Polarization-Dependent Photoinduced Bias-Stress Effect in Single

Sep 15, 2017 - Given the availability of a number of organic semiconductors for applications in organic electronics, understanding mechanisms of opera...
1 downloads 8 Views 946KB Size
Subscriber access provided by UNIV OF CAMBRIDGE

Article

Polarization Dependent Photoinduced Bias-Stress Effect in Single-Crystal Organic Field-Effect Transistors. Hyun Ho Choi, Hikmet Najafov, Nikolai Kharlamov, Denis V. Kuznetsov, Sergei I. Didenko, Kilwon Cho, Alejandro Briseno, and Vitaly Podzorov ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.7b11134 • Publication Date (Web): 15 Sep 2017 Downloaded from http://pubs.acs.org on September 17, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Applied Materials & Interfaces is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Polarization Dependent Photoinduced BiasStress Effect in Single-Crystal Organic FieldEffect Transistors. Hyun Ho Choi,1,2 Hikmet Najafov,1,♣ Nikolai Kharlamov,3 Denis V. Kuznetsov,3 Sergei I. Didenko,3 Kilwon Cho,2 Alejandro L. Briseno,4 Vitaly Podzorov1,3,5,* 1

Department of Physics, Rutgers University, Piscataway, New Jersey 08854, USA.

2

Department of Chemical Engineering, Pohang University of Science and Technology

(POSTECH), Pohang 37673, South Korea. 3

National University of Science and Technology MISiS, Moscow 119049, Russia.

4

Department of Polymer Science & Engineering, University of Massachusetts, Amherst,

MA 01002, USA. 5

Institute for Advanced Materials and Devices for Nanotechnology, Rutgers University,

New Jersey 08854, USA. ♣

Current affiliation: Coherent Advanced Crystal Group, 31 Farinella Drive, East

Hanover, NJ 07936, USA KEYWORDS

organic semiconductor; rubrene; molecular crystal; organic transistor;

bias-stress effect; photoinduced charge transfer; memory; mobility.

1 ACS Paragon Plus Environment

ACS Applied Materials & Interfaces

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ABSTRACT Photoinduced charge transfer between semiconductors and gate dielectrics can occur in organic field-effect transistors (OFETs) operating under illumination, leading to a pronounced bias-stress effect in devices that are normally stable while operating in the dark.

Here we report an observation of a polarization dependent photoinduced bias-

stress effect in two prototypical single-crystal OFETs, based on rubrene and tetraphenylbis(indolo{1,2-a})quinolin (TPBIQ). We find that the decay rate of the source-drain current in these OFETs under illumination is a periodic function of polarization angle of incident photoexcitation with respect to the crystal axes, with a periodicity of π. The angular positions of maxima and minima of the bias-stress rate match those of the optical absorption coefficient of the corresponding crystals. The analysis of the effect shows that it stems from a charge transfer of “hot” holes, photogenerated in the crystal within a very short thermalization length ( 1012 cm-2.12

However, under illumination with a visible light, the same type of devices exhibits a substantial decay of ISD.

For this effect to occur, OFETs must be illuminated through

the (semitransparent) gate electrode with photons of energy above the semiconductor’s band gap (that is, when VG is applied simultaneously with photo-excitation of charge carriers in the semiconductor).27-28 The dependences of the photoinduced bias-stress rate on the photoexcitation wavelength, λ, light intensity, P, and gate voltage, VG, suggest that a charge transfer of photogenerated carriers from the semiconductor to the gate dielectric, leading to the dielectric’s charging, is responsible for this effect.27 Depending on the polarity of VG applied during illumination, either electrons or holes are transferred to the dielectric, thus resulting in a positive or negative shift of OFET’s onset voltage, respectively.27 The new, optically “programmed” onset voltage remains rather stable in further measurements in the dark, but can be re-programmed by illuminating the OFET while a different VG is applied. Since the first demonstration of such “optically programmable” OFET memory devices,27 many similar “memories” were suggested, some using various polymeric or amorphous oxide gate dielectrics,29-39 some using inorganic oxide nanoparticles to enhance charging and widen the memory window.40-41 Furthermore, observation and understanding of this light-induced bias stress effect are 6 ACS Paragon Plus Environment

Page 6 of 32

Page 7 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

very important for future flexible and transparent organic optoelectronics, where some circuit components might be inevitably exposed to light either from external sources or from OLED pixels integrated as a part of the circuits. For instance, in using OFETs as switches controlling OLED pixels, one would face the problem of bias stress under illumination. Even though we have suggested an interpretation of this phenomenon based on a charge transfer of “hot” photocarriers, further studies are necessary to verify this mechanism and understand its dependence on various parameters, such as light polarization, energy, angle of incidence, crystal facets and molecular packing of the semiconductor.

Here, we performed a comparative study of photoinduced bias-stress effect in organic crystals that exhibit different packing motifs, rubrene and tetraphenylbis(indolo{1,2a})quinolin (TPBIQ). We have discovered that the rate of bias-stress decay of ISD under illumination in the corresponding OFETs is a periodic function of polarization angle of incident light with respect to the crystal axes, with maxima and minima in the bias-stress rate reflecting the specific molecular packing of the crystals.

In addition, we have

observed that the rate of this effect also depends on the angle of incidence, and this dependence is qualitatively different for s and p-wave polarized photoexcitations.

These

new observations support the mechanism of photoinduced bias stress effect based on a charge transfer of “hot” photocarriers, but also reveal how molecular packing affects this instability in OFETs. Our results also bear practical importance, as they suggest simple strategies for minimizing this undesirable parameter drift in applied organic optoelectronics for instance by controlling the angle of incidence or polarization of light emanating from OLED pixels and incident on OFETs. 7 ACS Paragon Plus Environment

ACS Applied Materials & Interfaces

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 32

EXPERIMENTAL SECTION

Device fabrication.

We used top-gated single-crystal OFETs with a polymer gate

insulator (parylene-N) and a transparent indium-tin oxide (ITO) gate to study the response of the source-drain current of these devices to a linearly polarized light. Single crystals of rubrene and tetraphenylbis(indolo{1,2-a})quinolin (TPBIQ) were grown by physical vapor transport in a stream of ultra-high purity He and Ar gases, respectively.42 A sublimed grade rubrene (Sigma-Aldrich) and TPBIQ synthesized and purified by sublimation as described in Ref.

43

were used for the crystal growth.

Temperatures in

the sublimation zone during the crystal growth were kept at 320 and 360 °C for rubrene and TPBIQ, respectively. The temperature gradient along the growth tube was ~ 5 °C/cm, and the He (Ar) flow rate was 150 (100) cc/min. The details of device fabrication can be found elsewhere.44 In brief, graphite source and drain contacts were painted onto (a,b) facet of rubrene and (a,c) facet of TPBIQ crystals, defining a channel with length L = 1 3 mm and width W = 50 - 200 µm (for TPBIQ) and 1 - 2 mm (for rubrene) devices. This was followed by a deposition of 1 - 1.6 µm-thick parylene film that served as a gate insulator, topped with a transparent 30-40 nm-thick ITO gate sputtered through a shadow mask.

Electrical measurements.

The devices were exposed to a linearly polarized light

incident on the smooth flat surface of the ITO gate electrode (inset in Fig. 1). We used a 20-W halogen lamp with a smooth emission spectrum in the visible range or a blue (460 nm) LED as the light sources.45 The emission of the lamp passed through a band-pass filter, selecting an excitation wavelength of 500 nm, close to the maxima of absorption in 8 ACS Paragon Plus Environment

Page 9 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

TPBIQ and rubrene. A linear polarizer was then used to polarize the light. Thus, the devices were illuminated with a monochromatic, linearly polarized light.

The

bandwidth of the band-pass filters was ∆λ = 10 nm. The calibrated angular scale of the polarizer had a resolution of about 1 - 2°. To ensure that all the light reaching the crystal/insulator interface in our OFETs is absorbed in the semiconductor, the single crystals used in our devices were chosen to be much thicker than the light penetration length, α-1, for all the used wavelengths and polarizations (α is the absorption coefficient). Typical absorption lengths for visible light in these organic crystals are in the range α-1 ~ 1 - 20 µm. All device measurements were performed in air, at room temperature, using Keithley K2400 source meters and K6514 electrometers.

Optical “programming” of the accumulation layers in OFETs.

Optical recording was

achieved by first illuminating the whole channel of a rubrene OFET with a white light at VG = - 50 V, which sets the device in depleted state, and then illuminating the device at VG = + 50 V through a shadow mask consisting of a narrow (150 µm-wide) slit.

This

creates a laterally patterned hole accumulation channel (due to photoinduced electron transfer to the gate dielectric), in the form of a narrow stripe, conducting at VG = 0 in the dark. The shadow mask was suspended above the OFET and could be moved along the ITO gate to “record” multiple conducting stripes at chosen locations.

Typical times of

erasing (depleting) and recording the OFETs at VG = ± 50 V were 5 - 10 min. retention times of the recorded structures were at least several days.

9 ACS Paragon Plus Environment

Typical

ACS Applied Materials & Interfaces

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

RESULTS AND DISCUSSION

Prior charge-transfer studies in rubrene OFETs have shown that these devices exhibit extremely small bias-stress effect in the dark

12

- the fact that has driven us to conclude

that thermalized gate-induced polarons can hardly cross the rubrene/parylene interface. A significant bias-stress effect in these OFETs can only be observed under illumination with photons of energy above the band gap of rubrene.27 In this process, non-equilibrium charges photogenerated in the crystal next to the interface have sufficiently high energy to overcome the tunneling barrier at the rubrene/parylene interface and be driven into the gate insulator by the applied transverse gate electric field, EG⊥ ≡ (VG - Vonset)/d, where d is the thickness of the gate dielectric (inset in Fig. 1). Note that Vonset in this expression corresponds to the onset (or threshold) voltage of the OFET determined from its linearregime transconductance characteristics, ISD(VG).

Non-zero Vonset is usually due to

screening of the gate electric field by the charges transferred and localized in the gate insulator (here, we assume that contribution of deep traps to the threshold voltage is negligible, as is typically the case in pristine single-crystal OFETs1). The Vonset in singlecrystal p-type OFETs can be negative or positive, depending on whether the gate insulator is charged with holes or electrons, respectively, and it is zero in pristine devices that were not subjected to any gate bias stressing (under illumination or in the dark). The screening or enhancement of the gate electric field in the transistor channel leads to a positive (electron transfer) or negative (hole transfer) shift of the p-type OFET’s onset, and correspondingly to an increase or a decrease of the source-drain current, ISD, in realtime measurements at fixed gate, VG, and source-drain, VSD, voltages under illumination.

10 ACS Paragon Plus Environment

Page 10 of 32

Page 11 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

The source-drain current of an operating transistor at fixed VSD and VG in the linear regime can be expressed as:

I SD =

W µCiVSD (VG − Vonset ) . L

(1)

In this expression, µ is the longitudinal FET carrier mobility, Ci is the gate-channel capacitance per unit area, and the onset voltage, Vonset(t), is changing with time due to the charge transfer to the gate insulator: Vonset(t) = en(t)/Ci, where n(t) is the projected (areal) density of the transferred charge (not to be confused with the density of mobile polarons in the accumulation channel, n2D = e-1Ci(VG - Vonset)), and e is the elementary charge. Therefore, the charge transfer rate, dn/dt, can be expressed through the measured decay rate of the source-drain current, dISD/dt, as:

 dn L  dI SD  = − . dt  eµWVSD  dt

(2)

An example of such a measurement under a linearly polarized photoexcitation normally incident on a rubrene OFET is shown in Fig. 1. The polarizer is set intermittently along a or b-axis of the crystal, while ISD(t) at fixed VSD and VG is recorded. The overall decrease of the source-drain current with time is due to the transfer of holes to parylene that occurs in this measurement at VG < 0 under illumination. We note that performing these measurements in the linear regime of OFET’s operation simplifies data analysis because of a nearly constant transverse (gate) electric field and a uniformly distributed (that is, without a pinch-off) longitudinal potential drop along the channel (since |VSD| > α-1, we consider that the upper integration limit can be ∞):



∞ dI SD αl ∝ n' (α ) = χ 0 Φ 0α ∫ exp(− αz ) ⋅ exp(− z / l )dz = χ 0 Φ 0 ⋅ 0 dt αl + 1

(5)

Therefore, the modulation contrast, η, of the bias-stress rate with polarization angle is:

η ≡

n' (b) − n' (a) α (α l + 1) = 1− a ⋅ b , n' (b) α b (α a l + 1)

(6)

where αa and αb are the absorption coefficients for light polarized along the a and b-axes of rubrene. If photocarrier thermalization length is much shorter than the light penetration

15 ACS Paragon Plus Environment

ACS Applied Materials & Interfaces

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

length, l > α-1, the modulation contrast becomes negligible, η ≈ 0, corresponding to the experimental situation in which we would not see any difference in the bias-stress rate while rotating the polarizer.

We now compare the modulation contrast of the photoinduced bias-stress rate, dISD(θ)/dt, in rubrene OFETs (Figs. 1 and 2) with that of the absorption coefficient α(θ) of the crystal. For light normally incident on the (a,b) facet, the absorption spectra of rubrene for the two orthogonal polarizations, along b and a-axes, can be found in Ref. 5. For photoexcitation with a green (500 nm) light used here, the modulation contrast of α is about 41% (the corresponding light penetration lengths are αb-1 = 1 and αa-1 = 1.7 µm).

Figure 1 shows that the measured modulation contrast of the rate dISD(θ)/dt is about 50%, which is close to the modulation of α. Here, it must be noted that oscillations of the bias-stress rate with θ are gradually diminishing, as θ is varied, simply because a negative gate voltage (VG < 0) is constantly applied under a cw illumination during this measurement, and a substantial charge is being transferred to the gate insulator, as the experiment proceeds, causing the bias-stress rate to gradually decrease with time. This effect can be seen in Fig. 2, where the overall rates and the oscillation amplitude are gradually diminishing with θ (see also measurements of other devices in Figs. S2-S5 of Supporting Information). For this reason, the contrast of the bias-stress rate oscillations, determined in such a measurement, might vary depending on the sample’s stressing history. Nevertheless, we can see that the contrast of oscillations of dISD(θ)/dt is 16 ACS Paragon Plus Environment

Page 16 of 32

Page 17 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

comparable to the contrast in α(θ), which suggests that the length-scale relevant to this effect is rather small (l αa. According to the careful spectroscopic measurements and calculations by Irkhin et al.,48 this is the case in rubrene in the wide range of excitation wavelengths.

The situation is qualitatively different, when a b-polarized light (s-wave) is used in this experiment. The M molecular dipole (c-axis absorption) should not be excited at all (for

18 ACS Paragon Plus Environment

Page 19 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

any φ), and thus the effective absorption coefficient will be a decreasing function of φ in the entire experimental range, 0 ≤ φ < π/2:

α s − wave (ϕ ) = α b ⋅ 1 − ε −1 ⋅ sin 2 (ϕ ) .

(8)

Figure 3 shows measurements of the photoinduced bias-stress rate, dISD/dt, performed in these two modes in a rubrene single-crystal OFET (symbols). It can be seen that the data are qualitatively well described by Eqs. 7 and 8 with ε = 2.94 and αc/αa = 8.1, αb/αa = 1.26 (dotted lines). These ratios of the absorption coefficients are within the qualitative agreement with the report of P. Irkhin et al., where αc/αa = 5, αb/αa = 1.5 were determined via optical spectroscopy.48 Note that the large increase of dn/dt for a p-wave excitation (blue squares) cannot be ascribed to the suppression of p-wave reflection (increased transmission) due to φ approaching the Brewster angle (the Brewster angle for rubrene is φB ≈ 66º). Indeed, according to the Fresnel equations, the reflection coefficient of a p-

wave is not greater than 5% at any angle in the range 0 < φ < φB. Thus, the reflection losses cannot account for the effects shown in Fig. 3.

We emphasize that this method

yields only approximate ratios of optical constants αb/αa, because of several possible sources of error, including the dependence of photoinduced bias-stress rate on the OFET’s stressing history, measurement sequence and duration at each polarization, as well as contribution from dark bias stress.

These measurements show that the photoinduced bias-stress effect in OFETs parallels the polarization and angular dependence of the absorption coefficient of the crystal, α(θ, φ), thus confirming the relevance of a thin charge-transfer region, with the thickness l