Chemical Physics Letters 444 (2007) 85–90 www.elsevier.com/locate/cplett

Reversible switching of tetra-tert-butyl-azobenzene on a Au(1 1 1) surface induced by light and thermal activation Sebastian Hagen, Felix Leyssner, Dhananjay Nandi, Martin Wolf, Petra Tegeder

*

Freie Universita¨t Berlin, Fachbereich Physik, Arnimallee 14, 14195 Berlin, Germany Received 19 April 2007; in final form 25 June 2007 Available online 7 July 2007

Abstract Two-photon photoemission spectroscopy is employed to analyze reversible changes in the electronic structure of the molecular switch tetra-tert-butyl-azobenzene (TBA) adsorbed on Au(1 1 1), which are induced by UV-light and thermal activation. Cycles of illumination and annealing steps confirm the reversibility of the switching process, which we assign to a trans/cis-isomerization of TBA molecules in direct contact with the Au(1 1 1) surface. Pronounced changes in the photoelectron spectra due to UV-light exposure allow to calculate an effective cross section of reff  8 · 1022 cm2 at 4.14 eV and reff  4 · 1021 cm2 at 4.4 eV, respectively, for the trans- to cisisomerization.  2007 Elsevier B.V. All rights reserved.

1. Introduction The switching of functional molecular properties via conformational changes is a well-known phenomena in nature. Many biological systems are based on the cooperation of individual molecular units whose specifically functions are enabled and controlled by defined changes of the molecular orientation. Such switching processes represent an attractive route for the development of new technologies with possible applications in molecular electronics [1]. Molecular switches, in particular azobenzene and its derivatives, are promising candidates for molecular electronic devices, which require the direct contact of the molecules with (metal) electrodes [1–3]. Azobenzenes undergo a reversible photo-induced conformational change at an appropriate wavelength between the planar trans-isomer and the three-dimensional cis-isomer [4–6]. Whereas the switching behavior of azobenzene in the gas phase and in solution is well studied, it is a central question how the switching properties change when the *

Corresponding author. Fax: +49 30 838 56059. E-mail address: [email protected] (P. Tegeder). URL: http://www.physik.fu-berlin.de/~femtoweb (P. Tegeder).

0009-2614/$ - see front matter  2007 Elsevier B.V. All rights reserved. doi:10.1016/j.cplett.2007.07.005

molecule is adsorbed on a substrate. Thereby the electronic coupling strength plays an important role since the lifetime of excited states of the molecule will decrease significantly when bound to a metal surface. Thus, the switching characteristics are presumably not preserved or rather different switching mechanisms might be accessible. In this Letter, we investigate by means of two-photon photoemission (2PPE) spectroscopy UV-light induced changes in the electronic structure of 3,3 0 ,5,5 0 -tetra-tertbutyl-azobenzene (TBA, see Fig. 1) adsorbed on Au(1 1 1) and the reverse process induced by thermal activation. We choose this molecule because the four lateral tertbutyl-groups act as ‘spacer legs’ [7–9] to reduce the electronic coupling between the optically active molecular orbitals and the metal substrate. In solution TBA exhibits the photochemical and thermal isomerization behavior typical for azobenzene derivatives [4–6]. Furthermore it is known from scanning tunneling microscopy (STM) that on the Au(1 1 1) surface deposition of TBA in the sub-monolayer regime leads to the formation of well-ordered islands with the molecules in a planar configuration, which is assigned to the trans-isomer [9,10]. In the gas and liquid phase this isomer is established to be the energetically favorable configuration [4–6].

86

S. Hagen et al. / Chemical Physics Letters 444 (2007) 85–90

respectively, and thermal activation for the reverse process is used. We apply 2PPE to demonstrate the switching since it is associated with reversible changes in the electronic structure of the system.

TBA/Au(111) TBA coverage 1.45 ML

QMS Intensity of mass 190

1.35 ML 1ML

2. Experimental

0.95 ML 0.9 ML

α1

α2

0.5 ML

α3

1ML

250

300

350

400

450

500

550

600

Temperature [K] Fig. 1. Thermal desorption spectra of TBA on Au(1 1 1) at different coverages recorded with a heating rate of 1 K/s at the fragment-mass of 190 amu.

It has been shown recently that the trans/cis-isomerization of azobenzene and its derivatives adsorbed on noble metals can be achieved by excitation with an STM-tip [9,11–13]. Isomerization of azobenzene [12] and the azobenzene derivative Disperse Orange 3 [11] induced by resonant and inelastic tunneling, respectively, have been proposed. In the case of TBA on Au(1 1 1) a reversible isomerization stimulated by the applied electric field [9] as well as by resonant electron attachment into the lowest unoccupied molecular orbital (LUMO) [14] have been discussed. Recently light-induced switching of individual TBA molecules adsorbed on Au(1 1 1) have been observed by Crommie and coworkers using STM [10]. This is remarkable, since the molecules are in direct contact with the gold surface. In the latter work, both the trans–cis- and the cis– trans-isomerization have been achieved by illumination with UV-light at the same wavelength of 375 nm (3.3 eV). This fact is surprising since at least for TBA in solution two different wavelength namely at 325 nm ð3:81 eV; trans ! cisÞ and 440 nm ð2:81 eV; cis ! transÞ are necessary to drive the reversible trans/cis-isomerization. However, the underlying mechanisms of the switching process of adsorbed TBA on Au(1 1 1) is not yet resolved. Generally, separate excitations, e.g. different wavelengths are essential to control the direction of switching between different molecular states. In this work, we employ two different excitation processes to stimulate a reversible conformational change in a well-ordered TBA ensemble adsorbed on Au(1 1 1). For the bidirectional switching UV-light at 4.14 and 4.4 eV,

The experiments were carried out in an ultrahigh vacuum (UHV) chamber with a base pressure <1 · 1010 mbar combined with an amplified tunable femtosecond laser system. The Au(1 1 1) crystal was mounted on a liquid nitrogen cooled cryostat which in conjunction with resistive heating enables temperature control from 90 K to 750 K. The crystal was cleaned by cycles of Ar+ sputtering and annealing. The TBA was evaporated from a home-built effusion cell onto the clean Au(1 1 1) surface held at 260 K. The coverage was analyzed with thermal desorption spectroscopy and work function measurements. For the illumination of the TBA-covered Au(1 1 1) laser pulses with photon energies of 4.14 and 4.4 eV and a power of 16.9 W/ cm2 and 1.7 W/cm2, respectively, were used. The annealing experiments were carried out by heating the sample from 90 K to 320 K with a heating rate of 1 K/s and subsequently cooling to 90 K. For the two-photon photoemission experiments the same photon energies (4.14 or 4.4 eV) are applied as for the illumination. Since the UVphotons caused a change in the 2PPE features during data collection, every 2PPE spectrum was recorded for only 5 s. The 2PPE spectra presented below are displayed as 2PPE intensity versus the final state energy with respect to the Fermi level (EF = 0 eV), Efinal  EF = Ekin + U, with U the work function and Ekin the kinetic energy of the photoelectrons. For a detailed description of the 2PPE experiment see Ref. [15]. The effective cross section (reff) for the UV-light induced trans- to cis-isomerization is determined as follows: The laser spatial profile is characterized by a CCD camera located at a position outside the UHV chamber, which is equivalent to the sample position and the fluence (Fi) for each camera pixel i is determined. It is assumed that the number of switched molecules (Ni) scales linearly with the fluence, i.e., dNi/dt  exp(reff Æ Fi Æ t) (first-order process). For the detection of the switching via the change in the photoemission intensity, a two-photon process is required (second-order process) and therefore the 2PPE intensity is proportional to the squared fluence (DI i ðtÞ / dN i =dt  F 2i ). The observed total change in intensity is then equal to the sum of the intensity change for each camera pixel i (DItotal = RDIi). By using an exponential saturation function for the first-order switching process one can estimate an effective cross section: X DI total ¼ F 2i DI 1 ½1  expðreff  F i  tÞ ð1Þ whereby DI1 corresponds to the asymptotic change of the photoemission intensity.

S. Hagen et al. / Chemical Physics Letters 444 (2007) 85–90

3. Results and discussion

A

0.9 ML TBA/Au(111)

5

2PPE Intensity [arb. units]

To obtain insights into the adsorption properties of TBA/Au(1 1 1) and to quantify the coverage, thermal desorption spectra are recorded as a function of coverage (see Fig. 1). At low coverage a broad desorption peak (a3) is observed around 525 K, which extends to lower temperatures with increasing coverage. Close to the saturation of this peak a second desorption feature a2 develops at 400 K. Further increase in coverage leads to saturation of the a3 and a2 peaks and to the appearance of a sharp desorption peak a1 around 314 K. The latter (a1) increases in height and width with increasing coverage, showing a typical zero-order desorption behavior. We therefore assign the a1 peak to desorption from the multilayer while the a2 and a3 peaks are associated with desorption from the first monolayer (ML). The a2 peak represents the desorption of 10% of the monolayer coverage. The origin of this peak might be attributed to a structural rearrangement within the first (e.g. compressed) TBA layer. Note that the thermal desorption behavior of TBA from Au(1 1 1) is very similar to that observed form TBA adsorbed on Ag(1 1 1) [16] and to those of other aromatic compounds on nobel metal surfaces [17–20]. Since it is known from STM that TBA on Au(1 1 1) forms well-ordered islands in the low-coverage regime with the molecules adsorbed in the planar configuration [9,10], we assume that up to 0.9 ML a well-ordered TBA layer in the planar (trans-) configuration exists. Therefore, all measurements are performed at a coverage of 0.9 ML which is prepared by heating the multilayer-covered surface to 420 K. Fig. 2a shows one-color 2PPE spectra of 0.9 ML TBA on Au(1 1 1) recorded at a photon energy of 4.14 eV before and after illumination for 350 s, respectively. From the non-illuminated adsorbate-covered surface (1. scan) four peaks labeled as A, B, SS, and IS are observed: The peaks A and B correspond to photoemission from the occupied dbands of Au(1 1 1). They are located at 2.55 eV (A) and 1.96 (B) below EF [21,22]. The peak SS results from the occupied Shockley surface state located in the L-projected band gap of the Au(1 1 1) surface. The energetic position is 0.48 eV with respect to EF in accordance with previous work [23]. These three peaks are also observed for the clean Au(1 1 1) surface (see inset of Fig. 2a). The peak labeled as IS arises from the n = 1 image potential state, which lies 0.6 eV below Evac. For comparison, the n = 1 image potential state on the clean Au(1 1 1) possesses a binding energy of 0.8 eV with respect to Evac [24], whereas on the n-heptane (1 ML) covered surface it is also located at 0.6 eV [25]. Illumination of the TBA/Au(1 1 1) system with UV-light (hm = 4.14 eV) caused significant changes in the 2PPE spectrum: (i) The work function (U) shifts to lower energies, (ii) the n = 1 image potential state loses intensity, and (iii) the photoemission intensity near the secondary edge increases strongly (see Fig. 2b). U is determined by the low energy cutoff (secondary edge or vacuum level, Evac) and high energy cutoff (Fermi

a

87 Au (111) hν = 4.14 eV B SS

8 7 6 EFinal - EF [eV]

hν = 4.14eV A

illumination time 350 s 1. Scan B

ΔΦ = 50meV

IS

SS

b difference spectrum

x4 5.0

6.0

7.0

8.0

EFinal - EF [eV] Fig. 2. (a) One-color 2PPE spectra of 0.9 ML TBA adsorbed on Au(1 1 1) before and after illumination for 350 s recorded at a photon energy of 4.14 eV. Inset: 2PPE spectrum of the clean Au(1 1 1) surface. (b) Difference spectrum obtained from the photoemission spectra shown in (a). This spectrum clarifies the changes in the photoemission due to UV-light exposure (see text).

edge, EFE = EF + 2hm) of the 2PPE spectra, by the relation U = 2hm  (EFE  Evac), where hm is the photon energy used in the one-color 2PPE experiment. Adsorption of 0.9 ML TBA leads to a decrease of U from 5.38 eV for the clean Au(1 1 1) to 4.55 eV. Illumination yields to a work function decrease by DU = 50 meV. It is likely to assume that the U drop is associated with a photo-induced conformational change (e.g. trans–cis-isomerization) of the TBA molecules. At low coverage TBA adsorbs in the planar (trans-) configuration [9,10]. In this configuration the molecule exhibits a vanishing dipole moment. On the other hand, the non-planar cis-isomer (free molecule) possesses a large dipole moment of 3.6 D [26]. The appearance of a permanent dipole moment after a conformational change of the adsorbed molecule could indeed be responsible for the observed decrease of U by 50 meV. In addition, UV-light exposure results in an intensity loss and smearing out of the image potential state. It is known that 2PPE spectroscopy of image potential states is a very sensitive probe of the adsorbate morphology

88

S. Hagen et al. / Chemical Physics Letters 444 (2007) 85–90

[27–30]. In the case of disordered adsorbate layers image potential state peaks are often suppressed or broadened, whereas from ordered layers these peaks are clearly visible. If the non-illuminated adsorbate is well-ordered (as known from STM experiments [9,10]) then the loss in intensity of the image state peak in consequence of the illumination could originate from a conformational change of the TBA molecules, viz. a change in the morphology of the adsorbate. Both effects, the work function decrease as well as the intensity loss of the image potential state, may be attributed to a UV-light stimulated trans- to cis-isomerization of the adsorbed TBA. This interpretation is supported by the light-induced (hm = 3.3 eV) trans/cis-isomerization of TBA observed by STM [10]. The appearance of the pronounced photoemission near the secondary edge may be assigned to an unoccupied final state of the TBA molecules in its modified configuration, i.e. cis-form, since the electron kinetic energy of this peak does not vary with the photon energy (4.14 and 4.4 eV) [16]. From the peak shape, the sharp rise on the low energy side, it is likely to assume that this state is located close to Evac, therefore we observe in the 2PPE process only the part of the state which lies above Evac. Since it is known that the reverse process, i.e. the cis- to trans-isomerization of azobenzene and its derivatives in the liquid phase can also be stimulated by thermal activation, annealing experiments were performed. In fact, these mea-

a

surements indicate that by annealing the sample to 320 K the same 2PPE spectrum with respect to intensity and shape can be received as in the non-illuminated case (1. scan) as can be seen in Fig. 3a. Repetition of the illumination and annealing steps clearly demonstrate the reversibility of the switching process. Note that the sample position during these measurements was not changed. On the basis of the work function change as a function of light exposure and annealing steps the reversibility of this process is demonstrated in Fig. 3b. We propose that the observed reversible changes in the electronic structure of the TBA adsorbed on Au(1 1 1) are due to a conformational change of the adsorbate. The UV-light induces the trans- to cisisomerization (as it has been observed in Ref. [10]) whereas the reverse process can be stimulated by thermal activation like it is illustrated in Fig. 3c. In order to quantify the light-induced trans/cis-isomerization we evaluated an effective cross section (reff) from the observed pronounced increase of the photoemission intensity near the secondary edge (see Fig. 2b). Provided that the increase correlates with the number of switched molecules and by using the exponential saturation function (see chapter 2, Eq. 1) we obtain reff for two different photon energies (4.14 and 4.4 eV) as shown in Fig. 4a. The solid lines represent a fit using the exponential saturation function (Eq. 1). From this fit one achieves an effective cross section of reff  8 · 1022 cm2 at a photon energy of 4.1 eV and 4 · 1021 cm2 at 4.4 eV, respectively. At 4.4 eV reff

b

0.9 ML TBA/Au(111)

0.9 ML TBA/Au(111)

Illumination: hν = 4.14 eV

Steps

2PPE Intensity [arb. units]

320 K

7

360 s

6

320 K

5

Work function [eV]

hν = 4.14 eV

Annealing: T = 320 K

4.58 4.56 4.54 4.52 4.50 1

c

435 s

4

320 K 3

224 s 1. Scan 5.0

6.0

7.0

8.0

2

3

4

5

6

7

Annealing-Illumination Steps

hν kT

2 1

9.0

EFinal - EF [eV] Fig. 3. Reversible switching of TBA molecules: (a) 2PPE spectra taken at a photon energy of 4.14 eV of 0.9 ML TBA/Au(1 1 1) before (step 1) and after illumination for 224 s (step 2), and cycles of illumination and annealing steps. (b) Dependency of the work function of the TBA-covered Au(1 1 1) on UVlight exposure and annealing. (c) Scheme of the reversible isomerization induced by UV-light and heat (kT).

S. Hagen et al. / Chemical Physics Letters 444 (2007) 85–90

a

b hν = 4.4 eV

1.7 W/cm

hν = 4.14 eV 16.9 W/cm

0

100

200

300 400 Time [s]

500

2

2

600

Absorbance [arb.units]

4.14 eV

trans

89

cis isomerization

before illumination after illumination @ 313 nm 4.4 eV

3.3 eV

250

300

350

400 λ [nm]

450

500

550

Fig. 4. (a) Change in the photoemission intensity as a function of illumination time for two different photon energies. The solid line represents an exponential fit using the saturation function equation (1). (b) UV–VIS absorption spectra of TBA in cyclohexane before and after illumination with UVlight at 313 nm. Exposure at 313 nm leads to a decrease of the absorbance around 325 nm which is due to the trans/cis-isomerization (see text).

exceeds the one at 4.1 eV by a factor of five. In comparison, in the trans/cis-isomerization process of TBA on Au(1 1 1) observed by Crommie and coworkers [10] 1 h exposure at a wavelength of 375 nm (3.3 eV) and a power of 90 mW/ cm2 induced switching of 4% of the monolayer, which yields an absolute cross section of 5 · 1021 cm2. This value is in the same range as reff at 4.4 eV extracted from our measurements. To evaluate the underlying mechanism responsible for the light-induced isomerization two possible scenarios can be considered: (i) The UV photons induce a direct (intramolecular) electronic excitation within the adsorbate and (ii) the incident photons indirectly drive the process, viz. a substrate-mediated photochemical process, where hot electrons are attached to the adsorbate, creating transient molecular anions [31]. In the gas and liquid phase direct electronic excitation induces isomerization of azobenzenes and its derivatives via a n ! p (S1) and p ! p (S2) electronic excitation, respectively. Fig. 4b shows UV–VIS spectra of TBA in solution. The strong absorption band at 325 nm belongs to the S2 transition, whereas the band around 440 nm corresponds to the S1 transition. The intensity loss of the absorption band at 325 nm due to illumination at 313 nm can be assigned to a trans- to cis-isomerization. The photon energies used in the present experiment and in Ref. [10] are marked in Fig. 4b. Assuming that the adsorption of TBA on Au(1 1 1) does not change the electronic structure significantly, i.e. the absorption bands are neither shifting nor broadening, then illumination at 4.4 eV as well as at 3.3 eV should be less efficient compared to exposure at 4.1 eV. This is obviously not the case, since reff at 4.1 eV is lower by a factor of five. Hence one might conclude that the intramolecular excitation is unlikely. The second possible mechanism for the optically-induced isomerization is the indirect excitation via attachment of excited electrons to the molecules. From scanning tunneling spectroscopy (STS) it is known that in TBA adsorbed on Au(1 1 1) the

switching process can be stimulated by resonant tunneling into the LUMO level [14]. This mechanisms has also been discussed in the case of the switching of azobenzene adsorbed on Au(1 1 1) [12]. However, in order to verify the validity of the two scenarios detailed wavelength dependent measurements over a wide energy range are required. In summary, we have demonstrated that the 3,3 0 ,5,5 0 tetra-tert-butyl-azobenzene (TBA) adsorbed on Au(1 1 1) can be reversible switched by UV-light and thermal activation. To distinguish between the two conformational states two-photon photoemission (2PPE) is used, since they differ in respect of the electronic structure. The differences which are observed in the photoemission spectra are a shift in the work function (DU = 50 meV), a significant change in the photoemission intensity near the secondary edge, and an intensity change of the n = 1 image potential state. The underlying mechanism for the optically-induced conformational change (trans- to cis-isomerization) has still to be elucidated from wavelength dependent measurements. Acknowledgments This work has been supported by the Deutsche Forschungsgemeinschaft through the SFB 658. We thank S. Hecht and M.V. Peters (Humboldt Universita¨t Berlin) for the preparation of the azobenzene derivative. References [1] M.R. Bryce, M.C. Petty, D. Bloor, Molecular Electronics, Oxford University Press, New York, 1995. [2] B.L. Feringa, Molecular Switches, Wiley-VCH, Weinheim, 2001. [3] Special Issue: Photochromism: Memories and Switches, M. Irie (Ed.), Chem. Rev. 100 (2000) 1683. [4] N. Tamai, O.H. Miyasaka, Chem. Rev. 100 (2000) 1875. [5] H. Rau, in: H. Du¨rr, H. Bouas-Laurent (Eds.), Photochromism – Molecules and Systems, Elsevier, Amsterdam, 2003, p. 165. [6] D. Fangha¨nel, G. Timpe, V. Orthman, in: A.V. El’tsov (Ed.), Organic Photochromes, Consultants Bureau, New York, 1990, p. 105.

90

S. Hagen et al. / Chemical Physics Letters 444 (2007) 85–90

[7] T.A. Jung, R.R. Schlittler, J.K. Gimzewski, Nature 386 (1997) 696. [8] F. Moresco, G. Meyer, K.-H. Rieder, H. Tang, A. Gourdon, C. Joachim, Phys. Rev. Lett. 86 (2001) 672. [9] M. Alemani, M.V. Peters, S. Hecht, K.-H. Rieder, F. Moresco, L. Grill, J. Am. Chem. Soc. 128 (2006) 14446. [10] M.J. Comstock et al., Phys. Rev. Lett., in press. [11] J. Henzl, M. Mehlhorn, H. Gawronski, K.-H. Rieder, K. Morgenstern, Angew. Chem. Int. Ed. 45 (2006) 603. [12] B.-Y. Choi et al., Phys. Rev. Lett. 96 (2006) 156106. [13] J. Henzl, T. Bredow, K. Morgenstern, Chem. Phys. Lett. 435 (2007) 278. [14] M. Alemani, L. Grill, M.V. Peters, S. Hecht, F. Moresco, private communication. [15] P.S. Kirchmann, P.A. Loukakos, U. Bovensiepen, M. Wolf, New. J. Phys. 7 (2005) 113. [16] P. Tegeder et al., Appl. Phys. A 88 (2007) 465. [17] M. Xi, M.X. Wang, S.K. Jo, B.E. Bent, P. Stevens, J. Chem. Phys. 101 (1995) 9122. [18] T. Vondrak, X.-Y. Zhu, J. Phys. Chem. B 103 (1999) 3449. [19] Q. Zhong, C. Gahl, M. Wolf, Surf. Sci. 496 (2002) 21. [20] T.J. Rockey, M. Yang, H.L. Dai, J. Phys. Chem. B 110 (2006) 19973.

[21] H. Eckhart, L. Fritsche, J. Noffke, J. Phys. F: Met. Phys. 14 (1984) 97. [22] R. Courths, H.-G. Zimmer, A. Goldmann, H. Saalfeld, Phys. Rev. B 34 (1986) 3577. [23] F. Reinert, G. Nicolay, S. Schmidt, D. Ehm, S. Hu¨fner, Phys. Rev. B 63 (2001) 115415. [24] T. Fauster, W. Steinmann, in: P. Halevi (Ed.), Electromagnetic Waves: Recent Developments in Research, Elsevier, Amsterdam, 1995. [25] C.D. Lindstrom, D. Quinn, X.-Y. Zhu, J. Chem. Phys. 122 (2005) 124714. [26] G. Fu¨chsel, T. Klamroth, J. Dokic´, P. Saalfrank, J. Phys. Chem. B 110 (2006) 16337. [27] C. Reuß, I.L. Shumay, U. Thomann, M. Kutschera, M. Weinelt, T. Fauster, U. Ho¨fer, Phys. Rev. Lett. 82 (1999) 153. [28] W. Berthold, F. Rebentrost, P. Feulner, U. Ho¨fer, Appl. Phys. A 78 (2004) 131. [29] K. Boger, M. Weinelt, T. Fauster, Phys. Rev. Lett. 92 (2004) 126803. [30] P. Tegeder, M. Danckwerts, S. Hagen, A. Hotzel, M. Wolf, Surf. Sci. 585 (2005) 177. [31] F.M. Zimmermann, W. Ho, Surf. Sci. Rep. 22 (1995) 127.