Morphology effects on spin-dependent transport and recombination in polyfluorene thin films

We have studied the role of spin-dependent processes on conductivity in polyfluorene (PFO) thin films by preforming continuous wave (cw) electrically detected magnetic resonance (EDMR) spectroscopy at temperatures between 10 K and room temperature using microwave frequencies between about 1 GHz and 20 GHz, as well as pulsed EDMR at the $X$ band (10 GHz). Variable frequency EDMR allows us to establish the role of spin-orbit coupling in spin-dependent processes whereas pulsed EDMR allows for the observation of coherent spin motion effects. We used PFO for this study in order to allow for the investigation of the effects of microscopic morphological ordering since this material can adopt two distinct intrachain morphologies: an amorphous (glassy) phase, in which monomer units are twisted with respect to each other, and an ordered (β) phase, where all monomers lie within one plane. In thin films of organic light-emitting diodes, the appearance of a particular phase can be controlled by deposition parameters and solvent vapor annealing, and is verified by electroluminescence spectroscopy. Under bipolar charge-carrier injection conditions, we conducted multifrequency cw EDMR, electrically detected Rabi spin-beat experiments, and Hahn echo and inversion-recovery measurements. Coherent echo spectroscopy reveals electrically detected electron-spin-echo envelope modulation due to the coupling of the carrier spins to nearby nuclear spins. Our results demonstrate that, while conformational disorder can influence the observed EDMR signals, including the sign of the current changes on resonance as well as the magnitudes of local hyperfine fields and charge-carrier spin-orbit interactions, it does not qualitatively affect the nature of spin-dependent transitions in this material. In both morphologies, we observe the presence of at least two different spin-dependent recombination processes. At room temperature and 10 K, polaron-pair recombination through weakly spin-spin coupled intermediate charge-carrier pair states is dominant, while at low temperatures, additional signatures of spin-dependent charge transport through the interaction of polarons with triplet excitons are seen in the half-field resonance of a triplet spin-1 species. This additional contribution arises since triplet lifetimes are increased at lower temperatures. We tentatively conclude that spectral broadening induced by hyperfine coupling is slightly weaker in the more ordered β-phase than in the glassy phase since protons are more evenly spaced, whereas broadening effects due to spin-orbit coupling, which impacts the distribution of $g$-factors, appear to be somewhat more significant in the β-phase.

Figure 1

(a) Illustration of the vertical device stack used in all measurements of this study where the active layer of PFO (blue) is sandwiched between a Ca layer (light gray) and a PEDOT:PSS layer (brown) for electron and hole injection, respectively. (b) Scanning electron microscope image of the active area of a standard device covered with SiN to insulate the ITO except for a small circular opening in the center, defining the active area. The small active device area atop the large substrate allowed for sufficient heat sinking of the power dissipated by the device under operation. (c) Photograph of a device under operating conditions with the blue PFO EL visible. The scale bars in (b,c) both represent 1 mm. The EL spectra for glassy, mixed, and β-phase devices are plotted in (d–f), respectively. Sketches of the polymer chain conformations for the two main phases are shown in their respective panels in order to illustrate the different states of molecular ordering.

Figure 2

Plots of transient changes to a steady-state device current of $I_0=20\mu \mathrm{A}$ after a 400 ns microwave excitation pulse was applied to the samples, as a function of the applied magnetic field which is represented by the vertical axis. The current transients were measured at RT (left) and 10 K (right), on devices containing PFO in the β-phase (a,b), a mixed phase containing both β-phase and glassy phase components (c,d) and the glassy phase (e,f). The insets of the panels display plots of the changes in device current as a function of the applied magnetic field for specific times after the microwave pulse indicated by the red arrows. Qualitatively, a sign reversal of the current change is seen between glassy and β-phases while for the mixed-phase device a sign change occurs between high and low temperature. The data also show that magnetic-resonance-induced current changes in PFO are more than a factor of 2 larger in β-phase PFO compared to glassy phase PFO. The maxima of the current changes occurred for $g$-factors of 2.003 27, 2.003 24, and 2.0033 for the room-temperature data shown in panels (a,c,e), respectively, and for $g$-factors of 2.004 23, 2.004 67, and 2.004 13 for the low-temperature data shown in panels (b,d,f), respectively, with errors below ${10}^{-4}$.

Figure 3

Plots of the measured maximal current change (black squares) after the resonant pulse excitation as a function of applied static field $B_0$ obtained from the data shown in Fig. 2 for β and glassy phases at temperatures of RT (a,b) and 10 K (c,d). The blue lines are fit results with double-Gaussian functions, representing electrons and holes, in which both functions have the same area as required for a pair process. The quality of the fit results is recognized by the fit residuals which are plotted in the insets of the panels. Weak but significant structure is discernible in the residual data sets, indicating that additional spin-dependent processes not described by a double-Gaussian line make minor contributions to the overall EDMR response of the devices. We note that within the given errors, no $g$-factor difference between the two resonance peaks was observed. For each data set, the $g$-factors resulting from both Gaussian functions are described by the values provided in Fig. 2.

Figure 4

Correlation between EL spectrum and the sign of the initial steady-state current change following magnetic resonant excitation at room temperature. The transient of current change is separated into two parts by its zero crossing and integrated to yield areas $A_1$ and $A_2$ as sketched in the top left inset (this example transient is not experimental data). The ratio of $A_1$ to the sum of the moduli $|A_1|+|A_2|$ gives a measure of the sign of the initial part of the transient, either current enhancement or quenching. This ratio is related to the fraction of glassy-phase EL which is derived by deconvoluting the EL spectrum into glassy phase emission and β-phase emission. The fraction of the glassy phase is defined as the ratio between the glassy-phase EL peak intensity $G$ divided by the total EL intensity $\left|G\right|+\left|\beta \right|$, as illustrated in the bottom right inset. The black line is a guide to the eye. The data reveal that the sign of the steady-state current change of PFO OLEDs is governed by the fraction of glassy to β-phase.

Figure 5

Plots of half-field resonance spectra for β-phase (red triangles) and glassy phase (blue circles) devices at RT (a,b) and 10 K (c,d). Each panel shows a change in steady-state current as a function of magnetic field. The fits in (c,d) were calculated using the easyspin matlab toolbox in order to determine the zero-field splitting parameter $D$. A bootstrap error analysis was used to establish the error in the fit parameters. No discernable signal was found for either phase at RT.

Figure 6

Multifrequency continuous wave (cw) EDMR spectra at room temperature obtained by using coplanar waveguide (CPW) resonators. (a) Continuous wave spectra of glassy and β-phase OLEDs, shown in blue and red, respectively. The change in current is shown as a function of both static magnetic field (bottom) and the corresponding magnetic resonance frequency for an uncoupled electron with the vacuum $g$-factor (top axis). Note that cw spectra have a differential line shape compared to the pulsed spectra in Fig. 3. (b) Plot of normalized resonance spectra for different magnetic-field scales as a function of the offset relative to the observed resonance center, obtained from devices with glassy (blue circles) and β-phase (red triangles) PFO. For each phase, the resonances for the lowest and highest frequency (2.33 and 17.89 GHz for the glassy phase; 1.15 and 19.88 GHz for the β-phase) are displayed. The solid black line represents the result of a global fit with multifrequency-dependent double-Gaussian derivatives that model both low- and high-frequency data. All fits reveal the superposition of a broad and a narrow Gaussian. (c) Plots of the widths $\mathrm{\Delta }B$ of the two Gaussians for the two phases as a function of the applied on-resonance magnetic field $B$ based on the fit results obtained from the global fit procedure. The shaded regions represent 95% confidence intervals resulting from the parameter uncertainties that were determined using a bootstrap analysis. The circles and triangles represent the values of the continuous red and blue plots highlighted for the magnetic fields at which experimental spin resonance data were obtained.

Figure 7

(a,d,g,j) Plots of the change in device current after a short, pulsed magnetic resonant excitation at $t=0$ as a function of time (horizontal axis) and pulse length (vertical axis) for both PFO phases at RT (top row) and 10 K (bottom row). The microwave pulse strength for all measurements was ∼560 μT. The current transients are baseline corrected using a second-order polynomial to highlight the Rabi oscillations. Panels (b,e,h,k) show corresponding slices along the dashed white line. Panels (c,f,i,l) show the real component of the Fourier transforms that were calculated using data slices without background subtraction. The Fourier amplitude is plotted as a function of frequency in units of spin-$\frac{1}{2}$ Rabi frequency $\gamma B_1$, where γ denotes the gyromagnetic ratio. Every frequency spectrum exhibits a peak in its signal intensity at the fundamental frequency of $\gamma B_1$ indicating the involvement of paramagnetic states with spin $s=\frac{1}{2}$. The 10 K measurements also show smaller peaks at $2\gamma B_1$, though less visible in the β-phase (f). This second harmonic is indicative of beating of the observed paramagnetic centers, the electron and hole spins.

Figure 8

Plots of electrically detected Hahn echo experiments observed in the time-integrated current of devices with both PFO phases at RT and at 10 K. The pulse sequence used for these experiments is sketched above the plots. Both plots show the integrated current (the charge) as a function of time difference τ–τ′ defined in the sketch of the pulse sequence. This difference was chosen such that the center of the electrically detected spin echoes occurs around τ–τ′ ≈ 0 for better comparison. The solid lines are fits with Gaussian curves and serve as a guide to the eye.

Figure 9

Plots of the decays of the Hahn echo envelopes measured at RT (a) and 10 K (b) as a function of 2τ (defined in the pulse sequence diagram above). (a) Data were recorded from a mixed-phase device (black pentagons), a glassy-phase device (blue circles), and a β-phase device (red triangles). All measured data sets were fitted with single exponential decay functions in order to determine the coherence times $T_2$ of the charge-carrier spin states.

Figure 10

Plots of the results of inversion-recovery experiments for glassy-phase (blue circles) and β-phase (red triangles) PFO OLEDs measured at RT. The time-integrated current (charge) is plotted as a function of mixing time ($T$) that follows the initial π pulse. An electrically detected Hahn echo sequence is used for readout. The EDMR pulse sequence used is shown above the plot. The β-phase OLED was forward biased so that the device current was 50 μA, with a circular active area of 500 μm diameter. The glassy-phase device had a smaller active area with 200 μm diameter. It was operated at a current of $I_0=20\mu \mathrm{A}$.

Figure 11

Plot of the stimulated and electrically detected echo measurements revealing the presence of an electron-spin-echo envelope modulation (ESEEM) at room temperature. The definition of the pulse parameters τ and $T$ is shown in the sketch of the pulse sequence used. Blue and red data points represent results from glassy and β-phase devices, respectively. Panels (a,b) show the stimulated echoes and their decays for on-resonance measurements at a magnetic field of 344 mT. The stimulated echo is measured with a constant mixing time $(T=180\mathrm{ns})$ and delay time τ while ${\tau }^{*}$ is varied (red and blue lines). For (b), $\tau ={\tau }^{*}=96\mathrm{ns}$, with $T$ being varied. In (c) the Fourier transforms of the envelope modulations contained in the data in (b) are plotted, revealing a strong peak at 14.5 MHz for the β-phase device. This frequency corresponds to the proton Larmor frequency for the applied magnetic field. No significant modulation is seen in the glassy-phase device because of an insufficient signal-to-noise ratio.

Figure 12

Histograms of $T_1$ values as determined using a bootstrap error analysis with 1000 iterations for both glassy phase (blue) and β-phase (red), using the datasets from Fig. 10. A few (4) $T_1$ values $>18\mu \mathrm{s}$ are omitted to improve clarity.