Spatially resolved work-function manipulation of azobenzene-functionalized self-assembled monolayers by optical stimulation Open Access

Azobenzene-based self-assembled monolayers (SAMs) are prototype systems for photoresponsive surfaces.^1–7 Their photochromic switching ability can, e.g., be used to tune level alignments at interfaces.^4–7 Here, we discuss local changes of the SAM's work function induced by altering the static dipole moment of the chromophores with controlled photoisomerization. Azobenzene is a conformational switch, consisting of two phenyl rings coupled via a dinitrogen bond (see inset in Fig. 1). In its trans form, the molecule is planar and the two rings are on opposite sides of the dinitrogen bond axis. In the cis conformation, both rings are on the same side of the reference plane. Therefore, the molecule has no static dipole moment in the trans configuration but is strongly polar in its cis state.⁸ Dissolved in solution, light in the blue to ultraviolet (UV) range triggers photoswitching between the two isomers.⁹ For fabricating photoresponsive surfaces, we linked azobenzene to Au(111)/Mica substrates via an alkanethiol chain following an elaborated procedure.^10–14 Utilizing a linker with a chain length of 11 carbon atoms, the azobenzene head groups are sufficiently decoupled from the surface (see Fig. 1 insets). We will denominate the azobenzene derivative 11-(4-(phenyldiazenyl)phenoxy)-undecane-1-thiol used in this work as Az11. It is essential to dilute the chromophore density to prevent steric hindrance.^12,15 We observed efficient photoswitching in mixed SAMs of Az11 and a pure alkanethiolate of comparable length (C12) prepared in a wet-chemical process.¹³ In this work, we used samples with 50% Az11 surface coverage.¹³ 

Near edge x-ray absorption fine structure (NEXAFS) spectroscopy showed that in a pure Az11 SAM the trans chromophores are tilted by about 30° from the surface plane.¹² In diluted SAMs, the chromophores tilt further toward the SAM surface.¹⁴ When switching into its cis form, the molecule flips the upper phenyl ring and the z-component of its static dipole moment points toward the surface. Thus, the work function of the SAM/Au(111) sample increases with the number of molecules in cis configuration.^4,7,16 This allows us to locally tune the work function and image areas of switched molecules in the SAM via spatially resolved photoelectron emission. Although external work function manipulations of azobenzene-functionalized surfaces have been studied by several groups,^{4–7,17–20} so far no analysis of the switched area was performed. Controlling homogeneity and locality of the switched area is highly relevant for future applications. Here, we use the photoemission electron microscopy (PEEM) mode of a time-of-flight momentum microscope for getting access to spatial modulations in the micrometer range. PEEM allows us to analyze the size and profile of switched areas. We demonstrate that the absolute value of the local work function and hence the isomerization profile as well as the time span to switch the work function can be tuned by laser illumination.

In Fig. 1, we compare the absorbance of Az11 molecules in the methanol solution and the differential reflectance of mixed SAMs from Ref. 13 with the spectrum of a mercury arc lamp. In the SAM, the S₂ band splits and its maximum shifts to higher energies with increasing Az11 percentage (Fig. 1, middle panel). This is attributed to excitonic coupling between the trans isomers, whereby the $π π ∗$ transition-dipole moments form an H-aggregate. Molecules switched to the cis state show no excitonic coupling.²¹ Irradiation at the low-energy edge of the S₂ absorption band at about 370 nm leads to trans–cis isomerization in the SAM. The cis–trans back-isomerization is triggered by illumination either at the high energy edge of the S₂ band or in the S₁ band, showing up at around 440 nm.²¹ 

We mainly used two complementary photon sources, a continuous wave (cw) laser (372 nm) and a Hg lamp for illumination in the S₂ and S₁ absorption bands, respectively. Simultaneous excitation facilitates optical tuning of the cis–trans ratio at the functionalized surface. The cw laser initiates the trans–cis isomerization only, whereas the Hg lamp contributes to both trans–cis and cis–trans configuration changes. As illustrated in Fig. 1, several emission lines of the mercury lamp are in the range of the S₁ and S₂ absorption bands of the Az11 SAM. The intense emission line at 436 nm illuminates in the S₁ band center, leading to efficient cis–trans isomerization. The emission line at 365 nm instead triggers trans–cis switching. Considering the relative photon flux of both emission lines we estimated the photostationary state (PSS) of the mercury lamp (Hg-PSS) as discussed in detail in the supplementary material (supplementary material, Sec. II). We find photon fluxes of $j 436 nm = 1.9 × 10 14$ cm⁻² and $j 365 nm = 3.3 × 10 14$ cm⁻² leading to a PSS with 67% cis molecules. Starting from this PSS, a maximal work function change of 30 meV can be reached when tuning toward a pure cis SAM by simultaneous laser illumination at 372 nm.²¹ 

The high-energetic lines of the mercury lamp lead to photoelectron emission from the functionalized surface.^16,21 We perform spatially resolved photoemission experiments in an ultra-high vacuum chamber using a momentum microscope.^22,23 The instrument has been set up in collaboration with Schönhense. In this work, we use the instrument's spatially resolving “Gaussian” imaging mode comparable to a photoelectron emission microscope.²² The field of view (FoV) on the sample surface is selected by real space apertures inserted between two zoom optics. Photoelectrons are detected after a time-of-flight tube by single event counting through a delay line detector.²⁴ A $( 5 × 5 )$ mm² chess-patterned gold-on-silicon sample (Chessy, Plano) was characterized for initial alignment and calibration of the focusing and magnifying electron optics, exploiting the instrumental resolution. Figure 2 shows spatially resolved electron images of the $( 10 × 10 )$ μm² chess patterns. With a FoV of 100 and 800 μm diameter, we obtained spatial resolutions below 3 and 16 μm, respectively (see the supplementary material, Sec. V). Since the measurements on the Az11 SAMs required a large FoV, we used the latter settings which still provide appropriate spatial resolution.

Trans–cis isomerization of the SAM in a selected area was triggered focusing the 372 nm laser beam onto the sample. An image of the laser spot recorded with a CCD camera is shown in Fig. 3(a) together with horizontal (x) and vertical (y) beam profiles taken across the spot center. The focus has an elliptical profile with full width at half maximum (FWHM) of 19 and $25 ± 1$ μm. Beam profiles were modeled using 2D Gaussian and Cauchy distributions. The latter accounts particularly better for the projection of the laser spot along the direction of incidence x.

To spatially resolve the work function shift across the sample, we recorded local electron yield (LEY) images in PEEM mode with and without laser illumination. The 372 nm beam was focused in the upper left part of the FoV and impinged at a grazing angle of $22 °$ onto the sample. This expands the horizontal spot profile to a FWHM of around 67 μm. Figures 3(b) and 3(c) show the detector sensitivity corrected normalized images [LEY(Hg + 372) − LEY(Hg)]/LEY(Hg) = (ΔLEY/LEY₀) for identical spot profiles, but changing laser power by one order of magnitude. Before image acquisition, the sample was exposed to laser illumination for about 300 s while the acquisition time for each images was about 150 s. The PEEM image in Fig. 3(b) was recorded with a laser power of $1.7 ± 0.3$ μW. This corresponds to a photon flux of $∼ 1 × 10 17 cm − 2 s − 1$ in the laser spot center, leading to about 1 switching event every 10 s.^13,21 At this threshold fluence, we start to observe a local shift of the PSS state toward higher work function (higher density of cis-isomers). From the horizontal and vertical profiles, we estimate a spot size of 60 × 200 μm², about $3 − 4$ times larger than the FWHM of the laser focus.

Increasing the 372 nm photon flux by a factor of 10, the area in Fig. 3(c) showing a decreased EY is more pronounced and significantly larger. Horizontal and vertical profiles reveal a magnification of the projected laser focus FWHM by a factor of 10. While the differential photoemission intensity ΔLEY changes gradually along the direction of laser incidence x, comparably sharp edges and a nearly flat top are observed along the y-direction.

The photon flux in the laser spot not only varies the local PSS but also the kinetics to reach this PSS. As a measure for the overall switching kinetics, we evaluate the transient of the total electron yield TEY(t) by summing up the electron count-rate in the complete FoV as a function of exposure time t. As shown in Fig. 5(a), TEY(t) decreases as soon as the laser beam illuminates the sample, since the Az11 SAM changes toward the new PSS that includes overall a higher fraction of cis isomers.^4,16,21 When turning laser illumination off, the ensemble of Az11 molecules relaxes back to the Hg-PSS.

The solid lines in Fig. 5(a) show the calculated change of the TEY for the fitted Cauchy and Gaussian laser spot profiles (cf. Fig. 3). Using the Cauchy distribution, we get poor agreement between modeled and measured switching kinetics (cyan line). The Gaussian beam profile (red line) shows a much better agreement. This matches the observation of sharp edges of the switched areas, shown by the line profiles in Fig. 3(c). We modeled the temporal evolution of the spatial work function distribution in a defined FoV to illustrate how the LEY gradually evolves under laser illumination. Figure 5(b) shows exemplary images after 1, 10, 50, and 150 s laser illumination with a Gaussian profile, Fig. 5(c), the corresponding kinetics curves for selected positions in real space. In the center of the beam, the PSS state is close to a pure cis SAM and is reached after few seconds for 48 μW laser power (blue curve). In the tails of the laser spot, the local PSS is not reached within 100 s (orange curve). The fast and precise work function tuning modeled for the spot center can be shown using photoemission from a pulsed laser source operated at a wavelength in the S₂ band. This gives excitation densities high enough to induce two-photon photoemission (2PPE) and enables direct tracing of the work function change as shown in Fig. 5(d) for 361 and 305 nm photons. Upon excitation with 305 nm, the low energy cutoff of the photoemitted electron spectrum shifts toward 4.22 eV with respect to 4.27 eV when illuminated by 361 nm. This indicates a work function shift of 50 meV. Enhancing the capabilities of the experimental PEEM setup with a new pulsed laser source (supplementary material, Sec. V) will enable spatially resolved measurements of the work function distribution, paving the path for transient μm work function patterning beyond the LEY signatures.

In conclusion, we showed that well-defined and homogeneous switched areas can be produced by illuminating an azobenzene-functionalized surface with a focused UV beam without the need of current patterning techniques with predefined vignettes,^25,26 enabling on-the-fly modifications. When shaping the beam profile, the absolute value of the local work function as well as the timescale required to switch the work function can be controlled. The switched area and isomerization profile depend on the laser spot profile paving the route toward designing cheap organic optoelectronic devices with an active area, tuning not only the work function, but, e.g., the lateral excitonic coupling among the photoswitches.

See the supplementary material for details on the 2D spot profile modeling, the determination of the mercury discharge lamp induced photostationary state, and the conversion of the work function changes into a photoelectron yield. These input parameters were used for simulating the spatially resolved work function changes of the diluted Az11 SAM. Furthermore, we show the characterization of the PEEMs spatial resolution and the extraction procedure for obtaining the difference PEEM images shown in the main text.

We thank Rafal Klajn for the kind supply of the Az11 molecules and Johannes Mosig for fruitful discussions and computational support. J.B. thanks the International Max Planck Research School for Elementary Processes in Physical Chemistry for the financial support in the initial phase of his Ph.D. This project was supported by Deutsche Forschungsgemeinschaft via the Collaborative Research Center TRR 227 on Ultrafast Spin Dynamics, projects A01 and B07. Part of the equipment was financed by the Deutsche Forschungsgemeinschaft via the program for major research instrumentation (91b GG).

The authors have no conflicts to disclose.

Jan Böhnke: Data curation (lead); Formal analysis (equal); Investigation (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Beatrice Andres: Conceptualization (equal); Data curation (equal); Investigation (lead); Supervision (equal); Writing – review & editing (equal). Larissa Boie: Investigation (equal); Writing – review & editing (equal). Angela Richter: Data curation (equal); Investigation (equal); Writing – review & editing (equal). Cornelius Gahl: Conceptualization (equal); Investigation (equal); Methodology (equal); Writing – review & editing (equal). Martin G. Weinelt: Conceptualization (equal); Funding acquisition (lead); Project administration (lead); Resources (lead); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal). Wibke Bronsch: Conceptualization (equal); Data curation (equal); Formal analysis (lead); Investigation (equal); Methodology (lead); Supervision (equal); Validation (equal); Visualization (lead); Writing – original draft (lead); Writing – review & editing (lead).

The data that support the findings of this study are available from the corresponding authors upon reasonable request.