Organic Bidirectional Phototransistors Based on Diketopyrrolopyrrole and Fullerene

In this work I demonstrate that a simple vertical D/A bilayer architecture is capable of fea-turing J-V-characteristics of a bidirectional organic phototransistor which may be switched by both electrical and optical means. The driving and switching voltages are in the order of 1 V.

The microscopic origin for this behaviour is based on the good charge transport properties of the donor, a small DPP molecule (Ph-TDPP-Ph),^370–375 and the efficient autoionization in the bulk of the acceptorC₆₀⁴⁶as well as its ability to transport both holes and electrons sufficiently well in this case (cf. chapter 5). Another important factor is the presence of an injection barrier to significantly reduce the dark current in the device. This behaviour is used to realize basic hybrid optical and electronic logic elements like NOT-, AND-, and OR-Gates with low driving voltage. Eventually, the corresponding set of logic operations is the basic requirement for the realization of advanced analog and digital applications. With this, the work demonstrates inter-esting applications of the light sensing ability of organic materials apart from the common use in standard organic solar cells.

An exemplary J-V characteristic of a ITO/MoO₃/Ph-TDPP-Ph/C₆₀/Al device under 1 sun illumination is shown in figure 8.13(a). Experimentally, I observe a very small current without illumination for devices with MoO₃ HTL and a pronounced photocurrent under forward bias.

Due to these properties, the corresponding device can be regarded as a vertical asymmetric bidirectional phototransistor, because current flow is possible under forward and reverse bias and can be controlled via illumination of the device. The on/off ratio, i.e the difference between current under illumination and in the dark, is in the order of 10⁵ (10³) at−1 V (+1 V). Notably, due to the low dark current no corrections were needed to evaluate the photocurrent in forward direction correctly.⁸⁷For monochromatic illumination above 2.25 eV I observe basically the same characteristic shape of the J-V curve as under sunlight conditions. Yet, when exciting below 2.25 eV the J-V characteristics feature a pronounced s-shape and very small current above V_oc (figure 8.13(b)). Consequently, the photocurrent is rather unidirectional under this condition.

These observations indicate that the photosensitivty of the device is switchable via the applied voltage. This aspect becomes especially evident from EQE spectra as a function of bias (figure 8.13(c)) The photocurrent spectrum at +1 V resembles the spectrum of single layer C₆₀devices and follows the absorption of C60. This indicates that charge generation under these conditions is due to autoionization of CT states in the bulk of C₆₀ (cf. chapter 5).⁴⁶ At 0 V and −1 V an additional contribution appears at higher wavelengths in the region where Ph-TDPP-Ph absorbs (8.13(c)). As autoionization in neat Ph-TDPP-Ph is very weak (figure 8.13(d)), the additional photocurrent has to be due to CT dissociation at the D/A interface. This dissociation process is very efficient as evidenced from the presence of a well defined saturation. Good charge transport properties as well as possibly CT delocalization may be responsible for this.84,131,162,163,174,428,429

Due to the reproducibility of the samples, I could combine two devices in a back-to-back con-figuration to achieve symmetric characteristics under illumination with respect to the origin, and a current saturation plateau for voltages above ±0.7 V, while the current is switched off in the dark. This results in a phototransistor like behaviour with an on/off ratio of 10³ (fig-ure 8.14(a)). As a first example of a logic gate, I used the tunability of the photosensitivity below 2.25 eV by the applied bias to realize a NOT-operation when the voltage is switched

be-tween 0 V (off state) and +1 V (on state) under constant illumination of a single device with monochromatic light in the region of 1.85−2.25 eV (figure 8.14(b)). Further applications that I successfully demonstrated in the work are AND- and OR-gates as well as a simple 4-bit analog ouput ADDER circuit (chapter 13). Consequently, the approach provides the full set of logic operations necessary to realize more complex analog and digital applications in future.

a)

c)

b)

d)

0.0 0.2 0.4 0.6 0.8 1.0

-3 -2 -1 0 1

current (µA)

voltage (V) total current dark current photo current

@1.91 eV

Figure 8.13.: (a) J-V-characteristic of a Ph-TDPP-Ph/C60 bilayer device with a MoO3 hole transport layer measured under AM1.5 conditions at 1 sun. (b) I-V-characteristic measured at 7^mW_cm² and an excitation energy of 2.76eV. The total current measured under illumination (grey solid line) is divided into its dark (light grey circles) and photocurrent contribution (red filled squares). EQE spectra of MoO3/Ph-TDPP-Ph/C60 devices as function of the applied voltage (left axis). Absorption spectra of the neat materials Ph-TDPP-Ph (dashed blue) and C60 (solid line) (right axis).(c) EQE measured under short circuit conditions for Ph-TDPP-Ph and C60

single layer devices (left axis). Absorption spectra of the neat materials Ph-TDPP-Ph (dashed blue) and C60 (solid line) (right axis).

The details about the underlying mechanism behind the observed transistor behaviour are in-ferred from the analysis of photocurrent spectroscopy, intensity-dependent J-V measurements as well as energy level considerations with two different hole transport layers (HTL),MoO₃ and PEDOT:PSS.

When MoO3 is replaced with PEDOT:PSS, the dark current is one to one and a half orders of magnitude higher. This indicates the presence of a considerable hole injection barrier in the case of an MoO₃ interlayer. It is further verified experimentally by the occurence and characteristic intensity dependent evolution of an s-shape in the J-V characteristics measured

78

below 2.25 eV^250,430 and explains the overall low dark current in the device. Furthermore, the analysis of the bias dependent EQE spectra revealed that (i) the photocurrent under forward bias is due to autoionization of charge transfer states in the bulk of C60, (ii) Ph-TDPP-Ph does not feature any significant intrinsic photogeneration, and (iii) dissociation under reverse bias is efficient at the D/A interface, as further evidenced by a good saturation of J-V-characteristics recorded below 2.25 eV.

a) b)

Figure 8.14.:(a) J-V-characteristic of the back-to-back configuration of two Ph-TDPP-Ph/C₆₀ devices measured at an excitation energy of 1.96eV (λ = 632nm). The total current measured under illumination (grey solid line) is divided into its dark (light grey circles) and photocurrent contribution (red filled squares). A schematic circuit diagram is shown as inset. (b) Output current IOut as a function of time for a NOT-Gate (black, bottom panel). The temporal course of the input voltage is displayed in blue (middle panel). The sample is constantly illuminated (red, top panel). Boolean states for input and output channels are indicated as binary numbers ([0]: off/low state, [1]: on/high state).

In view of these results, the observed bidirectional behaviour can be explained in terms of a photoenhanced recombination current mediated by the autoionization of CT states and thus the photogeneration of charge carriers within the C₆₀ layer.^87,88 Under reverse bias and excitation below 2.25 eV excitons are only generated in the donor (Ph-TDPP-Ph) and efficiently split at the D/A interface (figure 8.15 (a)). The generated charge then exit the device without encountering extraction barriers. At an excitation energy above 2.25 eV, the contribution from the autoionization in C60 simply adds to the total photogenerated current. Under forward bias and excitation below 2.25 eV excitons generated in the donor can only be split via electron transfer to the ITO/MoO₃ electrode. The remaining holes feature an extraction barrier at the D/A interface and accumulate, thereby giving rise to a space charge, which additionally impedes extraction.^86,429This results in a very small photocurrent. When exciting above 2.25 eV, however, CT states are autoionized in the C₆₀ layer so that electrons can now recombine with the accumulated holes on the donor side of the D/A interface. The remaining holes in the C₆₀ layer can then drift to the Al electrode (figure 8.15). This mechanism results in an observable net photocurrent that is directly visible when the dark current is very small. To a certain extent, this mechanism resembles a tandem solar cell. In all cases, additional losses due to low or imbalanced mobilities⁴²⁹ are avoided due to good charge transport in both C₆₀255,256,381 and Ph-TDPP-Ph.⁴³¹

Energy

ITO/MoO₃

vacuum DPP

C₆₀

+

- +

-

-1V

Energy

ITO/MoO₃

vacuum

DPP ^C60

Al

+ -

+ -

+1V

a) b)

Figure 8.15.: a) Simplified energy level diagram for ITO/MoO3/Ph-TDPP-Ph/C60/Al devices under a) reverse bias and b) forward bias.

80