Low dark leakage current in organic planar heterojunction photodiodesHimanshu Shekhar, Olga Solomeshch, Dan Liraz, and Nir Tessler

Citation: Appl. Phys. Lett. 111, 223301 (2017);View online: https://doi.org/10.1063/1.4996826View Table of Contents: http://aip.scitation.org/toc/apl/111/22Published by the American Institute of Physics

Articles you may be interested inSpatially inhomogeneous photoluminescence-voltage hysteresis in planar heterojunction perovskite-based solarcellsApplied Physics Letters 111, 223901 (2017); 10.1063/1.5001326

High-gain monolithic 3D CMOS inverter using layered semiconductorsApplied Physics Letters 111, 222101 (2017); 10.1063/1.5004669

Light-erasable embedded charge-trapping memory based on MoS2 for system-on-panel applicationsApplied Physics Letters 111, 223104 (2017); 10.1063/1.5000552

Wavelength-selective enhancement of photo-responsivity in metal-gated multi-layer MoS2 phototransistorsApplied Physics Letters 111, 223106 (2017); 10.1063/1.5003315

Heterojunction p-Cu2O/n-Ga2O3 diode with high breakdown voltageApplied Physics Letters 111, 222104 (2017); 10.1063/1.4998311

A crystalline germanium flexible thin-film transistorApplied Physics Letters 111, 222105 (2017); 10.1063/1.5007828

Low dark leakage current in organic planar heterojunction photodiodes

Himanshu Shekhar, Olga Solomeshch, Dan Liraz, and Nir Tesslera)

Microelectronic & Nanoelectronic Centers, Electrical Engineering Department, Technion Israel Instituteof Technology, Haifa 32000, Israel

(Received 19 July 2017; accepted 14 November 2017; published online 30 November 2017)

It is often suggested that the dark leakage current of organic photodiodes is due to extrinsic leakage

paths that do not involve the electronic junction. By studying a series of devices, where the acceptor

is kept constant (C70) and the donor material is varied, we find a direct correlation between the

strength of the sub-gap signature of the charge-transfer states and the leakage current. Attributing the

differences in the sub-gap absorption to the donor’s sub-gap states suggests that the donor’s side of

the junction should be made longer, to push the Fermi level at V¼ 0 towards the acceptor’s LUMO,

and thus, an optimized value of 800 Pacm�2 at V ¼ �1 V is reported. Published by AIP Publishing.https://doi.org/10.1063/1.4996826

Organic photodiodes (OPDs)1–4 are being presented as a

promising complementary or an alternative to inorganic based

photodetectors for detection/imaging applications.5–8 A major

challenge preventing polymer or small molecule based OPDs

from producing a high-quality photodetector is their dark cur-

rent level under reverse bias, which is usually several orders

of magnitude higher than the required value. Low reverse bias

dark current, often referred to as leakage current, is essential

for higher specific detectivity D* and large dynamic ranges

which are the key figure of merits of any photodetector. We

note that we are interested in the actual reverse bias dark cur-

rent and are not to be confused with the ideal-diode’s reverse

dark saturation current density, J0, which is the preexponent

factor of the forward current. It has been shown in some

works that by employing thick active layers,9 electron/hole

blocking layers,10–12 and different donor active materials,13

the leakage current to some extent can be suppressed. Despite

these successes, it remains a challenge because the physical

processes which govern the losses and determine the dark

leakage current are relatively unclear, and hence, it is still an

active area of research. Also, not always care is taken that

while minimizing the reverse (leakage) current the forward

diode current will not be compromised.

Here, we report the dependence of the reverse bias dark

current of an acceptor C70 based planar heterojunction (PHJ)

diode on different donor molecules [copper phthalocyanine

(CuPc), boron subphthalocyanine chloride (SubPc), 1,1-bis

[(di-4-tolylamino) phenyl] cyclohexane (TAPC), tetraphenyl-

dibenzoperiflanthene (DBP), and chloroaluminum phthalocya-

nine (ClAlPc)]. Similar comparisons were studied before but

in the context of solar cells’ open circuit voltage14,15 and less

in the context of the reverse bias dark current. It is common

(to many technologies) that reverse leakage current has a low

dependence on both voltage and temperature.16,17 This has led

to a range of explanations and models some of which are con-

cerned with the junction itself18–20 while others deal with

extrinsic factors and thus often lump it under the notion of

parallel or shunt resistance.15,16,21 To decide whether our

diodes belong to the former or latter group, we study identical

device structures and use the dark current at �1 V as an indi-

cation for the level of leakage current. We found that the leak-

age current reduced from CuPc/C70 to TAPC/C70 and to

SubPc/C70 with an overall reduction of 2 orders of magnitude.

To study the physical mechanism behind this result, we per-

formed both temperature dependent J-V and subgap external

quantum efficiency (EQE) analyses. Finally, as TAPC lends

itself easier (compared to SubPC) for device optimization, we

report a value of 800 pA cm�2 at �1 V for PHJ TAPC/C70

devices.

All devices were fabricated on top of an indium tin

oxide (ITO) coated glass substrate. The ITO substrates were

cleaned & dried as described elsewhere.22 Ultra-high purity

(UHP) grade organic materials CuPc, C70 (CreaPhys),

SubPc, TAPC, DBP, and ClAlPc (Lumtec) were used as

received. The first device structure used consists of a pat-

terned ITO anode, a 70 nm solution deposited thick film of

copper thiocyanate (CuSCN, Sigma 99%) as the hole trans-

port layer (HTL), a 20 nm thick film of donor, a 40 nm thick

film of C70 as an acceptor, a 8 nm thick film of bathocuproine

(BCP, Lumtec 99.99%) as the hole/exciton blocking layer,

and 30 nm thick Mg covered by a 60 nm thick Ag as the cath-

ode. To avoid perimeter leakage, the edges of the patterned

ITO were covered by a 100 nm polyimide layer, and the

device area was 18.4 mm2 (variation of leakage current

density between 1, 4, and 18 mm2 devices was less than a

factor of 2). The dark current-voltage of all the devices was

characterized using a semiconductor parameter analyzer

(B1500 A, Agilent Technologies) inside a nitrogen filled glo-

vebox. Temperature dependent dark J-V characteristics were

analyzed using a K-20 temperature controller (MMR

Technologies), and the signal was read using a Keithley

2612 source meter. Spectrally resolved external quantum

efficiency (EQE) was obtained outside the glove box with

measured samples kept in the nitrogen atmosphere inside a

holder. Light from the monochromator (Oriel, CS130) was

chopped at 80 Hz, and the signal was amplified (Low-Noise

Current Preamplifier SR570) and finally read using a lock-in

amplifier (EG&G 7265). The samples for cyclic voltammetry

(CV) measurements were prepared as evaporated films on

a)Author to whom correspondence should be addressed: nir@ee.technion.

ac.il. URL: http://webee.technion.ac.il/orgelect/.

0003-6951/2017/111(22)/223301/5/$30.00 Published by AIP Publishing.111, 223301-1

APPLIED PHYSICS LETTERS 111, 223301 (2017)

ITO substrates with ITO acting as the working electrode. All

measurements were made inside a glove box in an inert

atmosphere of N2.

Figure 1(a) shows the molecular structural formula of

the active materials used in this study, and the energy level

diagram of all the materials is shown in Fig. 1(b). Figure

1(c) shows the schematic of the investigated PHJ device in a

stack ITO/CuSCN (70 nm)/Donor (20 nm)/C70 (40 nm)/BCP

(8 nm)/Mg (30 nm)/Ag (60 nm), using the donor materials

CuPc, TAPC, SubPc, DBP, and ClAlPc. Because of the

higher and extended absorption spectrum in the visible

range, C70 was used in place of the commonly used C60

acceptor. To place the HOMO/LUMO values in the proper

context, Fig. 1(d) shows the cyclic voltammogram of two of

the active materials. It shows that when using the tangential

to estimate the onset, the two materials are assigned the

same energy level. However, as was shown for fullerene,22 it

is also clear that this simple procedure does not capture the

entire picture or it may explain the range of values reported

in the literature.

The measured dark and light performance characteristics

of the devices are shown in Fig. 2. Organic photovoltaics

(OPVs) with the structure similar to those studied here, using

SubPc/C70 and CuPc/C70, have been reported before.23–25

The solar cell characteristic parameters such as open circuit

voltage (Voc), short circuit current (Jsc), and power conver-

sion efficiency (PCE) of these devices are consistent (listed

in Table I) with the literature values. The dark J-V character-

istics (see Fig. 2) show typical diode behavior, with the leak-

age current part at low forward voltage being clearly visible

for TAPC & SubPc donor devices. The ideal-diode’s reverse

dark saturation current density, J0, of the devices was

obtained by extrapolating the exponential part, which is lin-

ear on the semi-log scale, and finding the intercept on the y-

axis (see the inset in Fig. 2). The actual slope defines the ide-

ality factor, n.

In this paper, the term leakage current is used for cur-

rents that are not part of the main (“ideal”) diode characteris-

tics. As discussed in the introduction, such leakage currents

could be due to shunt paths,16,26 thermal activation of car-

riers at the D/A interface,15 or thermally and/or electrically

activated injection of carriers at the contacts.27,28 To quantify

the leakage, we defined the actual dark leakage currents as

the value measured under a reverse bias of �1 V. The

extracted currents’ values are also collected in Table I.

Since, in the ideal case, there is a direct relationship between

kT�log(J0) and the open circuit voltage (VOC),29 we added

the VOC values to Table I too.

Examining the obtained values in Table I, we note that

there is a rough correlation between kTlog(J0), VOC, and the

change in the energy difference between acceptor LUMO

and the donor’s HOMO. Namely, J0 is relatively well

behaved and can be safely associated with excitation across

the junction. The actual leakage current (Jleak) is however

2–8 orders of magnitudes higher. Specifically, the change

from CuPc to TAPC reduced Jleak only by a factor of �15,

compared to 107 in J0. Namely, while by going from CuPc to

TAPC and SubPc, there was a significant improvement in

the leakage current, there is still significant room for

improvement. This is especially true for the TAPC based

device which happens to also have the lowest ideality factor.

FIG. 1. (a) Molecular structure of

organic molecules used in this study.

(b) Energy levels of all the materials

used in this work. (c) Schematics of the

investigated device structure. (d) Cyclic

voltammetry of TAPC and SubPc.

FIG. 2. Current density vs voltage characteristics in the dark (closed sym-

bols) and under one sun, AM 1.5G, illumination (open symbols) of investi-

gated devices for different donors. The built-in voltage values for CuPc,

TAPC, and SubPc donor devices are marked with black arrows. The device

structure is shown in Fig. 1. The inset shows the extrapolation method used

to obtain the ideal-diode leakage current for CuPc, TAPC, and SubPc donor

devices.

223301-2 Shekhar et al. Appl. Phys. Lett. 111, 223301 (2017)

In an attempt to better identify the physical origin of the

currents at low bias, we obtained the temperature dependence

of the dark J-V between 200 k and 300 k. The ideal-diode’s

reverse dark saturation current density, J0, is thermally acti-

vated and can be expressed as17

J0¼ J00 exp�/B

KT

� �; (1)

where /B is the activation energy, J00 is the prefactor, and

KT is the thermal energy. It is also accepted that for a diode

with an ideality factor n, the actual energy barrier DE is DE

¼ /B�n.17 We found that close to room temperature, the acti-

vation energies of J0 for (CuPc, TAPC, and SubPc) devices

are /B ¼ (0.45 eV, 0.5 eV, and 0.65 eV), and using the n val-

ues and Eq. (1), we get for the energy barrier DE ¼ (0.75 eV,

0.65 eV, and 1.0 eV). Namely, there is relative correlation

between the activation energies and VOC, suggesting that

both are determined by the same processes at the junction.

Returning to the main concern of this paper, the measured

dark currents at�1 V and �0.3 V, Jleak, and the activation ener-

gies close to room temperature are found to be /B_leak(�1 V)

¼ (0.45 eV, 0.28 eV, and 0.16 eV) and /B_leak(�0.3 V)

¼ (0.46 eV, 0.30 eV, and 0.18 eV) for (CuPc, TAPC, and SubPc)

devices, respectively. It seems strange that the device show-

ing the highest leakage also shows the highest activation

energy. However, this very device also exhibits a tempera-

ture dependent activation energy which reduces to �0.25 eV

below 250 k.

Turning the focus to our main interest, dark current

under reverse bias, Jleak, we note that the activation energies

are rather low. They are low compared to any energy differ-

ence that can be evaluated using the energetics presented in

Fig. 1(b), be the HOMO-LUMO differences or reverse injec-

tion barriers. Studies of inorganic planar heterojunctions

have shown that if the junction is the source of leakage, low

activation energies are attributed to tunneling being part of

the generation/recombination process and that it requires the

presence of deep sub-gap states and/or defects in the vicinity

of the junction.17,19,30 With the above in mind, we decided to

look into the sub-gap range using EQE spectra extending to

long wavelengths.

Figure 3(a) shows the zero bias spectral photocurrent

response of all the donor devices studied in this work. From

the measured spectral photocurrent response, two regions are

clearly distinguishable. A relatively flatter part, in the higher

energy region, corresponds to bulk absorption in the donor

and\or acceptor. The declining range, lower energy region,

corresponds to the excitation via sub-gap states.31 The photo-

current response observed in the lower energy part is often

referred to as charge transfer (CT) excitation and serves as a

measure for the absorption of these states.32,33 The primary

interest from this measurement was the estimation of the

subgap density of states in different donor’s based devices

and to reach energies below those of the CT state associated

with J0 (note the new feature that is revealed beyond

1200 nm or below 1 eV). The results in Fig. 3(a) can be

divided into three groups. In the first, there is the CuPc/C70

device showing a very broad response. In the second are

grouped the TAPC, DBP, and ClAlPc which show a second

feature beyond 1200 nm. The SubPc donor devices have the

lowest absorption tail, and beyond 1000 nm, it is below our

detection limit. Since C70 is the common acceptor in all the

devices, we attribute the pronounced subgap absorption to

the donor material in all devices. We note that the relative

strength of the sub-gap states [Fig. 3(a)] and that of the leak-

age currents [Fig. 3(b)] are correlated and are grouped

according to the low energy sub-gap states. This suggests

that the leakage current is associated with excitations across

the D/A interface through sub-gap or CT states. To test this

hypothesis, we inserted a thin (7 nm) layer of TAPC between

CuPc and C70. The EQE of this triple active layer is shown

as the dashed black line in Fig. 3(a), and its J-V curve is

shown (symbols) in Fig. 3(b). We note that both the sub-gap

EQE and the reverse leakage current changed to match those

of the TAPC/C70 bi-layer device. Careful examination of the

sub-gap EQE of the triple active layer (dashed line) shows

some remains of CuPc at �1100 nm and close to our set-up

noise margin, indicating the quality of the surface coverage

of CuPc by TAPC.

TABLE I. Dark and light characteristic parameters for devices, shown in Fig. 1, with various donors.

Donor HOMO (eV) J0 (A/cm2) n Jleak at �1 V (A/cm2) VOC (V) JSC (mA/cm2) PCE (%) FF (%)

CuPc 5.3 2 � 10�8 1.7 1 � 10�6 0.52 4.84 1.33 53

TAPC 5.5 2 � 10�16 1.3 6 � 10�8 0.98 5.24 1.83 36

SubPc 5.5 1 � 10�16 1.55 1 � 10�8 1.12 5.21 1.96 34

FIG. 3. (a) External quantum efficiency (EQE), extended to the sub gap

range, for devices of different donors. (b) J-V curves PHJ based on different

donors. Both (a) and (b) also show a CuPc/C70 device where a thin (7 nm)

TAPC layer was inserted inbetween (black, dashed line). (c) Sketch of phys-

ical mechanisms at play at the donor-acceptor interface in the measured

devices.

223301-3 Shekhar et al. Appl. Phys. Lett. 111, 223301 (2017)

Figure 3(c) shows schematically the physical picture we

suggest for the source of the leakage current in the above

diodes. Moving from CuPc to TAPC and to SubPc, the den-

sity of the donor’s tail states reduces and consequently the

probability for low energy CT states that could promote exci-

tation across the D/A interface reduces. Namely, we claim

that the increased leakage current is due to a higher genera-

tion/recombination at the junction and that the latter is

enhanced by the donor’s deep sub-gap states tailing from the

HOMO level.

If the main issues are the tail states of the donor’s

HOMO level, then it would be beneficial to position the

Fermi level at V¼ 0 between those states and the acceptors

LUMO. This can be achieved by lengthening the hole’s side

of the junction. Attempting to optimize the SubPc/C70 cell

was not successful due to the SubPc properties being highly

dependent on the film thickness.23 The optimized TAPC/C70

device structure was ITO/CuSCN (150 nm)/TAPC (60 nm)/

C70 (40 nm)/BCP (8 nm)/Mg (30 nm)/Ag (60 nm) which

consisted primarily of enhancing the thickness of a high

mobility hole transporting layer. The J-V characteristics of

this device are shown in Fig. 4 along with those of a device

having about half the CuSCN film thickness. Also com-

pared to the previous figures, we note that Jleak, at �1 V,

has reduced to 800 pA cm�2, while the exponential part of

J-V in the forward direction is not affected. Statistic across

3 substrates or 6 devices showed the ranges to be 3–8 nA

cm�2 and 800–1100 pA cm�2.

In conclusion, different donor based planar heterojunc-

tion devices are studied for their reverse bias leakage current.

We find that temperature dependent dark J-V exhibits activa-

tion energies that are too low to be correlated with common

energetics of the materials used to construct the devices (i.e.,

HOMO-LUMO differences or reverse injection barriers). We

have shown that the deep tail states at the donor/acceptor

junction are efficient generation-recombination centers and

are one of the main sources of the leakage current. Also, as

our study indicated that most of the subgap tail states are due

to the donor material, we found that positioning the Fermi

level (at V¼ 0) through device design can farther suppress

the leakage current. From the practical point of view, we

show that by minimizing the tail state density, i.e., by using

a less electronically disordered material, leakage current can

be greatly reduced. Also, by designing the device such that

most of the built-in voltage drops across the donor side, an

ultralow leakage current of 800 pA cm�2 at �1 V for TAPC/

C70 is obtained. Deciphering the details of the leakage mech-

anism at the junction requires further studies and will be

reported at a later stage.

This research was supported by the Israel Science

Foundation (Grant No. 488/16) and the Adelis Foundation for

renewable energy research within the framework of the Grand

Technion Energy Program (GTEP). O.S. acknowledges the

support by the Center for Absorption in Science of the

Ministry of Immigrant Absorption under the framework of the

KAMEA Program. Part of the device fabrication process was

performed at the Technion’s micro-nanofabrication & printing

unit (MNF&PU).

1S. F. Tedde, J. Kern, T. Sterzl, and P. Lugli, Nano Lett. 9(3), 980–983

(2009).2R. A. Street, M. Mulato, R. Lau, J. Ho, J. Graham, Z. Popovic, and J. Hor,

Appl. Phys. Lett. 78(26), 4193–4195 (2001).3P. Peumans, A. Yakimov, and S. R. Forrest, J. Appl. Phys. 93(7),

3693–3723 (2003).4C. K. Renshaw, X. Xu, and S. R. Forrest, Org. Electron. 11(1), 175–178

(2010).5T. N. Ng, W. S. Wong, M. L. Chabinyc, S. Sambandan, and R. A. Street,

Appl. Phys. Lett. 92(21), 213303 (2008).6D. Baierl, L. Pancheri, M. Schmidt, D. Stoppa, G.-F. Dalla Betta, G.

Scarpa, and P. Lugli, Nat. Commun. 3, 1175 (2012).7X. Tong and S. R. Forrest, Org. Electron. 12(11), 1822–1825 (2011).8G. Yu, J. Wang, J. McElvain, and A. J. Heeger, Adv. Mater. 10(17),

1431–1434 (1998).9A. Armin, M. Hambsch, I. K. Kim, P. L. Burn, P. Meredith, and E. B.

Namdas, Laser Photonics Rev. 8(6), 924–932 (2014).10P. E. Keivanidis, S.-H. Khong, P. K. H. Ho, N. C. Greenham, and R. H.

Friend, Appl. Phys. Lett. 94(17), 173303 (2009).11D.-S. Leem, K.-H. Lee, K.-B. Park, S.-J. Lim, K.-S. Kim, Y. W. Jin, and

S. Lee, Appl. Phys. Lett. 103(4), 043305 (2013).12F. Guo, Z. Xiao, and J. Huang, Adv. Opt. Mater. 1(4), 289–294 (2013).13R. F. Bailey-Salzman, B. P. Rand, and S. R. Forrest, Appl. Phys. Lett.

91(1), 013508 (2007).14M. D. Perez, C. Borek, S. R. Forrest, and M. E. Thompson, J. Am. Chem.

Soc. 131(26), 9281–9286 (2009).15W. J. Potscavage, S. Yoo, and B. Kippelen, Appl. Phys. Lett. 93(19),

193308 (2008).16S. Dongaonkar, J. D. Servaites, G. M. Ford, S. Loser, J. Moore, R. M.

Gelfand, H. Mohseni, H. W. Hillhouse, R. Agrawal, M. A. Ratner, T. J.

Marks, M. S. Lundstrom, and M. A. Alam, J. Appl. Phys. 108(12), 124509

(2010).17A. L. Fahrenbruch and J. Aranovich, in Solar Energy Conversion: Solid-

State Physics Aspects, edited by B. O. Seraphin (Springer, Berlin,

Heidelberg, 1979), pp. 257–326.18K. W. Mitchell, A. L. Fahrenbruch, and R. H. Bube, J. Appl. Phys. 48(10),

4365–4371 (1977).19A. R. Riben and D. L. Feucht, Solid-State Electron. 9(11), 1055–1065

(1966).20N. C. Giebink, G. P. Wiederrecht, M. R. Wasielewski, and S. R. Forrest,

Phys. Rev. B 82(15), 155305 (2010).21K. Ki-Beom, T. Yoon-Heung, H. Yoon-Soo, B. Kwang-Heum, Y. Myung-

Hee, and L. Moon-Ho, Jpn. J. Appl. Phys., Part 2 42(4B), L438 (2003).22H. Shekhar, L. Tzabari, O. Solomeshch, and N. Tessler, J. Appl. Phys.

120(15), 155501 (2016).23Y. Haomiao, Y. Ruichen, Z. Jiawei, Y. Anran, P. Huan, Q. Jiajun, and H.

Xiaoyuan, J. Phys. D: Appl. Phys. 49(20), 205105 (2016).24X. Xi, W. Li, J. Wu, J. Ji, Z. Shi, and G. Li, Sol. Energy Mater. Sol. Cells

94(12), 2435–2441 (2010).

FIG. 4. Current density vs voltage characteristics in the dark for the structure

ITO/CuSCN (150 nm)/TAPC (60 nm)/C70 (40 nm)/BCP (8 nm)/Mg (30 nm)/

Ag (60 nm). The dashed line shows the characteristics of a similar device

but with only 70 nm thick CuSCN.

223301-4 Shekhar et al. Appl. Phys. Lett. 111, 223301 (2017)

25H.-Y. Wei, J.-H. Huang, C.-Y. Hsu, F.-C. Chang, K.-C. Ho, and C.-W.

Chu, Energy Environ. Sci. 6(4), 1192–1198 (2013).26C. M. Proctor and T.-Q. Nguyen, Appl. Phys. Lett. 106(8), 083301 (2015).27S. Yoon, J. Cho, K. M. Sim, J. Ha, and D. S. Chung, Appl. Phys. Lett.

110(8), 083301 (2017).28W. Xiao, H. Jiang, L. Jie, and Y. Junsheng, J. Phys. D: Appl. Phys. 49(7),

075102 (2016).29N. Tessler, J. Appl. Phys. 118(21), 215501 (2015).30A. R. Riben and D. L. Feucht, Int. J. Electron. 20(6), 583–599 (1966).

31R. A. Street, A. Krakaris, and S. R. Cowan, Adv. Funct. Mater. 22(21),

4608–4619 (2012).32K. Vandewal, S. Albrecht, E. T. Hoke, K. R. Graham, J. Widmer, J.

D. Douglas, M. Schubert, W. R. Mateker, J. T. Bloking, G. F.

Burkhard, A. Sellinger, J. M. J. Fr�echet, A. Amassian, M. K. Riede,

M. D. McGehee, D. Neher, and A. Salleo, Nat. Mater. 13(1), 63–68

(2014).33F. Piersimoni, D. Cheyns, K. Vandewal, J. V. Manca, and B. P. Rand,

J. Phys. Chem. Lett. 3(15), 2064–2068 (2012).

223301-5 Shekhar et al. Appl. Phys. Lett. 111, 223301 (2017)