Article pubs.acs.org/JPCC
Plasmon-Enhanced Single-Molecule Electroluminescence: A Computational Study Yuan Zhang,† Yaroslav Zelinskyy,†,‡ and Volkhard May*,† †
Institut für Physik, Humboldt−Universität zu Berlin, Newtonstraße 15, D-12489 Berlin, Germany Bogolyubov Institute for Theoretical Physics, National Academy of Sciences of Ukraine, Metrologichna str., 14-b, UA-03143, Kiev, Ukraine
‡
ABSTRACT: Photoemission due to current formation through a molecular junction is studied theoretically. Placing a single molecule between two spherical leads in which plasmon excitations are in near-resonance to the molecular excitation energy, an increase in the emission intensity by more than three orders of magnitude is demonstrated. Such a plasmon-enhanced singlemolecule electroluminescence can be obtained for a small as well as a large reorganization energy upon molecular excitation. In the latter case, a certain part of the vibrational progression in the emission spectrum undergoes an increase. For large charging energies of the molecule, electroluminescence enhancement can be observed in a voltage range around 2 V.
I. INTRODUCTION Plasmon enhancement of molecular photoemission is an intensively studied phenomenon in the field of plasmonics. (For a recent review, see refs 1 and 2.) The enhancement is particularly strong if the molecule is placed between two spherical metal nanoparticles (MNPs), forming a so−called plasmonic nanoresonator.3,4 In any case, such experiments presuppose initial optical excitation of the molecule before emission enhancement can be studied. However, excited states of the molecule can also be prepared by external charge injection, which requires to contact the molecule by nanoelectrodes and to apply a voltage. Such type of a molecular junction formed by a single molecule attached to two leads has been studied in detail within the last two decades. (For recent reviews, see refs 5−8.) Experiments were also reported repeatedly where excited states of the molecule have been addressed and where the subsequent electroluminescence could be detected.9−12 In the following, we demonstrate how a huge plasmon enhancement of single-molecule electroluminescence can be achieved. Therefore, we suggest a molecular junction where the plasmon excitations of the leads are in near-resonance to the molecular excitation energy. This can be best realized in using spherical Au nanoparticles as leads (cf. Figure 1). Choosing a particular radius, the plasmon excitation can be placed in the S0−S1 transition region of many molecules. Then, excited molecular states can be populated by charge transmission if a voltage is applied in a suitable range. The molecular junction we have in mind has not been reported in literature so far, but an ensemble of spherical MNPs separated by organic molecules and contacted by electrodes was already described in refs 13−15. © 2012 American Chemical Society
Figure 1. Molecular junction with two leads and a single molecule sandwiched between them. Upper panel: pyramidal leads with plasmon resonances strongly separated from the molecular electronic excitation. Lower panel: spherical leads (Au with 20 nm diameter) with surface-to-surface distance ΔX and with hybrid dipole plasmon excitations resonant to the molecular electronic excitation.
To compare electroluminescence with and without plasmon enhancement, two types of junctions, as displayed in Figure 1, will be considered. The molecule is contacted by two leads with a pyramidal shape to which a spherical MNP may be attached. If the spherical MNPs are absent (upper panel of Figure 1), then we do not expect plasmon enhancement. (The collective excitations of the leads shall be out of resonance with respect to the molecular transition.) If spherical MNPs are placed at the end of the leads (lower panel of Figure 1), then emission Received: October 9, 2012 Revised: November 22, 2012 Published: November 22, 2012 25962
dx.doi.org/10.1021/jp309987c | J. Phys. Chem. C 2012, 116, 25962−25968
The Journal of Physical Chemistry C
Article
enhancement becomes possible. (The molecular transition shall be in resonance to the MNP dipole plasmon.) Let us briefly comment on the enhancement effect. First, we emphasize the need to treat the molecular excitations and the MNP plasmons as a uniform quantum system. Because the extension of the molecule lead system is small enough that the respective interaction does not exhibit any retardation effect. The residual (instantaneous) Coulomb coupling between the molecular and the lead−plasmon excitations is dominated by excitation energy exchange processes.16 If those are strong enough, then common states of the molecule lead−plasmon system are formed. As a consequence, photons are emitted from a uniform system with a rate much larger than the one of the isolated molecule (decoupled from the lead plasmons).18 This enhancement of molecular emission can be understood as an oscillator strength transfer from the lead−plasmon excitations. The enhanced release of radiation energy due to an increased probability of photon emission is achieved by a larger work per time performed by the current when moving through the junction. Because the plasmon excitations in the two leads are coupled to each other, hybrid excitations are formed. (We will study the case where the molecular excitation is in resonance to the lower MNP hybrid plasmon level.) In the following description, the role of the leads is two-fold. The leads act as a reservoir of electrons responsible for the current through the molecule. At the same time, collective plasmon excitations are possible, which, however, are assumed to be decoupled from the charge-transmission process. For the simulations, we use our junction model with two spherical leads as introduced in ref 17 and utilize the density matrix theory described in refs 19 and 20. (We do not repeat respective formulas but refer to our older studies.) The subsequent section introduces the used junction model. In Section 3, we briefly comment on the density matrix approach and the way to determine the current. The computation of the emission spectrum is explained in Section 4. All results are presented in Section 5. The article ends with some concluding remarks quoted in Section 6
Table 1. Used Parameters (For Explanation See Text) ΔE10 ℏωvib Q0g Q0e Q1g Q1e dmol γmol ℏΓ Epl dpl γpl kBT 2πJ
0.05 ... 1 eV 50 meV 0 4 6 2 8D 3 meV 1 meV 2.5949 eV 2893 D 28.5 meV 25 meV 1 meV
transfer coupling. It is characterized by the common coupling parameter Γ. (It should not be different for different electronic levels of the molecule and for the two leads.) We assume that in the considered voltage range only the singly negatively charged state of the molecule is involved in the current formation. The molecular energies are ENa. The charging number N = 0,1 counts the number of excess electrons in the molecule, and the electronic quantum number a refers to the ground state a = g and an excited state a = e. If the molecule is charged, then an electrostatic coupling to the leads becomes possible, which shall be included in the definition of the energies E1a. The difference ΔE10 = E1g − E0g − μ0 gives the so−called relative charging energy. (μ0 is the uniform chemical potential of the leads.) Because less is known on these energies, assumptions have to be taken. We set ΔE10 > 0; that is, charging is impossible in the absence of an applied voltage. The additional assumptions E2g − E1g ≫ μ0 and E0g − E−1g ≪ μ0 should guarantee that double charging and hole transfer, respectively, can be ignored for all applied voltages. Processes of molecular charging and discharge as considered in the following are presented in Figure 2. Those are accounted for by standard expressions for the rates of charging and discharge being all proportional to the common coupling parameter Γ. (See, for example, refs 19 and 20.) The applied voltage shall enter the description by replacing the Fermi energy of the left lead by EF + |e|V/2 and that of the right lead by EF−|e|V/2 (model of a symmetrically applied voltage). As indicated in Figure 2, we will consider a voltage range, where also the excited state of the molecule (being in either its neutral or singly charged state) is populated. Then, photon emission becomes possible. Note that the configuration in the lower panel of Figure 2 has been suggested in literature (direct population of the singly charged molecule’s excited state; cf. refs 9−11). It corresponds to an applied voltage according to |e|V/2 ≥ E1eg + ΔE10 (note ENeg = ENe − ENg). A novel configuration also resulting in electroluminescence is given in the middle panel of Figure 2. Here the excited state of the neutral molecule is populated via discharge that appears at | e|V/2 ≥ E0eg − ΔE10. If ΔE10 is much smaller than E0eg, then photon emission may start at an applied voltage of 3 ... 4 V. This voltage can become smaller if ΔE10 comes into the range of E0eg. Because charging of the molecule starts at a voltage according to |e|V/2 = ΔE10 we shall search for systems with ΔE10 ≈ E0eg/2. Then, we obtain the voltage range to observe electroluminescence at V ≈ E0eg/|e|, that is, a value possibly below 2 V.
II. MODEL OF THE MOLECULAR JUNCTION Different types of molecules have been used so far in the experiments on single-molecule electroluminescence. Molecular junctions with tetraphenyl porphyrins were described in refs 10−12, and measurements on junctions with magnesium porphine were reported in ref 9. The experiments of ref 13 have been based on ethyne−bridged (porphinato)zinc(II)oligomers, and in ref 4, emitted photons of the fluorescence dye 2.5 Cy3 could be detected. Related basic electronic excitation energies extend from ∼1.59,13 to 1.9 eV.10−12 To cover the variety of these molecules and also to study different energetic configurations of the molecule lead−plasmon system we do not focus on a junction with a particular molecule. Instead, a parametrized model is introduced, which offers the freedom to vary some basic parameters easily (cf. Table 1). Such an approach, however, requires assumptions related to more subtle quantities introduced below (for example the relative charging energy or the different types of reorganization energies). For the spherical leads, we take Au nanoparticles with a diameter of 20 nm and with a dipole plasmon excitation energy around 2.6 eV. (Concerning the life−time broadening γpl and the transition dipole moment dpl, see also Table 1.) In line with the studies reported in ref 9, our focus is on sequential charge transmission valid in the case of weak molecule−lead electron 25963
dx.doi.org/10.1021/jp309987c | J. Phys. Chem. C 2012, 116, 25962−25968
The Journal of Physical Chemistry C
Article
Because the molecular excitation energies ENeg should be in resonance to the plasmon excitation energy Epl of the spherical leads, molecular de-excitation due to plasmon excitation will also take place. The molecule−lead interaction responsible for this excitation energy exchange is given as a coupling between the molecular transition dipole (it shall point in the x direction, see Figure 1) and the MNP transition multipoles.16 In contrast, the left lead X = L interacts with the right lead X = R via all transition multipoles. As demonstrated in ref 16 for a spherical Au nanoparticle of 20 nm diameter, the molecule−MNP interaction is dominated by the dipole−dipole coupling part if the molecule is placed 2.5 nm (or more) apart from the MNP surface. We will concentrate on similar molecule−MNP distances in the subsequent studies. The MNP transition dipole moments read as dXI = dpl eXI with amplitude dpl. I = x,y,z labels the three different dipole plasmons, and the eXI are respective unit vectors of a Cartesian coordinate system (referring to lead X = L,R). To stay simple, the plasmon energy exchange coupling between the two leads will be described in the same approximation resulting in the coupling VLI,RI′. It follows the formation of MNP dipole plasmon hybrid levels.16,21 Only certain elements of VLI,RI′ remain finite. We have: VLx,Rx = −2Vpl, VLy,Ry = VLz,Rz = Vpl (with Vpl = d2pl/X3lead, where Xlead is the distance between the centers of mass of the two spherical leads). Obviously, plasmon excitations of different orientation are decoupled, but those of the same orientation may be hybridized: Exφ=± = Epl ± 2Vpl and Eyφ=± = Ezφ=± = Epl ± Vpl. These formulas offer a qualitative correct picture for the following discussion. Because the molecular transition dipole moment points in the x direction, it couples to the hybrid state with energy Ex−. Assuming a lead surface-to-surface distance ΔX of 5 nm Ex− amounts to 1.925 eV, and for ΔX = 2 nm, we get Ex− = 1.610 eV. Accordingly, it becomes possible to come into resonance with the excited states of those molecules previously mentioned. Fast nonradiative plasmon decay is accounted for by the rate 2γpl (two times the dephasing rate, cf. Table 1). The radiative decay is considered according to 2γrad = 4ω3pld2pl/3πc3ℏ (it amounts to ∼30% of the nonradiative decay rate; ωpl = Epl/ℏ). To include molecular vibrations, we concentrate on a single active vibrational coordinate. This is confirmed by experiments that indicate the dominance of one vibrational mode when charging or discharge of the molecule takes place (different type of copper phthalocyanine).22 The single-coordinate Q should undergo harmonic vibrations; that is, the potential energy surfaces involved take the form UNa (Q) = U(0) Na + ℏωvib(Q − QNa)2/4.23 According to ref 22, it should describe a high-frequency mode. The vibrational energy ℏωvib has been taken in a form independent of the electronic state and between 50 and 100 meV. The QNa are the electronic state-dependent equilibrium values of Q, and U(0) Na ≡ ENa is the energy at the equilibrium configuration (electronic transition energies ENge (0) now take the form U(0) Na − UNg ). Reorganization energies related to intramolecular electronic transitions follow as λMa,Nb = ℏωvib/ 4 × (QMa − QNa)2. They may refer to charging (M = 1, N = 0) or to molecular excitation (M = N, a = e, b = g). To have a pronounced nuclear rearrangement upon charging, we set λ1g,0g = 450 meV while the λ0e,0g and λ1e,1g related to electronic transition are set equal to 50 meV. (See also Table 1.) Moreover, the molecule−lead energy exchange coupling has to include vibrational overlap expressions (Franck−Condon factors). Intramolecular vibrational energy redistribution
Figure 2. Energy level scheme of the junction at three different applied voltages V (E1eg = E0eg, vibrational levels not shown, dashed red line: Fermi−energy at V = 0). In the left lead L the energies E0a of the neutral molecule are combined with the energy Eel of an electron below or at the Fermi−edge (blue region). For the right lead, R molecular energies are combined with the energy Fel of an electron above the Fermi−edge (light gray region). The central part M shows the energies E1a of the singly charged molecule. Charging and discharge proceed as horizontal transitions. Upper panel: current formation starts via the electronic ground-state population of the charged molecule (V = 2ΔE10/|e|). Middle panel: population of the neutral molecule’s excited state sets in via discharge (V = 2(E0eg − ΔE10)/|e|). Lower panel: population of the charged molecule’s excited state becomes possible (V = 2(E1eg + ΔE10)/|e|; level E0e not shown). If the molecule is in the excited state, then a de-excitation and subsequent excitation of a hybrid plasmon may also take place (not shown). 25964
dx.doi.org/10.1021/jp309987c | J. Phys. Chem. C 2012, 116, 25962−25968
The Journal of Physical Chemistry C
Article
refer to the electronic states with quantum numbers NA and Ng, respectively. The density matrix elements ρ(N) Cκ,Aμ account for the actual steady-state of the coupled molecule−lead system (including lead−plasmon excitations) when computing Iem (eq 1). The functions σ(N) gμ,Cκ (t;B) correlate the ground state to the various excited states. They are also propagated by the density matrix equations but starting with the initial values δ C , B ⟨χNgν|χNCκ⟩.18,20 The Fourier transformation of the σ functions results in the frequency dependence of the emission. The combination of dipole moments characterizing either the molecular excitation or the lead plasmon excitations guarantees signatures of plasmon hybrid level formation as well as Fanolike structures in the emission. (Accordingly, the respective offdiagonal density matrix elements are essential.)
(IVR) is considered by coupling the single vibrational coordinate to a reservoir of further vibrations that do not couple to the considered electronic transition.20 This coupling is determined by the single value J = J(ωvib) of a common spectral density (cf. Table 1, it results a lifetime ℏ/2πJ of the first excited vibrational state of ≈0.6 ps).
III. DENSITY MATRIX THEORY AND CURRENT FORMULA All described processes of molecular charging, discharge, IVR, coupling to the lead plasmons, and interlead coupling are accounted for by a properly formulated density matrix theory.19,20 Here we concentrate on the steady-state of the system that is achieved if a certain voltage has been applied. Because coherences among different charging states are of no importance, the density matrix is diagonal with respect to the charging number. The density matrix elements ρ(N) gμ,gν describe the molecule as well as the leads in their ground state. (Different vibrational states with quantum numbers μ and ν are possible.) Excited-state density matrix elements are ρ(N) Aμ,Bν. The quantum numbers A and B label either the excited state of the molecule (which is in the charged state with charging number N) or the dipole plasmon excitations I of lead X. If A = e and B = XI, then the density matrix is responsible for molecule−lead coupling, that is, respective excitation energy exchange processes. Interlead coupling and hybrid level formation is accounted for by density matrix elements with, for example, A = (N) LI and B = RI. The coherences ρAμ,gν characterize photoemission of the coupled molecule−lead system. (See the next section; the molecular part decays according to the dephasing rate γmol and the MNP part according to γpl.) (0) The diagonal density matrix elements ρaμ,aν and ρ(1) aμ,aν enter the current formula by defining the partial currents IX,0→1 and IX,1→0, respectively. Both describe charge motion between lead X and the molecule. IX,0→1 is responsible for charging and IX,1→0 for discharge.19,20 The net current moving from lead X into the molecule follows as IX(t) = IX,0→1(t) + IX,1→0(t), and the steady−state current I is I = IL = −IR. (It is obtained by propagating the density matrix in the presence of an applied voltage up to a time where its elements do not change.)
V. RESULTS We start the presentation of results by considering a molecule with small reorganization energies (λMa,Nb ≈ 0). This allows us to discuss the simple case where any vibrational contribution is absent. Moreover, we take the small value of 50 meV for the relative charging energy ΔE10. To have reference data at hand, we first focus on the junction where the dipole plasmon resonance is out of the range of molecular excitation energies (junction type shown in the upper panel of Figure 1). Figure 3
Figure 3. IV−characteristics of the junction with pyramidal leads (absence of a resonant coupling of the lead plasmons to the molecule, cf. Figure 1). The different current plateaus correspond to the different voltage regimes of current formation, as discussed in Figure 2 (note the cut in the voltage axis).
IV. EMISSION SPECTRUM The steady-state emission spectrum is introduced as F(ω) = 4ω3Iem (ω)/3πc3ℏ, where ∫ dω F(ω) gives the total probability of photon emission per time (inverse radiative life−time of the junction excitations).18,20 The emission line shape follows as
displays the IV characteristics, and the steady-state emission spectrum can be found in Figure 4. (Note the additional assumption E1eg = E0eg.) Because of the chosen small value of ΔE10, current starts to flow at 0.1 V, which corresponds to the junction energy level configuration of the upper panel in Figure 2 (charging via the electronic ground state). If the applied voltage is further increased, then the steady-state current moves to a second plateau (cf. middle panel of Figure 2). Here the neutral molecule’s excited state becomes populated and photon emission appears. A closer inspection of the molecular level populations PNa reveals that P0e ≈ 0.24 but P1e ≈ 0.08. So, emission from the charged molecule also contributes. The last plateau of the IV characteristics corresponds to a junction energy level configuration of the lower panel of Figure 2 (charging also via the excited electronic state). We have P1e = P0e = 0.25. The molecule in its neutral and single charged state
Iem(ω) = Re(∑ [dAd*B] ∑ ∑ ⟨χNAμ |χNgν ⟩ A ,B
×
∑∫ C ,κ
0
∞
N
μ,ν
dt e−iωt σg(νN, C) κ ,(t ; B) × ρC(κN,)Aμ ) (1)
The dA and dB are transition dipole moments that are responsible for molecular transitions (dmolex) or lead transitions (dXI). Their products [dAd*B ] are weighted by the density matrices σ and ρ. (To keep it simple, dmol does not depend on the molecular charging.) Emission from higher lead multipole excitations contributes but can be ignored in this first attempt. The χNAμ and χNgν forming the overlap expression are (reaction coordinate) vibrational wave functions (with vibrational quantum numbers μ and ν). The vibrational wave functions 25965
dx.doi.org/10.1021/jp309987c | J. Phys. Chem. C 2012, 116, 25962−25968
The Journal of Physical Chemistry C
Article
Figure 4. Steady-state emission spectra of the molecular junction formed by two spherical leads versus frequency (related to the lower plasmon hybrid energy Ex−). Different detunings between the molecular excitation energy E0eg and Ex− are taken (applied voltage 4.1 V). Upper left panel: Spherical MNP surface-to-surface distance ΔX = 5 nm, E1eg = E0eg, curves from background to foreground: Ex− − E0eg = 0, 20, 50, and 100 meV, the foremost red filled curve shows the emission (increased by ∼1500) for the junction with pyramidal leads (absence of a resonant coupling of the lead plasmons to the molecule). Upper right panel: ΔX = 2 nm, E1eg = E0eg, curves from background to foreground: Ex− − E0eg = 0, 20, 50, 100, and 200 meV. Lower panel: ΔX = 5 nm, E1eg − E0eg = 100 meV, curves from background to foreground: Ex− − E0eg = 50, 70, 100, 150, and 250 meV.
structure (lower panel of Figure 4). A similar change of the detuning as discussed beforehand results in an intermediate broadening of the upper peak and finally also shows a separate signature of the plasmon hybrid level emission. Emission spectra that display vibrational satellites are presented in Figure 5. To achieve a pronounced steady-state population of excited vibrational levels if a current flows through the molecule, we use rather large values of reorganization energies (λ1g,0g = 450 meV, λNe,Ng = 50 meV).19 The applied voltage shall realize a direct population of the excited electronic level of the charged molecule. Therefore, its energy Ele + ℏωvib(1/2 + μ) has to be compared with EF + |e|V/2 + E0g + ℏωvib/2 (cf. the lower panel of Figure 2). According to the chosen voltage of 4.4 V, the energy range between the fifth and sixth excited vibrational state is directly addressed by charge injection. For reference, the foremost curve of the lower panel of Figure 5 shows the emission in the absence of lead plasmon excitations. The large applied voltage and the subsequent population of excited vibrational levels results in a number of vibrational satellites. (We draw the emission versus ℏω − E0eg, so the so-called 0−0−transition is positioned at 0.) This emission curve has been enlarged by a factor of 4000 to be comparable with those spectra including the coupling to the lead plasmons. It can be directly compared with the background curve in the upper panel of Figure 5 (case Ex− = E0eg). Those emission lines are mostly enhanced, which are close to the resonance with the plasmon hybrid level. Such a local enhancement moves through the vibrational progression if E0eg is decreased (increasing detuning between molecular excitation and plasmon hybrid level).
contributes to the emission spectrum (see upper panel, foremost curve of Figure 4). Next, we discuss the spectra obtained for the case where the molecular excitation is in resonance with the hybrid plasmon excitation of the spherical leads. (See Figure 4; note that for ΔX = 5 nm, we get Ex− = 1.925 eV.) The molecular excitation energy E0eg = E1eg is stepwise moved downward out of the lower plasmon hybrid resonance Ex−. This displacement goes along with a narrowing and intermediate increase in the emission. In any case, the emission intensity is more than three orders of magnitude larger than the emission without plasmon excitation indicating a huge enhancement effect. For ΔX = 2 nm (Ex− = 1.610 eV) and in the resonant case E0eg = Ex−, a splitting of the emission line is observed, reflecting a splitting of the molecular excitation and the plasmon hybrid level. For larger detunings between E0eg and Ex−, a weak plasmon emission remains visible. Interestingly, the smaller value of ΔX does not result in an increased emission intensity. (The stronger molecule plasmon coupling leads to smaller excited level populations of the molecule.) The related IV−characteristics at E0eg = Ex− (not shown) are only slightly affected by the coupling to the lead plasmons. (The second plateau of Figure 3 is somewhat reduced in contrast with P1e and P0e, which are about one order of magnitude smaller.) The intermediate maximum of the emission peaks results from an interplay of line narrowing, which increases the peak height and excitation energy transfer as well as increases detuning, which both decrease the peak height. If we let the excitation energy of the molecule in its neutral state differ from that in the singly charged state (here E0eg < E1eg), then the emission spectrum displays a double-peak 25966
dx.doi.org/10.1021/jp309987c | J. Phys. Chem. C 2012, 116, 25962−25968
The Journal of Physical Chemistry C
Article
Figure 7. Steady-state emission spectra of the molecular junction formed by two spherical leads versus frequency (related to the lower plasmon hybrid energy Ex−). Case of large relative charging energy ΔE10 = 1 eV (applied voltage 2.2 V, ΔX = 5 nm, E1eg = E0eg). Different detunings between the molecular excitation energy E0eg and Ex− are taken. Curves from background to foreground: Ex− − E0eg = 0, 20, 50, and 100 meV; the foremost red filled curve shows the emission for the junction with pyramidal leads (increased by ∼1500).
Figure 7. They look similar to the spectra displayed in Figure 4, but there a voltage two times that used here has been applied (4.4 V; be aware that the emission intensity is also three times larger).
Figure 5. Steady-state emission spectra (related to E0eg) as in the upper left panel of Figure 4 but for finite molecular reorganization energies (see text; applied voltage 4.4 V). Upper panel: from background to foreground Ex− − E0eg = 0, 25, and 75 meV. Lower panel: from background to foreground Ex− − E0eg = 125 and 225 meV. The foremost red filled curve gives the emission spectrum in the absence of lead plasmon excitations (increased by a factor of 4000).
VI. CONCLUSIONS The electroluminescence due to the current formation through a molecular junction has been discussed. The emission signal can be enhanced more than three orders of magnitude if the molecular transition comes into resonance with plasmon excitations of the leads. This has been demonstrated for a junction formed by two spherical Au leads of 20 nm diameter with a single molecule in between. The emission enhancement appears for small as well as rather large molecular reorganization energies and for a chosen lead surface-to-surface distance of 5 nm. In this configuration, we expect fewer effects of so-called spill-out electrons that enter the outer region of the spherical leads as a result of plasmon decay.24 Lead plasmonenhanced molecular electroluminescence can be observed in a low voltage region if the molecular excited-state population is achieved via discharge. Moreover, the relative charging energy has to lie in the range of half of the molecular electronic excitation energy.
The foregoing examples are based on a rather small value of the relative charging energy ΔE10. Next, we increase this quantity to 1 eV and expect photoemission at voltages according to |e|V/2>E0eg − ΔE10 = 0.925 eV (note E0eg = Ex− and ΔX = 5 nm). The resulting voltage is somewhat smaller than the one responsible for molecular charging (|e|V/2 + ΔE10). Accordingly, IV−characteristics (in the absence of lead plasmon excitation) display a nonvanishing current for an applied voltage larger than 2 V (cf. Figure 6; photoemission for this case is shown in the foreground of Figure 7). The spectra of electroluminescence at an applied voltage of 2.2 V and including the lead plasmon enhancement effect are shown in
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS Financial support by the China Scholarship Council (Y. Zhang) and by the Deutsche Forschungsgemeinschaf t through Sfb 951 (Y. Zelinskyy) is gratefully acknowledged.
■
REFERENCES
(1) Halas, N. J. Nano Lett. 2010, 10, 3816. (2) Odom, T. W.; Schatz, G. C. Chem. Rev. 2011, 11, 3667. (3) Bek, A.; Jansen, R.; Ringler, M.; Mayilo, S.; Klar, T. A.; Feldmann, J. Nano Lett. 2008, 8, 485.
Figure 6. IV characteristics of the junction with pyramidal leads (absence of a resonant coupling of the lead plasmons to the molecule, cf. Figure 1) and for a large relative charging energy ΔE10 of 1 eV. 25967
dx.doi.org/10.1021/jp309987c | J. Phys. Chem. C 2012, 116, 25962−25968
The Journal of Physical Chemistry C
Article
(4) Ringler, M.; Schwemer, A.; Wunderlich, M.; Nichtl, A.; Kürzinger, K.; Klar, T. A.; Feldmann, J. Phys. Rev. Lett. 2008, 100, 203002. (5) Introducing Molecular Electronics; Cuniberti, G., Fagas, G. F., Richter, K., Eds.; Lecture Notes in Physics 680; Springer: New York, 2005; p 1. (6) Selzer, Y.; Allara, D. L. Annu. Rev. Chem. 2006, 57, 593. (7) Galperin, M.; Ratner, M. A.; Nitzan, A. J. Phys.: Condens. Matter 2007, 19, 103201. (8) Zimbovskaya, N. A.; Perderson, M. R. Phys. Rep. 2011, 509, 1. (9) Wu, S. W.; Nazin, G. V.; Ho, W. Phys. Rev. B 2008, 77, 205430. (10) Tian, G.; Liu, J.-C.; Luo, Y. Phys. Rev. Lett. 2011, 106, 177401. (11) Zhang, C.; Zhang, R.; Jiang, S.; Zhang, L.; Gao, H. Y.; Zhang, X. L.; Chen, L. G.; Liao, Y.; Dong, Z. C. Appl. Phys. Lett. 2012, 100, 073111. (12) Schneider, N. L.; Berndt, R. Phys. Rev. B. 2012, 86, 035445. (13) Banerjee, P.; Conklin, D.; Nanayakkara, S.; Park, T.-H.; Therein, M. J.; Bonnell, D. A. ACS Nano 2010, 4, 1019. (14) Mangold, M. A.; Calame, M.; Mayor, M.; Holleitner, A. W. J. Am. Chem. Soc. 2011, 133, 12185. (15) Mangold, M. A.; Calame, M.; Mayor, M.; Holleitner, A. W. ACS Nano 2012, 6, 4181. (16) Zelinskyy, Y.; Zhang, Y.; May, V. J. Phys. Chem. A 2012, 116, 11330−11340. (17) Zelinskyy, Y.; May, V. Nano Lett. 2012, 12, 446. (18) Zhang, Y.; Zelinskyy, Y.; May, V. J. Nanophotonics 2012, in press. (19) Wang, L.; May, V. Chem. Phys. 2010, 375, 252. (20) Wang, L.; May, V. Phys. Chem. Chem. Phys. 2011, 13, 8755. (21) Nordlander, P.; Oubre, C.; Prodan, E.; Li, K.; Stockman, M. I. Nano Lett. 2004, 4, 899. (22) Qiu, X. h.; Nazin, G. V.; Ho, W. Phys. Rev. Lett. 2004, 92, 206102. (23) May, V.; Kühn, O. Charge and Energy Transfer Dynamics in Molecular Systems; Wiley-VCH: Weinheim, Germany, 2011. (24) Huschka, R.; Zuloaga, J.; Knight, M. W.; Brown, L. V.; Nordlander, P.; Halas, N. J. J. Am. Chem. Soc. 2011, 133, 12247.
25968
dx.doi.org/10.1021/jp309987c | J. Phys. Chem. C 2012, 116, 25962−25968