Result and discussion - 有機金屬螯合物之時間解析光激螢光研究

4-1 Power dependence of photoluminescence

We measured the PL spectra and TRPL under different pump fluence from 4.02 kW/cm² to 80.46 kW/cm². The PL peak intensities of the four samples with gold and without gold show the linear relations to the excitation power as shown in Fig. 4-1, 4-2, 4-3, and 4-4, indicating that there is no thermal heating effect on the samples due to the excess pump fluence or amplified spontaneous emission (ASE) phenomenon.

As the pump fluences increase, the samples with gold film have a lager slope than the ones without gold film, and the change of slope implies that the samples are under the surface plasmon resonance of gold. The slopes of the samples without and with gold are changed from 14.6, 17.5, and 28.5 to 30.6, 25.2, and 30.7 for sample A, B, and C respectively. For sample D, the slope changes of the first and the second peaks are 6.8 to 16.3 and 4.2 to 13.6, respectively. The ratios of the two slopes for sample A, B, and C are 2.1, 1.4, and 1.1, respectively, and those for the first and second peak of sample D are 2.4 and 3.3. The largest change in slope for sample D indicates that sample D has the largest PL enhancement resulting from surface plasmon coupling with gold film, especially the second peak of sample D.

Samples C and D contain the same organic material, but sample D shows a larger slope change than sample C. This result shows that the slope change is not related to the PL enhancement ratio, but is related to the metal structure of surface plasmon resonance and will be discussed later.

Fig. 4-1 PL peak intensity against pump fluence of sample A.

Fig. 4-2 PL peak intensity against pump fluence of sample B.

Fig. 4-3 PL peak intensity against pump fluence of sample C.

Fig. 4-4 PL peak intensity against pump fluence of sample D (a) first peak, (b) second peak.

When the carrier recombination process in organic layer starts, the energy from the carrier recombination process will generate photons, moreover, for the sample that is coated with metal, if the interaction of photons and the surface plasmons matches the Momentum Conservation Principle condition, the photons will be absorbed by the surface plasmons and exist in the form of surface plasmon polaritons. That is, the energy of the external electromagnetic field will be absorbed by surface plasmon, resulting in the electromagnetic enhancement of the resonance-influenced area. After the surface plasmon resonance, the enhanced surface plasmons can be turned into photons as light when matching the Momentum Conservation Principle again. So there are two ways to generate the photons when coupling with the metal and comes out the PL enhancement. So the emission enhancement of the organic dye on gold film can be attributed to excited organic dye molecules coupling to the electron vibration energy of surface plasmon. As shown in Fig. 4-5, a surface plasmon is produced during the molecular relaxation process and leads to an increase in the spontaneous recombination rate. Excitons are generated in organic layer by optical- or electrical-excitation. For the sample that is not coated with metal, theses excitons are terminated by the radiative or nonradiative recombination rates. For the sample that is coated with metal, when the bandgap energy ( ) of organic dye is close to the electron vibration energy of surface plasmon ( at the metal-semiconducting interface, the energy can transfer to the surface plasmon with surface plasmon coupling rate ( . PL decay rates are enhanced by surface plasmon coupling rate, as values are expected to be very fast.

Large density of states from the surface plasmon dispersion diagram shown in Fig. 4-6 introduces high electromagnetic fields and increases . In this figure, surface plasmon enhancement method can be applied in the range of visible to ultraviolet depending on the metal chosen, where aluminum is for UV, silver for the

blue region, and gold for the red region. We chose the gold for surface plasmon enhancement because the surface plasmon resonance wavelength of gold ranges from about 500nm to 700nm so that is suited for the PL wavelength range of our samples.

If the metal surface is perfectly flat, the surface plasmon energy will dissipate thermally because the plasmon wave is a nonpropagating evanescent wave. So by providing rough or nanostructure metal layer allows the SP energy be extracted as light and make SPs of high momentum to scatter, lose momentum, and couple to radiative light^[25].

Fig. 4-5 Physical mechanism of the electron-hole recombination and surface plasmon coupling.

Fig. 4-6 Surface plasmon dispersion diagram^[26], interfaced with Au (dash line), Ag (solid line), and Al (dotted line).

The main peak originated from the host emitter and a shoulder (side peak) locates at 628nm on the PL spectra are illustrated in Fig. 4-7, 4-8, 4-9, and 4-10.

Except sample D, the second peak of which is enhanced by the surface plasmon resonance, the PL spectra of samples A, B, and C increase uniformly over the whole wavelength. The main PL peak and the shoulder peak at 628nm may be attributed to the intermolecule and intramolecule emission branching^[27]. Intramolecule singlet excitons can be generated by the absorption of photons or the carrier injection followed by electron-hole pair formation on a single molecule; and the intermolecule excitation can be formed by resonantly interacting between two molecules in different sites. So the first emission peak of the PL spectra is the intramolecule process of the host materials and the second peak is due to the intermolecule process of the energy transfer mechanism from the host emitter to DCM. Two different molecules, one neutral and the other in the excited state, can form an intermolecule excitation as an excimer. Excimer shows a long radiative lifetime, which is because the transition from the excimer state to the ground state is usually forbidden by symmetry, and this is consistent with the TRPL measurement for sample D (see Fig. 4-30). Because of the SPR, we could observe the red shift of the PL spectrum with respect to the center wavelength of the host dyes. The shift range of the main peak are from 555nm to 560nm, and 591nm to 600nm for samples A and D respectively, but samples B and C do not show red shift at the main peak because the main peak at 490nm for sample B is not in the surface plasmon resonance wavelength of gold and sample C does not show much enhancement by surface plasmon resonance. The shift of spectra is clearer for PL spectra measured with and without gold as shown in Fig. 4-15 to 4-18.

Fig. 4-7 Power-dependent PL of sample A (a) without gold film, (b) with gold film.

628nm

Fig. 4-8 Power-dependent PL of sample B (a) without gold film, (b) with gold film.

628nm 628nm

Fig. 4-9 Power-dependent PL of sample C (a) without gold film, (b) with gold film.

628nm

Fig. 4-10 Power-dependent PL of sample D (a) without gold nanoplates, (b) with gold nanoplates.

628nm

The PL enhancement ratio as a function of pump fluence is plotted in Fig. 4-11 to 4-14 for samples A-D. It can be observed that the enhancement ratios of all the samples start to get saturated at pump fluence of ~48 kW/cm². And the results in the following sections are measured under this saturation pump fluence. While sample C shows nearly no enhancement, sample B shows a large enhancement at 550 nm, instead of its main peaks at 490nm. Sample D not only shows the largest overall enhancement, but also has a large enhancement at its second peak (shoulder) rather than at its main peak. The difference of the enhancement ratio between sample C and D is attributed to the structure of the metal^[28]. For small nanostructure such as nanorods or nanoparticles, which have the diameter or side length shorter than 500nm, the scattering cross sections are small due to the smaller volumes in these kinds of structures, so that the surface plasmon coupling is not significant. But for the case of large nanocrystals such as hexagonal or triangular plates (500nm~few μm of side length), it is necessary to mainly concern the wavevector matching condition of plasmons in gold structure and the scattered photons. However, the surface plasmon energy can be extracted as light only when appropriate wavevector matching conditions are met. Compared with individual gold nanostructures with finite sizes, the wavevector matching condition of plasmons with outcoupling photons in continuous gold structures (gold film) is more stringent. Thus in the coupling with Alq₃:DCM dye, the gold nanoplates (about 5~10μm of side length) structure shows the better wavevector matching condition than the gold film resulting in a higher enhancement ratio and a larger scattering cross section although the nanoplates are much smaller than film.

According to the measurement results of sample C and the description above, Alq3:DCM dye may not have apparent surface plasmon enhancement coupling with gold film. Sample D has the largest enhancement among the four samples and the

second peak increases considerably, while the shoulder peak of samples A, B, and C does not show obviously sharp peak after the enhancement by the surface plasmon resonance (Fig. 4-18). So we can suppose that the intermolecule process of sample D gets larger surface plasmon resonance enhancement coupling with gold nanoplates.

Fig. 4-11 Enhancement ratio against pump fluence of sample A.

Fig. 4-12 Enhancement ratio against pump fluence of sample B.

Fig. 4-13 Enhancement ratio against pump fluence of sample C.

Fig. 4-14 Enhancement ratio against pump fluence of sample D (a) first peak, (b) second peak.

Fig. 4-15, 4-16, 4-17, and 4-18 show the PL spectra of dyes measured with gold and without gold at pump fluence of 48 kW/cm². Samples A, B, and D have the considerable enhancement at the main PL peak while PL enhancement of sample B occurs at 550nm. The PL enhancements by surface plasmon resonance are 1.81 times, 1.34 times, and 1.87 times (first peak) 2.15 times (second peak) for sample A, B, and

D respectively. For sample C, however, there is only 1.06 times of enhancement at the main PL peak. The PL spectra of sample A, C, and D cover the spectral range from 500nm to 700nm while that of sample B covers from 400nm to 700nm. The PL enhancement of sample B starts from 510nm, the middle of the spectrum (Fig. 4-16), so there is nearly no enhancement at the first half of sample B and the enhancement occurs at the latter half of the spectrum. Besides, sample A and sample D also get the overall enhancement start from ~500 nm and it is in agreement with the surface plasmon resonance wavelength of gold and the surface plasmon dispersion diagram (Fig. 4-6). These results suggest that the enhancement is due to the surface plasmon resonance coupled with gold.

It is very interesting that the sample B has a large enhancement at 550nm instead of the overall enhancement. The lifetime of sample B is at the order of nanoseconds implying that this peak cannot be the phosphorescence because the lifetime of the phosphorescence is usually a few microseconds or even milliseconds. As a matter of fact, the enhancement ratio of sample B is similar to those of other samples that follow the process of being enhanced gradually from the beginning and then reached to the maximum enhancement point and then gradually decayed to the original state.

That is, the spectral ranges of the other three samples match gold’s surface plasmon resonance wavelength so that get the overall enhancement. But for sample B, the enhancement of sample B is limited by the gold’s surface plasmon resonance wavelength ranging from about 510nm to 700nm, and the maximum enhancement point of gold is located at nearly 550nm. So sample B is enhanced just from the middle of its spectrum not from the beginning (400nm), leading us to think that there is a sudden enhancement appears at 550nm.

Fig. 4-15 PL spectra comparisons of dyes with gold and without gold of sample A at pump fluence 48 kW/cm².

Fig. 4-16 PL spectra comparisons of dyes with gold and without gold of sample B at pump fluence 48 kW/cm².

Fig. 4-17 PL spectra comparisons of dyes with gold and without gold of sample C at pump fluence 48 kW/cm².

Fig. 4-18 PL spectra comparisons of dyes with gold and without gold of sample D at pump fluence 48 kW/cm².

The enhancement ratio and the peak shift of PL excited by a continuous (CW) laser are far different from those excited by a pulse laser, as shown in Fig. 4-19, 4-20, 4-21, and 4-22. There are at least 2.2 times of PL enhancements for sample A, B, and D at the main PL peak, while enhancement in sample B by the surface plasmon resonance occurs at 550nm. The PL enhancements by surface plasmon resonance are 8 times, 2.2 times, and 6 times for sample A, B, and D respectively. Sample C, which gets nearly no enhancement in pulsed laser measurement, shows ~1.2 times of enhancement at the main peak in CW measurement. Although the spectral dependence is similar, SPR induced PL enhancement is more significant under CW laser excitation..

Surface plasmon-induced enhancement under the CW pump leads to the shift of the spectra. Sample D shows the red shift as much as ~30 nm, but sample A shows slight blue shift. Meanwhile, samples B and C do not show any peak shift at the main peak since the spectra is not influenced by the surface plasmon resonance of gold.

Fig. 4-19 PL spectra comparisons of dyes with gold and without gold of sample A measured by CW 405nm diode laser. (From Prof. S. Gwo’s lab in National Tsing Hua University.)

Fig. 4-20 PL spectra comparisons of dyes with gold and without gold of sample B measured by CW 405nm diode laser. (From Prof. S. Gwo’s lab in National Tsing Hua University.)

Fig. 4-21 PL spectra comparisons of dyes with gold and without gold of sample C measured by CW 405nm diode laser. (From Prof. S. Gwo’s lab in National Tsing Hua University.)

Fig. 4-22 PL spectra comparisons of dyes with gold and without gold of sample D measured by CW 405nm diode laser. (From Prof. S. Gwo’s lab in National Tsing Hua University.)

In order to understand the emission dynamics, we measured the time-resolved PL of each sample. Fig. 4-23, 4-24, 4-25, and 4-26 illustrate the results of TRPL of samples A, B, C, and D measured at different pump intensity. There were two slopes of the decay paths, which indicated that the effective lifetimes was determined by double exponential fitting, , where and and

and represent the PL intensities and decay time constants of the fast and slow components, respectively. While PL signals of samples A, B, and C undergo double exponential decays, that of sample D shows a single exponential decay curve, .

At low pump fluences, PL signals of sample A, B, and C without gold film decay with a single time constant. As the pump fluence increases, double exponential decay becomes obvious. For samples A and B with gold film, PL signals show two-step decay for all pump fluence, but that of sample C with does not change the radiative decay path. Figures 4-23, 4-24, 4-25, and 4-26 show that the fast decay paths of samples A and B with gold film show much steeper slope than the ones without gold film and it implies that the decay lifetime become faster due to the Purcell effect. But for samples C and D, the changes of the decay times are not obvious because sample C does not get much surface plasmon enhancement and sample D has exhibited fast decay rates (see Fig. 4-30) so that is not strongly affected by the Purcell effect, although the PL of sample D is enhanced by surface plasmon resonance. So samples A and B show the stronger Purcell effect than samples C and D.

Fig. 4-23 TRPL measurement of sample A (a) without gold film, (b) with gold film.

Fig. 4-24 TRPL measurement at 550nm of sample B (a) without gold film, (b) with gold film.

Fig. 4-25 TRPL measurement of sample C (a) without gold film, (b) with gold film.

Fig. 4-26 TRPL measurement of sample D (a) first peak without gold nanoplates, (b) first peak with gold nanoplates, (c) second peak without gold nanoplates, (d) second peak with gold nanoplates.

The relation between the power and the time constants is shown in Fig. 4-27, 4-28, 4-29, and 4-30. The excitation power and the decay time are in linear relationship, implying that the high pump fluence can make the recombination rate increase. No saturation at high pump power indicates no heating phenomenon on the samples due to the excess pump fluence or amplified spontaneous emission phenomenon.

In the figures, the black lines with triangles are for samples without gold and the red lines with circles correspond to samples with gold, and the time constants of fast and slow decays are clearly separated. For all samples with or without gold film, the slow decay has the time constants of the order of tens of ns. For the fast decay parts, the time constants are of the order of 1 or 2 ns. For both fast and slow decay parts, the time constants with gold (red lines) are faster than those without gold. The range of the changes in time constant among sample A, B, and C is A>B>C and it is consistent with the PL enhancement. Although sample B shows the larger time constants than sample A, but the ratio of change in lifetimes is still smaller than sample A. Sample D changes a little in time constants due to its weak Purcell effect and is inconsistent with the PL enhancement as above mentioned.

Fig. 4-27 Time constant of power-dependent PL of sample A.

Fig. 4-28 Time constant of power-dependent PL at 550nm of sample B.

Fig. 4-29 Time constant of power-dependent PL of sample C.

Fig. 4-30 Time constant of power-dependent PL of sample D.

In the TRPL measurement, the double exponential decay curve is related to the Förster energy transfer mechanism. When doping of a host emitter with a highly emissive guest emitter, Förster energy transfer can take place and result in transfer of the excited-state energy from the host to the guest and subsequent emission from the guest. The Förster energy transfer rate via induced dipole-dipole interactions between organic materials is given by^[29]

eq. (1)

eq. (2)

where is the normalized fluorescence spectrum of the host material, is the normalized optical absorption cross section of the dopant, is the natural radiative lifetime of the host material, and is the mean distance between host and dopant molecules. For efficient energy transfer (i.e., for large) the overlap between and must be large, as in the case of Alq₃ and DCM [Fig. 4-31 (a) and (b)]. In Fig. 4-31 (c), the energy transfer process illustrates that the pump energy is absorbed by Alq₃molecule and is subsequently transfers to the DCM molecule. This exciton then recombines and emits a photon.

In double exponential decay curve, the two lifetime components can be attributed to different physical processes^[30]. The fast decay path represents the host molecule excitations in close proximity to the nearest guest molecules which nonradiatively (Förster-type) transfer their energy to them. The slow decay path may represent the host excitations that are further away from the guest molecules and, depending on the intermolecule distance between the host and guest molecules, either migrate within the host before being transferred to the guest or decay radiatively and/or nonradiatively. Both processes contribute to the energy transfer rate. The significant

contribution from the slow components to the total lifetime suggests that some excitations remain on the hosts and radiatively decay there, leading to the emission of the host. So this is consistent with the PL spectral (Figs. 4-7 to 4-9) that the main peaks originate from the host material and the shoulders or second peaks fixed at 628nm are due to the Förster energy transfer from the host to the guest.

在文檔中有機金屬螯合物之時間解析光激螢光研究 (頁 36-74)