changes in both WF and R using one parameter. Additionally, the initial coverage, as well as the thermally-induced stages of the sample are more precisely and repeatedly established by AFW. The overlap of WF and R causes better separation of thermally-induced stages [43,44]. AWF after

adsorption of 0.3 ML and 1.0 ML is 0.22 eV and 0.47 eV respectively. For 2.0 ML AWF is 0.11 eV (determined with respect to 1.0 ML). After constant temperature treatment of 0.3 ML at temperature above 680 K, R saturates at values 49%–51% i.e. around 5% below the level characteristic of the clean Ag(100) surface, see dashed line in Fig. 5a. For 1.0 ML the R saturates at 40%–45% i.e. around 12% below the level characteristic of the clean Ag(100) surface, see dashed line in Fig. 5b. This is an indication of a decomposition of CoPc molecules during the desorption. The effect is more pronounced during the thermal treatment of 1.0 ML because there are more molecules on the surface, and more parts of the molecules stay on the surface, reducing the R of the sample. The thermally-induced stages chosen for our investigations are marked by horizontal solid lines in Fig. 5. As it can be easily seen, lower annealing temperatures require longer time to transform between two thermally-induced stages. Such behaviour is typical for thermally activated processes. In case of multilayer the thermally-induced stages are correlated with coverage of CoPc remaining on the surface, since we expect desorption of intact molecules. In case of the sub- and monolayers, the thermally-induced stages cannot be directly associated with coverage. Happened it because decomposition of molecules occurring during desorption. Since WF and R are the same for every thermallyinduced stage, we expect that the comparable amount of intact and broken parts of molecules are present on the surface. Based on time needed for transition between stages, we constructed the Arrhenius plots (Fig. 6). Due to participation of two processes (desorption and decomposition) during annealing of the sample above 680 K, we only can determine effective activation energies and frequency prefactors from Arrhenius plots, responsible for transition between stages. The error bars on y-axis in Fig. 6 are estimated from the time intervals needed for the transition between the stages, assuming the accuracy of AWF as ±0.002 eV. The values obtained from Arrhenius plot for multilayer desorption are 2.70 ± 0.35 eV and 1024.3.±3.5 s−1 . The activation energy and

900

A. Sabik et al. / Applied Surface Science 435 (2018) 894–902

Fig. 6. Arrhenius plots for 0.3 ML, 1.0 ML and 2.0 ML of CoPc on Ag(100) obtained from time intervals necessary for the transition between the two thermally-induced stages shown in Fig. 5.

Fig. 5. Time evolutions of AWF during the annealing of CoPc layers on Ag(100) at fixed temperatures for initial coverages: (a) 0.3 ML; (b) 1.0 ML; (c) 2.0 ML. The selected thermally-induced stages are marked by horizontal solid lines. Dotted/dashed horizontal lines indicate the R for initial molecular states; whereas the dashed horizontal lines the R for the final states after the thermal treatment.

frequency prefactor in this range are expected for thermal desorption of molecule of CoPc size. The increase in desorption activation energy and frequency prefactor with number of carbon atoms in the molecule was reported previously, see Table 1. For thermal desorption of ovalene (the molecule with the same number of carbon atoms as CoPc) from inert HOPG surface the activation energy is 2.2 eV and prefactor is 1021 s−1 [33]. CoPc additionally to 32 carbon atoms, consists of one Co atom and eight nitrogen atoms, which likely play role in desorption process as well. We believe that for multilayer desorption we cope with thermal desorption of intact molecules. Additionally that conclusion is supported by observation of LEED patterns of structures typical for 1 ML after multilayer desorption. Now we compare our value with available in the literature data. Unfortunately, the desorption parameters for MPcs are scantily reported in the literature and frequently frequency prefactors are assumed. Our multilayer desorption activation energy

and prefactor are within a comparable range to the values reported for a multilayer of CuPc from Ag(111), which are 2.20 ± 0.05 eV and 1.5 × 1020 s−1 [38]. The activation energies reported for the desorption of multilayer of CoPc and CuPc from graphene/Ir(111) were 2.3 ± 0.1 eV and 2.2 ± 0.1 eV, respectively [40]. The activation energy for the desorption of 2.0 ML of CuPc from Cu(100) was established at 1.73 eV [35]. Struzzi et al. reported the desorption of a multilayer of FePc from HOPG, graphene/Ir(111) and Au(110) [39] at 2.3 – 2.4 eV. For desorption of multilayer of CoPc, FePc and CuPc from Au(110) the activation energy of 2.27 ± 0.05 eV was reported [41]. For investigations of desorption from Au(110), HOPG, graphene/Ir(111) authors assumed prefactor in the range 1018 s−1 . That might be responsible for differences in determined activation energy, since modification by two orders in magnitude of the prefactor causes around 10% change in activation energy estimated using the Redhead equation [40]. Our prefactor is determined directly from Arrhenius dependence, is around five orders of magnitude higher than assumed 1018 s−1 , and is in accord with the values previously reported for the multilayer desorption of large organic molecules [14,26–32,42]. The values obtained from Arrhenius plots after annealing of 0.3 ML above 680 K are 2.97 ± 0.27 eV and 1015.6 ± 1.6 S −1 , whereas after annealing 1.0 ML are 2.62 ± 0.29 eV and 1013.7 ± 1.7 s−1 . The values for both initial coverages are the same, suggesting the same process responsible for changes. Saturation of R below value characteristic for clean surface inform us that decomposition of CoPc is present at this temperature range. Linearity of Arrhenius plot suggests that there is a dominant process at investigated temperature range or processes have comparable activation energies and prefactors. Additionally the frequency prefactors determined in our study are within a range 1013 s−1 . For thermal desorption, the increase in both: activation energy and frequency prefactor with size of molecule (number of molecule carbon atoms) desorbed was reported previously for polyaromatic hydrocarbons, see Table 1 [33]. We expect then that prefactor in the range of 1021 s-1 should be typical for molecule of this size. Observed by us prefactor is not typical for thermal desorption of large molecule. At the same time the prefactor in range 1013 s-1 together with linear Arrhenius plot suggests that the decomposition of molecule (not desorption) is a rate limiting process for transition between thermally-induced stages. That means that activation energies obtained from Arrhenius plot are likely associated with decomposition not desorption of the molecules. The activation energy for thermal desorption of

A. Sabik et al. / Applied Surface Science 435 (2018) 894–902

intact CoPc in contact with metal substrate, likely would be higher due stronger molecule-metal bounding in the system. Looking at the available literature, we see that values obtained for monolayer desorption from unreactive surfaces, where decomposition is not present (HOPG, Au(110), graphene/Ir(111)), are around 3.2 eV (higher than value observed by us) and are due to thermal desorption of intact molecules [40], see Table 2. The reactive surfaces likely cause weakening of molecule inner bounds, what results in decomposition of molecule. The relationship between multilayer and (sub-) monolayer desorption parameters was reported for the desorption of 2,4 -BTP from HOPG [42]. The activation energies quoted in the literature, 2.48 eV for multilayer and 2.69 eV for sub-monolayer, the frequency prefactors: between 1015 s−1 for sub-monolayer and 1024 s−1 for multilayer. The authors attributed such a trend to difference in initial molecular states for desorption, i.e. translationally/rotationally mobile, flat-lying oligopyridine at sub-monolayer and upright standing immobile molecule at multilayer [42]. In the case of CoPc on Ag(100), we know that at sub-monolayer regime the molecules lie flat and move in a complex rotational/translational fashion [54]. For 1.0 ML, the molecules lie flat and form well-ordered structures described in the previous section. Notwithstanding scanty information about the arrangement of CoPc molecules above 1.0 ML, the STM studies of MPc multilayers revealed that starting from the second layer the molecules may be upstanding or tilted in relation to the surface [56]. An indication of the tilted adsorption configurations in multilayer regimes were reported for CoPc on Au(111) [57], Si(111)-(1 × 1)Pb [58] and Bi(111) [59] substrates. The report of Roos et al. [42] suggested that the various initial states influences the prefactor determined from Arrhenius plot [42], what could be also the case in our study. However, we have evidence of decomposition of CoPc molecules at temperatures above 680 K. We suspect that decomposition of molecule is influencing the prefactor more significantly than entropy change caused by different initial states for desorption. Moreover, it was shown that on reactive surfaces, such as Cu(100), Cu(111) and Ag(111) there is fragmentation (decomposition) of Pc molecules during thermal treatment [35–38]. On less reactive surface, such as graphen/Ir(111), the decomposition of CoPc and FePc was observed at higher temperature than desorption, whereas was not observed for CuPc, proving non-negligible role of the metal atom assembled in cavity of Pc [40]. The Ag(100) surface is an example of the reactive surface for which the decomposition plays an important role. Unfortunately, there is no values available for comparison for desorption of CoPc from more reactive surfaces, however clearly reactivity of the surface play the role during thermal treatment above 680K.

4. Conclusion We have determined energetic parameters of processes occurring during thermal treatment of three initial coverages of CoPc on Ag(100): sub-monolayer, monolayer and multilayer. The desorption temperature of multilayer was measured at around 500 K. The thermal stability of the CoPc molecules that remain in direct contact with the substrate (sub- and monolayer) is higher than of molecules in multilayers. For monolayer we observed the changes in surface state above 680 K. Additionally, the elevated WF and diminished R (in relation to clean Ag(100)) after the sample constant temperature treatment above 680 K, indicate the decomposition of CoPc molecule. We established effective activation energies from each initial molecular phase at 2.70 eV (multilayer), 2.62 eV (monolayer) and at 2.97 eV (sub-monolayer). The activation energies are within a comparable range and in the case of multilayer it is associated with simple thermal desorption. The frequency prefactors are in the 1013 s−1 range for (sub-) monolayer and around eleven order

901

of magnitude higher (1024 s−1 ) for multilayer. We believe that the decomposition of CoPc molecule during desorption is responsible for such huge difference in the frequency prefactors. It is likely that for layers in direct contact with the substrate, the decomposition and desorption are interrelated and desorption would not happened without decomposition at investigated temperature range. The deposition of 1.0 ML of CoPc onto Ag(100) at RT results in the formation of two commensurate structures (5 × 5)R0 and (5 × 5)R ± 37◦ , the latter not previously reported for this system. The thermal treatment leads to the emergence of an additional molecular arrangement: (5 × 5)R ± 7◦ . After the desorption of multilayer, we identified (5 × 5) rotated structures, which are characteristic of 1.0 ML of CoPc.

Acknowledgments We would like to thank Zbigniew Juszczyk and Piotr Wieczorek for their technical assistance. This work was cofinanced by National Science Centre, Poland project number 2015/19/N/ST3/01044 and research grant number 0420/2015/16.

References [1] G. de la Torre, Ch.G. Claessens, T. Torres, Phthalocyanines: old dyes, new materials. putting color in nanotechnology, Chem. Commun. 20 (2007) 2000–2015. [2] C. Iacovita, M.V. Rastei, B.W. Heinrich, T. Brumme, J. Kortus, L. Limot, J.P. Bucher, Visualizing the spin of individual cobalt-phthalocyanine molecules, Phys. Rev. Lett. 101 (2008), 116602. [3] J. Bartolomé, C. Monton, I.K. Schuller, Molecular Magnets, Springer, Berlin Heidelberg, 2014, pp. 221–245. [4] D.D. Klyamer, A.S. Sukhikh, P.O. Krasnov, S.A. Gromilov, N.B. Morozova, T.V. Basova, Thin films of tetrafluorosubstituted cobalt phthalocyanine: structure and sensor properties, Appl. Surf. Sci. 372 (2016) 79–86. [5] W. Auwärter, D. Écija, F. Klappenberger, J.V. Barth, Porphyrins at interfaces, Nat. Chem. 7 (2015) 105–120. [6] J.M. Gottfried, Surface chemistry of porphyrins and phthalocyanines, Surf. Sci. Rep. 70 (2015) 259–379. [7] F. Petraki, H. Peisert, F. Latteyer, U. Aygül, A. Vollmer, T. Chassé, Impact of the 3d electronic states of cobalt and manganese phthalocyanines on the electronic structure at the interface to Ag(111), J. Phys. Chem. C 115 (2011) 21334–21340. [8] M. Schmid, A. Kaftan, H.-P. Steinrück, J.M. Gottfried, The electronic structure of cobalt(II) phthalocyanine adsorbed on Ag, Surf. Sci. 606 (111) (2012) 945–949. [9] E. Salomon, P. Amsalem, N. Marom, M. Vondracek, L. Kronik, N. Koch, T. Angot, Electronic structure of CoPc adsorbed on Ag(100): evidence for molecule-substrate interaction mediated by Co 3d orbitals, Phys. Rev. B 87 (2013), 075407. [10] B.W. Caplins, A.J. Shearer, D.E. Suich, E.A. Muller, Ch.B. Harris, Measuring the electronic corrugation at the metal/organic interface, Phys. Rev. B 89 (2014), 155422. [11] M. Toader, P. Shukrynau, M. Knupfer, D.R.T. Zahn, M. Hietschold, Site-dependent donation/Backdonation charge transfer at the CoPc/Ag(111) interface, Langmuir 28 (2012) 13325–13330. [12] A. Mugarza, R. Robles, C. Krull, R. Korytár, N. Lorente, P. Gambardella, Electronic and magnetic properties of molecule-metal interfaces: transition-metal phthalocyanines adsorbed on Ag(100), Phys. Rev. B 85 (2012), 155437. [13] E. Salomon, D. Beato-Medina, A. Verdini, A. Cossaro, D. Cvetko, G. Kladnik, L. Floreano, T. Angot, Correlation between charge transfer and adsorption site in CoPc overlayers adsorbed on Ag(100), J. Phys. Chem. C 119 (2015) 23422–23429. [14] A. Winkler, Initial stages of organic film growth characterized by thermal desorption spectroscopy, Surf. Sci. 643 (2016) 124–137. [15] K.R. Paserba, A.J. Gellman, Kinetics and energetics of oligomer desorption from surfaces, Phys. Rev. Lett. 86 (2001) 4338. [16] K.R. Paserba, A.J. Gellman, Effects of conformational isomerism on the desorption kinetics of n-alkanes from graphite, J. Chem. Phys. 115 (2001) 6737–6751. [17] J.L. Brand, M.V. Arena, A.A. Deckert, S.M. George, Surface diffusion of n-alkanes on Ru(001), J. Chem. Phys. 92 (1990) 5136–5143. [18] A.V. Teplyakov, A.B. Gurevich, M.X. Yang, B.E. Bent, J.G. Chen, NEXAFS and TPD studies of molecular adsorption of hydrocarbons on Cu(100): segmental correlations with the heats of adsorption, Surf. Sci. 396 (1998) 340–348. [19] S.L. Tait, Z. Dohnálek, Ch.T. Campbell, B.D. Kay, n-alkanes on MgO(100). I. coverage-dependent desorption kinetics of n-butane, J. Chem. Phys. 122 (2005), 164707.

902

A. Sabik et al. / Applied Surface Science 435 (2018) 894–902

[20] S.L. Tait, Z. Dohnálek, Ch.T. Campbell, B.D. Kay, n-alkanes on MgO(100). II. Chain length dependence of kinetic desorption parameters for small n-alkanes, J. Chem. Phys. 122 (2005), 164708. [21] R.Z. Lei, A.J. Gellman, B.E. Koel, Desorption energies of linear and cyclic alkanes on surfaces: anomalous scaling with length, Surf. Sci. 554 (2004) 125–140. [22] K.A. Fichthorn, R.A. Miron, Thermal desorption of large molecules from solid surfaces, Phys. Rev. Lett. 89 (2002), 196103. [23] K.A. Fichthorn, K.E. Becker, R.A. Miron, Molecular simulation of temperature-programmed desorption, Catal. Today 123 (2007) 71–76. [24] K.E. Becker, K.A. Fichthorn, Accelerated molecular dynamics simulation of the thermal desorption of n-alkanes from the basal plane of graphite, J. Chem. Phys. 125 (2006), 184706. [25] K.R. Paserba, N. Vaidyanathan, A.J. Gellman, Conformational entropy effects on the desorption kinetics of polyethers from graphite, Langmuir 18 (2002) 9799–9809. [26] S. Müllegger, I. Salzmann, R. Resel, G. Hlawacek, C. Teichert, A. Winkler, Growth kinetics, structure, and morphology of para-quaterphenyl thin films on gold, J. Chem. Phys. 121 (111) (2004) 2272–2277. [27] S. Müllegger, A. Winkler, Hexaphenyl thin films on clean and carbon covered Au(1 1 1) studied with TDS and LEED, Surf. Sci. 600 (2006) 1290–1299. [28] P. Frank, N. Koch, M. Koini, R. Rieger, K. Müllen, R. Resel, A. Winkler, Layer growth and desorption kinetics of a discoid molecular acceptor on Au(1 1 1), Chem. Phys. Lett. 473 (2009) 321–325. [29] P. Frank, T. Djuric, M. Koini, I. Salzmann, R. Rieger, K. Müllen, R. Resel, N. Koch, A. Winkler, Layer Growth, Thermal stability, and desorption behavior of hexaaza-triphenylene-hexacarbonitrile on Ag(111), J. Phys. Chem. C 114 (2010) 6650–6657. [30] B. Scherwitzl, W. Lukesch, A. Hirzer, J. Albering, G. Leising, R. Resel, A. Winkler, Initial steps of rubicene film growth on silicon dioxide, J. Phys. Chem. C 117 (2013) 4115–4123. [31] B. Scherwitzl, R. Resel, A. Winkler, Film growth, adsorption and desorption kinetics of indigo on SiO2 , J. Chem. Phys. 140 (2014), 184705. [32] B. Scherwitzl, R. Lassnig, M. Truger, R. Resel, G. Leising, A. Winkler, Adsorption, desorption, and film formation of quinacridone and its thermal cracking product indigo on clean and carbon-covered silicon dioxide surfaces, J. Chem. Phys. 145 (2016), 094702. [33] R. Zacharia, H. Ulbricht, T. Hertel, Interlayer cohesive energy of graphite from thermal desorption of polyaromatic hydrocarbons, Phys. Rev. B 69 (2004), 155406. [34] Th. Wagner, H. Karacuban, R. Möller, Analysis of complex thermal desorption spectra: PTCDA on copper, Surf. Sci. 603 (2009) 482–490. [35] T.J. Schuerlein, N.R. Armstrong, Formation and characterization of epitaxial phthalocyanine and perylene monolayers and bilayers on Cu(100): low-energy electron diffraction and thermal desorption mass spectrometry studies, J. Vac. Sci. Technol. A 12 (1994) 1992–1997. [36] O. Snezhkova, F. Bischoff, Y. He, A. Wiengarten, S. Chaudhary, N. Johansson, K. Schulte, J. Knudsen, J.V. Barth, K. Seufert, W. Auwärter, J. Schnadt, Iron phthalocyanine on Cu(111): coverage-dependent assembly and symmetry breaking, temperature-induced homocoupling, and modification of the adsorbate-surface interaction by annealing, J. Chem. Phys. 144 (2016), 094702. [37] K. Manandhar, T. Ellis, K.T. Park, T. Cai, Z. Song, J. Hrbek, A scanning tunnelling microscopy study on the effect of post-deposition annealing of copper phthalocyanine thin films, Surf. Sci. 601 (2007) 3623–3631. [38] S. Thussing, P. Jakob, Structural and vibrational properties of CuPc/Ag(111) ultrathin films, J. Phys. Chem. C 120 (2016) 9904–9913. [39] C. Struzzi, M. Scardamaglia, M. Angelucci, L. Massimi, C. Mariani, M.G. Betti, Adsorption energy of iron-phthalocyanine on crystal surfaces, Nuovo Cimento C 36 (2013) 51–57. [40] M. Scardamaglia, C. Struzzi, S. Lizzit, M. Dalmiglio, P. Lacovig, A. Baraldi, C. Mariani, M.G. Betti, Energetics and hierarchical interactions of metal–phthalocyanines adsorbed on Graphene/Ir(111), Langmuir 29 (2013) 10440–10447.

[41] M. Massimi, M. Angeluci, P. Gargiani, M.G. Betti, S. Montoro, C. Mariani, Metal-phthalocyanine ordered layers on Au(110): metal-dependent adsorption energy, J. Chem. Phys. 140 (2014), 244704. [42] M. Roos, A. Breitruck, H.E. Hoster, R.J. Behm, Entropic stabilization of large adsorbates on weakly binding substrates—a thermal desorption and scanning tunneling microscopy study, Phys. Chem. Chem. Phys 12 (2010) 818–822. [43] A. Sabik, F. Gołek, G. Antczak, Note: work function change measurement via improved Anderson method, Rev. Sci. Instrum. 86 (2015), 056111. [44] A. Sabik, F. Gołek, G. Antczak, Reduction of the low-energy electron reflectivity of Ag(100) upon adsorption of cobalt phthalocyanine molecules, J. Phys. Chem. C 120 (2016) 396–401. [45] F. Bischoff, K. Seufert, W. Auwärter, S. Joshi, S. Vijayaraghavan, D. Écija, K. Diller, A.C. Papageorgiou, S. Fisher, F. Allegretti, D.A. Duncan, F. Klappenberger, F. Blobner, R. Han, J.V. Barth, How surface bounding and repulsive interactions cause phase transformations: ordering of a prototype macrocycling compound on Ag(111), ACS Nano 4 (2013) 3139–3149. [46] C. Wagner, D. Kasemann, C. Golnik, R. Forker, M. Esslinger, K. Müllen, T. Fritz, Repulsion between molecules on a metal: monolayers and submonolayers of hexa-peri-hexabenzocoronene on Au(111), Phys. Rev. B 81 (2010), 035423. [47] I. Kröger, B. Stadtmüller, Ch. Stadler, J. Ziroff, M. Kochler, A. Stahl, F. Pollinger, T.-L. Lee, J. Zegenhagen, F. Reinert, Ch. Kumpf, Submonolayer growth of copper-phthalocyanine on Ag(111), New. J. Phys. 12 (2010), 083038. [48] S.C. Bobaru, E. Salomon, J.-M. Layet, T. Angot, Structural properties of iron phtalocyanines on Ag(111): from the submonolayer to monolayer range, J. Phys. Chem. C 115 (2011) 5875–5879. [49] I. Kröger, P. Bayersdorfer, B. Stadtmüller, Ch. Kleimann, G. Mercurio, F. Reinert, Ch. Kumpf, Submonolayer growth of H2-phthalocyanine on Ag(111), Phys. Rev. B 86 (2012), 195412. [50] V. Feyer, M. Graus, P. Nigge, M. Wießner, R.G. Acres, C. Wiemann, C.M. Schneider, A. Schöll, F. Reinert, Adsorption geometry and electronic structure of iron phthalocyanine on Ag surfaces: a LEED and photoelectron momentum mapping study, Surf. Sci. 621 (2014) 64–68. [51] G. Antczak, K. Boom, K. Morgenstern, Revealing the presence of mobile molecules on the surface, J. Phys. Chem. C 121 (2017) 542–549. ´ [52] G. Antczak, W. Kaminski, K. Morgenstern, Stabilizing CuPc coordination networks on Ag(100) by Ag atoms, J. Phys. Chem. C 119 (2015) 1442–1450. ˜ A. Martn-Jimenez, J. [53] B. Cirera, N. Giménez-Agulló, J. Björk, F. Martinez-Pena, Rodriguez-Franandez, A.M. Pizarro, R. Otero, J.M. Gallego, P. Ballester, J.R. Galan-Mascaros, D. Ecija,Nat. Commun. 7 (2016), 11002. ´ [54] G. Antczak, W. Kaminski, A. Sabik, Ch. Zaum, K. Morgenstern, Complex surface diffusion mechanisms of cobalt phthalocyanine molecules on Ag(100), J. Am. Chem. Soc. 137 (2015) 14920–14929. [55] A. Mugarza, N. Lorente, P. Ordejón, C. Krull, S. Stepanow, M.-L. Bocquet, J. Fraxedas, G. Ceballos, P. Gambardella, Orbital specific chirality and homochiral self-assembly of achiral molecules induced by charge transfer and spontaneous symmetry breaking, Phys. Rev. Lett. 105 (2010), 115702. [56] Y. He, J. Kröger, Y. Wang, Organic multilayer films studied by scanning tunneling microscopy, Chem. Phys. Chem. 18 (2017) 429–450. [57] M. Takada, H. Tada, Low temperature scanning tunneling microscopy of phthalocyanine multilayers on Au(1 1 1) surfaces, Chem. Phys. Lett. 392 (2004) 265–269. [58] X. Chen, Y.-S. Fu, S.-H. Ji, T. Zhang, P. Cheng, X.-C. Ma, X.-L. Zou, W.-H. Duan, J.-F. Jia, Q.-K. Xue, Probing superexchange interaction in molecular magnets by spin-flip spectroscopy and microscopy, Phys. Rev. Lett. 101 (2008), 197208. [59] M.-L. Tao, Y.-B. Tu, K. Sun, Y. Zhang, X. Zhang, Z.-B. Li, S.-J. Hao, H.-F. Xiao, J. Ye, J.-Z. Wang, Structural transitions in different monolayers of cobalt phthalocyanine film grown on Bi(1 1 1), J. Phys. D 49 (2016), 015307.