Organic semiconductors with a charge carrier life time of over 2 hours at room temperature

Recently, Gao et al. reported being able to measure significant quantities of photogenerated charge up to one hour after it had been generated in an organic semiconductor device. The aim of this paper is twofold; (a) to provide conclusive experimental evidence to support the picture of device operation; and (b) to understand and demonstrate how changes to the device structure and materials can be used to tune the charge carrier lifetime. By tuning both the materials used, and the device structure we are able to observe a charge carrier life time of over 2 hours and still extract significant amounts of charge from the device after 5 hours. This is achieved by engineering the band structure of the device to control the spatial overlap of the stored photoexcited electron and hole populations and thus the recombination rate. By performing lifetime measurements as a function of charge carrier density and applied voltage we find the recombination rate has a 0th order dependence on carrier density, and elucidate the mechanisms responsible for these long charge carrier life times. This work is of technological significance for the development of organic electronic high sensitivity photodetectors and memory elements.


Abstract:
Recently, Gao et al. reported being able to measure significant quantities of photogenerated charge up to one hour after it had been generated in an organic semiconductor device. The aim of this paper is twofold; a) to provide conclusive experimental evidence to support the picture of device operation; and b) to understand and demonstrate how changes to the device structure and materials can be used to tune the charge carrier lifetime. By tuning both the materials used, and the device structure we are able to observe a charge carrier life time of over 2 hours and still extract significant amounts of charge from the device after 5 hours. This is achieved by engineering the band structure of the device to control the spatial overlap of the stored photoexcited electron and hole populations and thus the recombination rate. By performing lifetime measurements as a function of charge carrier density and applied voltage we find the recombination rate has a 0 th order dependence on carrier density, and elucidate the mechanisms responsible for these long charge carrier life times. This work is of technological significance for the development of organic electronic high sensitivity photodetectors and memory elements.

Introduction:
Since the emergence of the transistor in the late 1940s [1], silicon based components have been at the heart of all modern electronic devices. However, recently there has been considerable academic and industrial interest in developing a new class of electronic devices made from carbon based conducting organic molecules [2 -4]. These new organic electronic devices have many advantages over their silicon counterparts; they have the potential to be fabricated from solution using traditional printing techniques [5]; the semiconductors can be produced from low energy wet chemistry [6]; and the final device can be flexible enabling building and product integration [7]. Despite these advantages, only organic light emitting diodes (OLEDs) [8,9] have as of yet reached the mass market [10][11][12][13][14].
The primary reason for this is the low performance of organic semiconductor devices, due to their low charge carrier mobilities [15] and high recombination rates [16]. In organic semiconductors charge is usually localized on individual semiconducting molecules and conduction occurs via thermally assisted hopping [17] between neighboring molecules. This hopping process is inherently slower [18] than the band like conduction process which occurs in silicon devices, resulting in charge carrier mobilities which are typically two orders of magnitude [15] lower than their silicon counterparts [19]. Furthermore, not all molecules in an organic electronic device are electrically well connected to their neighboring molecules, resulting in configurational and energetic dead ends [17], where charge carriers can become trapped and eventually become annihilated through recombination [16]. Thus recombination rates in organic materials are typically higher than in their inorganic counterparts [20], with charge carrier life times ranging from a few picoseconds for bound electron hole pairs [ 21] to a maximum of a second for charges confined to deep energetic traps [22]. The last decade has seen significant attention focused on understanding the mechanisms responsible for the low mobilites and high recombination rates in organic semiconductors [23][24][25][26][27][28][29][30][31][32]. Usually in organic semiconductors, the recombination rate is defined as R=k (n,p )np (1) where k(n,p) is a carrier density dependent quantity, n is the electron density and p is the hole density.
In 2015, Gao et al. [33], reported a device design strategy to reduce the recombination rate in an organic semiconductor devices and as a result were able to extract significant quantities of charge up to one hour after photoexcitation. They proposed a sandwich structure of  very close to the HOMO edge at the center of the device, meaning there will be very few electrons for the holes to recombine with. The doping is believed to come from by-products of the synthesis of NPB see SI of [33]. By solving Possion's equation across the device in steady state, we found that the presence of doping was essential for the formation of the potential hill, and the closer the electron affinities of the contacts the more symmetric the potential hill would be. Results: Previously, Gao et al. [33], suggested that the higher the permittivity of the insulating layer, the smaller the potential drop over it would be and the more bent the bands would become over the semiconductor. He also suggested that this increased band bending would lead to longer charge carrier life times. However, in the previous work no high permittivity insulators were used (such as metal oxides) so it was not possible to determine if this trend would continue to hold as permittivity was increased and if even longer charge carrier lifetimes could be observed.     Previously, the recombination mechanism in organic semiconductors has typically (and broadly speaking [37]) been described as bi-molecular [38], meaning that the recombination rate is proportional to the square of the carrier density (R~n 2 ). To determine if the recombination mechanism in the present device is also bi-molecular, we fabricated a Glass/ITO/SiO2/NPB/Al structure, and photoexcited it with a range of light intensities over two orders of magnitude. The charge left in the device as a function of time can be seen in figure   6, the points represent experimental data while the lines represent lines with slope, =1 hour which act as a guide for the eyes. As in figure 4 it can be seen that the there is a relatively fast phase to recombination just after photoexcitation, followed by a slower phase. We understand the initial faster recombination phase as electrons which did not get swept out the device annihilating the stored hole population. It can also be seen that the overall shape and slope of the curves do not change as a function of initial charge density, suggesting that recombination rate in the present device is not dependent upon the carrier density. This 0 th order dependence of charge lifetime on carrier density, further supports the assertion of Gao [33], that a unipolar reservoir of charge is present at the center of the device and that very few photogenerated carriers stored in this reservoir are lost to electron-hole recombination.  thickness of the NPB semiconductor. The first general trend that can be seen by looking at the early time charge densities is that the thinner the device, the less charge is generated; this is due to not all the photons being absorbed by the active layer. Also, the thicker the device, the steeper the decay of carrier density between 50 s and 1000 s. By changing the thickness of the device we are effectively tuning the curvature of the band structure. For the device to be charge neutral, a thin device will have a much more curved band structure than thick device. A steeper band structure (blue line Figure 7b) will make it energetically easier for electrons to escape the device and thus the charge carrier life time will be longer, this result further supports the picture of device operation as described by Gao at al. The above experiments run over a very long time period, we therefore thought that degradation of the device through the generation of oxygen/water defect states could be responsible for some of the decay dynamics we see. To ensure this was not the case, we encapsulated an NPB device and compared the extracted charge carrier life times to those of a non encapsulated device. We found that encapsulation did not significantly change the charge stored as a function of time, and therefore we do not think degradation is influencing our results (see SI for figure).
In all the above experiments, the between time when the laser pulse was applied and the application of the charge extraction ramp, the device was kept at short circuit. If rather than keeping the device at short circuit we apply a small offset voltage (Voffset), we should be able to change how the bands are bent within the device and change the position of the point where the HOMO is at its maximum. This in turn will change the position where the photogenerated charge packet is kept within the device and its spatial overlap with any background charge (see figure 1d). We should therefore be able to adjust the charge carrier life time in this way, and if this works this will further support our picture of device operation presented above.   Conclusions: By altering both the device structure and measurement conditions we were able tune the charge carrier lifetime in the device from a few hundred milliseconds to over two hours. We find charge carrier life time within the structure depends on the permittivity of the insulating layer, the applied bias, device thickness, and the LUMO/HOMO level of the semiconductor used. By performing life time measurements as a function of laser intensity we were able to show that recombination is not bi-molecular in this device, in fact it is 0 th order, i.e. not a function of charge carrier density all, further supporting the idea of a reservoir of one charge carrier species being present in the device. Very long charge carrier lifetimes could be find application in high sensitivity photodetectors and memory elements.