Quantum Confinement of Hybrid Charge Transfer Excitons in GaN

Nov 27, 2017 - †Department of Materials Science and Engineering, ‡Department of Physics, and §Department of Electrical Engineering and Computer S...
0 downloads 18 Views 527KB Size
Subscriber access provided by READING UNIV

Communication

Quantum confinement of hybrid charge transfer excitons in GaN/InGaN/organic semiconductor quantum wells Anurag Panda, and Stephen R. Forrest Nano Lett., Just Accepted Manuscript • DOI: 10.1021/acs.nanolett.7b04122 • Publication Date (Web): 27 Nov 2017 Downloaded from http://pubs.acs.org on November 30, 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.

Nano Letters 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 29 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

Nano Letters

Quantum confinement of hybrid charge transfer excitons in GaN/InGaN/organic semiconductor quantum wells Anurag Panda1 and Stephen R. Forrest1,2,3* 1

Department of Materials Science and Engineering, University of Michigan, Ann Arbor, Michigan 48109, USA 2

3

Department of Physics, University of Michigan, Ann Arbor, Michigan 48109, USA

Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, Michigan 48109, USA Abstract

*Corresponding author: [email protected] Keywords: organic-inorganic heterojunction, wide energy gap, Poole-Frenkel emission, exciton dissociation, recombination We investigate hybrid charge transfer exciton (HCTE) confinement in organicinorganic (OI) quantum wells (QWs) comprising a thin InGaN layer bound on one side by GaN and on the other by the organic semiconductors, tetraphenyldibenzoperiflanthene (DBP) or 4,4′-bis(N-carbazolyl)-1,1′-biphenyl (CBP). A binding energy of 10 meV is calculated for the Coulombically bound free HCTE state between a delocalized electron in GaN and a hole localized in DBP. The binding energy of the HCTE increases to 165 meV when the electron is confined to a 1.5 nm In0.21Ga0.79N QW (HCTEQW). The existence of the HCTEQW is confirmed by measuring the voltage-dependent DBP exciton dissociation yield at the OI heterojunction in the QW devices that decrease with increasing In concentration and decreasing electric field, matching the trends predicted by Poole-Frenkel emission. Combining spectroscopic measurements with optical models, we find that 14 ± 3% of the excitons that reach the GaN/DBP heterojunction form HCTEs and dissociate into free charges,

1

ACS Paragon Plus Environment

Nano Letters 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

while the remainder recombine. A high non-radiative recombination rate through defect states at the heterointerface account for the lack of observation of HCTEQW photoluminescence from GaN/InGaN/CBP QWs at temperatures as low as 10K.

2

ACS Paragon Plus Environment

Page 2 of 29

Page 3 of 29 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

Nano Letters

Introduction Heterojunctions between organic and inorganic semiconductors are of interest for both their fundamental properties and potential applications. The different bonding characteristics of the two material systems result in vastly different optoelectronic properties including their dielectric constants, exciton binding energies, and charge mobilities.1,2 For example, type II organic-inorganic heterojunctions (OI-HJs) serve as exciton dissociation sites in dyesensitized solar cells3 and organic photovoltaic cells (OPVs), while type I and type III OI-HJs are utilized as charge injection and generation layers in quantum dot solar cells4, organic light emitting devices5 and OPVs. Recently there has also been interest in reduced dimensional, engineered OI-HJs that tune properties of the hybrid states through both strong and weak optical coupling.6,7 In past work we developed a comprehensive framework for understanding current transport and charge recombination processes in diodes based on type II OI-HJs8,9. A firstprinciples model was also developed for predicting the properties of the hybrid charge transfer exciton (HCTE)10, the quasi-particle that governs both exciton-to-charge conversion and charge recombination at the HJs. It was found that the HCTEs exist at HJs between inorganics such as ZnO10, TiO29, InP9, WS27, and organic semiconductors. Further, recombination at the HJ can be mediated by trapped HCTEs that are bound states of localized electrons on the inorganic semiconductor surface and holes in the organic semiconductor bulk. On the other hand, dissociation is mediated by free HCTEs whose electron is delocalized in the inorganic semiconductor. The low binding energy and oscillator strength of the free HCTE are determined by the asymmetry in the effective masses and dielectric constants of the two

3

ACS Paragon Plus Environment

Nano Letters 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 4 of 29

contacting materials. The free HCTE also has a dipole moment aligned perpendicular to the interface, allowing for tuning of its properties by an externally applied field. To gain further insight into HCTEs, in this work we explore exciton dynamics at an unusual hybrid quantum well (QW) where a relatively narrow energy gap InGaN layer is bounded on one side by GaN and on the other by either of the organic semiconductors, tetraphenyldibenzoperiflanthene (DBP) or 4,4′-bis(N-carbazolyl)-1,1′-biphenyl (CBP). The energetics of the HCTE are tuned by the degree of quantum confinement determined by the energy barriers on either side of the InGaN layer and the well thickness. Similar to ZnO, nitride-based semiconductors are suitable for studying HCTEs11 due to their low dielectric constant12 relative to other III-V compound semiconductors, ensuring a high binding energy for the state. Furthermore, nitride-based semiconductor composition can be modified to absorb in the ultraviolet, ensuring that the photoresponse of a visible-light-absorbing organic semiconductor can be spectrally resolved. Similar to ZnO, the GaN family of semiconductors exhibits rich defect phenomena due to lattice mismatch with the sapphire substrate.13 We calculate a 10 meV binding energy for the unbound, or free HCTE (HCTEF) at the GaN/DBP

heterointerface,

which

increases

to

165

meV

when

bound

in

a

GaN/In0.21Ga0.79N/DBP QW (HCTEQW). Voltage and QW In-concentration-dependent EQE measurements confirm the existence of a bound HCTEQW, whose dissociation efficiency is shown to be determined by Poole–Frenkel emission. Combining spectrally resolved photoluminescence quenching14,15 (SR-PLQ) and external quantum efficiency (EQE) measurements of GaN/DBP OI-HJs, we find 14 ± 3% of the organic excitons that diffuse to the heterointerface form HCTEs that subsequently dissociate to contribute to the junction photocurrent.

4

ACS Paragon Plus Environment

Page 5 of 29 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

Nano Letters

Further, we investigate photoluminescence (PL) from HCTEQW in GaN/InGaN/CBP QW structures. We observed no PL at temperatures as low as T = 10 K, whereas electroluminescence (EL) from trapped HCTEs (HCTET) is observed from T = 294 K to T = 10 K. This suggests that charge recombination at the OI-HJ occurs via electron trap states at the nitride surface on a time scale much shorter than the HCTEQW radiative recombination time. To our knowledge this is the first report of exciton confinement within a hybrid OI QW. Methods The epitaxial nitride layers are grown on sapphire substrates (University Wafer Inc., Boston, MA) in the (0001) orientation using metalorganic chemical vapor deposition (CVD) with Si as the n-type dopant. Prior to growth, the substrates are diced into 1 cm2 squares and sequentially sonicated in detergent, deionized water, acetone and isopropanol. The organic semiconductors (Luminescence Technology Corp., New Taipei) are purified once using vacuum thermal gradient sublimation16. The structures used for SR-PLQ measurements shown as insets of Fig. 1a consist of sapphire/4-5 µm, n++ GaN (1x1018 cm-3)/750 nm, n-GaN (1x1016 cm-3) and a combination of organic layers (DBP or CBP) that either block or quench DBP excitons. The OI diodes used for HCTE characterization consist of sapphire/4-5 µm, n++ GaN (1x1018 cm3

)/cathode/inorganic layers/30 nm DBP/30 nm MoO3/100 nm Al, as shown in Fig. 1b, inset.

For organic exciton dissociation efficiency characterization, the inorganic layer is 750 nm nGaN (1x1017 cm-3), whereas for HCTEQW characterization, the inorganic layer is 50 nm nGaN (1x1016 cm-3)/10 nm, undoped GaN/1.5 nm, undoped InxGa1-xN (x = 0.11 or 0.21). The DBP, MoO3 and Al layers are deposited by vacuum thermal evaporation at 1 Å/s at a

5

ACS Paragon Plus Environment

Nano Letters 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

background pressure 10-7 torr. The 5 nm Ti/50 nm Au cathode is deposited by e-beam evaporation at 0.5 Å/s with a background pressure 8x10-8 torr, on a region of the substrate where the n++ GaN is photolithographically exposed via plasma etching. Both the anode and cathode are deposited through shadow masks to define a 0.79 mm2 device area. Similar structures are used for EL spectral measurements except that the DBP layer is replaced by a 20 nm thick CBP layer. For hard-contact patterned devices used for low temperature EL measurements, a 1 µm thick SiO2 layer is deposited on the nitride surface using plasmaenhanced CVD, followed by photolithographically defining the device active area via etching in buffered HF. The sample is illuminated for SR-PLQ measurements from the DBP side at a wavelength of

442 nm using a 100 µW/cm2 He-Cd laser at 65o from normal. The PL is

collected at normal incidence using a fiber-coupled monochromator (Princeton Instruments SP-2300i) equipped with a Si CCD detector array (PIXIS:400). A similar procedure is followed for the temperature dependent PL characterization, with the exception that the pump intensity is 15 mW/cm2 at

325 nm. The EL spectral measurements use a two-stage,

closed-cycle Janis He cryostat using the same optical system as for the PL measurements. The temperature is maintained with a thermally controlled stage heater, allowing 45 min for stabilization at each temperature. The EQE measurements employ a monochromated (5 nm spectral resolution) Xe lamp chopped at 200 Hz, a lock-in amplifier, and a current amplifier (Keithley Model 428). A National Institutes of Standards and Technologies-traceable Si photodetector is used to calibrate the intensity at each wavelength. A parameter analyzer and AM 1.5G solar illuminator are used to obtain dark and light current density vs. voltage (J-V) characteristics. 6

ACS Paragon Plus Environment

Page 6 of 29

Page 7 of 29 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

Nano Letters

The power is adjusted to 100 mW/cm2 (1 sun intensity) using a National Renewable Energy Laboratory calibrated reference cell. Quantum state calculations combine molecular dynamics (MD) simulations using Materials Studio® of the organic molecules deposited on the InGaN (0001) surface, density functional theory (DFT) calculations of cationic molecular orbitals, and solutions to Schrödinger’s equation10. The DFT calculations use Gaussian 09®, with the B3LYP functional and the 3-21G basis set. The time-independent Schrodinger equation is solved using COMSOL Multiphysics®. The parameters used for simulation are listed in Table 1, and energies of the QW are reproduced in Fig. 2. Theory The proposed energy level diagrams for the QW devices based on vacuum level alignment of the semiconductor energy levels and experimental offsets reported for nitride semiconductor systems17 are shown in Fig. 2. We use DBP as the organic semiconductor since its energy gap is less than that of InxGa1-xN (x = 0.11 or 0.21) and GaN, ensuring that the voltage-dependence of its photoresponse can be spectrally resolved from the inorganic semiconductor photoresponse. In contrast, the wide energy gap of CBP maximizes ∆EOI, the energy difference between the GaN conduction band minimum and the CBP highest occupied molecular orbital (HOMO), as well as ∆EcL, the energy difference between the GaN conduction band minimum and the CBP lowest unoccupied molecular orbital (LUMO). These energy barriers effectively confine the injected charges at the InGaN/CBP HJ, and allow for the possibility of observing the HCTEQW by PL or EL.

7

ACS Paragon Plus Environment

Nano Letters 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 29

The nitride semiconductor surfaces are expected to efficiently dissociate organic excitons due to their large ∆EcL for both CBP and DBP. The resulting HCTEF and HCTEQW PL and EL spectral peaks are expected at ∆EOI – EB and ∆EOI – ∆Ec + EE1 – EB, respectively. Here, EB is the binding energy of the state, ∆Ec is the conduction band offset between InGaN and GaN, and EE1 is the first quantized electron state energy in the QW relative to the InGaN conduction band minimum energy. The spectral peak of HCTET is determined by the electron Fermi level, Ef,n, at the inorganic surface as determined by the surface trap density of states.10 The lowest energy eigenvalues (for the singlet states: 1HCTEF and 1HCTEQW) obtained from simulations for the DBP based OI-HJs are shown in Fig. 3. The EB = 10 meV for the 1

HCTEF delocalized in GaN, and 127 meV for the 1HCTEQW confined in the In0.11Ga0.89N

well. When the In composition increases to x = 0.21, ∆Ec increases from 0.3 eV to 0.5 eV, resulting in an increase of the EB to 165 meV. The electron wavefunction penetration into the organic semiconductor increases from 0.1% in 1HCTEF to 4.5% in 1HCTEQW in In0.11Ga0.89N, which gives rise to an increase in oscillator strength of the state. Increasing the In concentration from x = 0.11 to x = 0.21 increases the electron wavefunction penetration to 6.0%. When DBP is replaced with CBP, ∆EcL increases from 0.6 eV to 1.7 eV. As a result, the EB = 97 meV and 136 meV for 1HCTEQW in the x = 0.11 and x = 0.21 InGaN samples, respectively. The decrease in binding energy for CBP as compared to DBP results from the decrease in electron wavefunction penetration into the organic semiconductor, which are 1.2% and 2% for x = 0.11 and x = 0.21 InGaN samples, respectively. The HCTEQW spectral peak is expected to shift with voltage due to the voltage dependence of EE1 and

. At high electric fields, the change in both

and EE1 can be

significant. For example, for the GaN/In0.11Ga0.89N/CBP QW at 104 V/cm (Va = 0.1 V), 105 8

ACS Paragon Plus Environment

Page 9 of 29 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

Nano Letters

V/cm (Va = 1 V), 106 V/cm (Va = 10 V), we calculate an increase of EE1 of 3 meV, 24 meV, and 163 meV respectively, and an increase in

of 2 meV, 12 meV and 56 meV,

respectively. The wavefunction penetration into the organic semiconductor side of the HJ remains unchanged at 1.5%. Spontaneous and piezoelectric polarizations existing within the nitride semiconductors will have an effect of similar magnitude on the properties of the HCTEQW. In our measurements, however, they are screened by the externally applied field. Results The experimental and calculated SR-PLQ signal intensity ratio of DBP on a quenching and blocking GaN surface are shown in Fig. 1a.13 In the quenching sample, DBP forms a type II HJ where excitons recombine or dissociate. The CBP layer inserted between GaN and DBP forms a type I HJ with DBP, and thus block DBP excitons. The Fabry-Pérot oscillations result from the index of refraction contrast between GaN and the adjacent layers. A DBP diffusion length of LD = 10 ± 1 nm18 is found from the data using the method of Bergemann et al.,14 assuming 100% quenching at the GaN interface. The match between the calculated and experimental ratios indicates that all the excitons generated in DBP that reach the GaN surface either dissociate or recombine. The EQE spectrum of the OI-HJ device results solely from generation and dissociation of DBP excitons, since GaN is transparent at

365 nm. We fabricated a DBP

photoconductor to ensure that the EQE in the OI-HJ device is not due to DBP exciton dissociation in the bulk, a process which has recently been reported to yield EQE > 10%.19 The EQE spectra of the GaN/DBP OI-HJ and DBP photoconductor are shown in Fig. 1b. The

9

ACS Paragon Plus Environment

Nano Letters 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

photoconductor is forward biased to 0.66 V to match the built-in field in the DBP layer in the OI-HJ at 0 V. Since the thickness of the inorganic layer is larger than its depletion width, the built-in field across the organic layer is determined by self-consistently solving for the voltages across the organic and inorganic layers using the depletion approximation in the inorganic, and the field continuity condition (Poisson equation) and uniform field approximation in the organic.8 The equations is iteratively solved until the sum of the voltage drops across the two layers gives the built-in voltage due to the difference in the work function of the electrodes (1.1 V). The calculations indicate that 40 % of the built-in voltage is dropped across the organic layer. The field in the photoconductor is determined using the uniform field approximation, assuming a built-in voltage of 1 V. The peak EQEs at

615 nm of the OI diode and photoconductor are 3.6 ± 0.5% and

0.22 ± 0.02%, respectively. Optical modeling coupled with solutions to the exciton diffusion equation20,21 are used to determine the fraction of excitons that reach the HJ (see Fig. 1b). The exciton diffusion equation is solved using LD = 10 ± 2 nm, assuming that the quenching efficiencies at both the GaN and MoO3 surfaces are 100%. This analysis shows that, at λ = 615 nm, the fraction of light absorbed in DBP and the fraction of excitons that diffuse to the OI-HJ are 79.5 ± 2.5% and 22.8 ± 2.5%, respectively. The J-V characteristics in the dark and at 1 sun intensity, simulated AM1.5G radiation for GaN/InGaN/DBP QWs are shown in Fig. 4a. The devices have < 10-3 mA/cm2 reverse bias leakage current up to an applied voltage of Va =

8 V. Both devices show charging at low

current due the traps, and as a result do not reach J = 0 at Va = 0 in the dark with a

10

ACS Paragon Plus Environment

Page 10 of 29

Page 11 of 29 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

Nano Letters

measurement integration time of 17 ms at each voltage. The In0.21Ga0.79N QW device has a more abrupt turn-on under forward bias as compared to the x = 0.11 QW device. Both have similar ideality factors of n = 3.2 ± 0.1. The x = 0.21 device also has a larger slope (between 2 V and +1 V) and lower photocurrent as compared to the x = 0.11 QW. Figure 4b shows the EQE vs. wavelength and voltage of the QW devices. The spectral peaks at

475 nm are due to absorption by InGaN, and between

475 nm and

675

nm are solely due to DBP. The x = 0.21 device has a lower EQE from both the InGaN and DBP as compared to the x = 0.11 device, although the proportional increase with voltage is similar for both structures. This suggests that the increased In concentration and correspondingly increased ∆Ec increases electron confinement in the QW. The InGaN photoresponse peaks show minor shifts with voltage due a combination of screening of the built-in polarization fields and the quantum-confined Stark effect17, which confirm that DBP serves as an effective electron barrier at its interface with InGaN. The proportional increase in InGaN EQE with voltage is also lower than the increase in DBP photoresponse, suggesting an additional voltage-dependent mechanism for DBP exciton dissociation. The EL spectra at T = 294 K and Va = 6 V, and at T = 10 K and Va = 12 V of the GaN/InGaN/CBP QW devices are shown in Fig. 5. In addition to the organic and inorganic semiconductor bulk emission peaks, an additional peak is observed at λ

600 nm (2.1 eV)

that is attributed to the InGaN/CBP HCTE. The emission intensities of the bulk semiconductors increases with V and decreasing T. The intensity of the HCTE peak also increases with decreasing T. However, at a given V, the HCTE spectral intensity relative to the bulk emission decreases with T. The HCTE spectra remain broad and the spectral peak is independent of both voltage and temperature. At the lowest measurement temperature (T = 10 11

ACS Paragon Plus Environment

Nano Letters 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 12 of 29

K), the HCTE spectral peaks shift to the blue by approximately 75 nm (x = 0.11) and 40 nm (x = 0.21). No HCTE PL is observed between T = 294 K and T = 10 K. Discussion The SR-PLQ data confirm that excitons from DBP and CBP that form type II HJs with nitride semiconductors either dissociate or recombine at the OI-HJ. The difference in EQE between the device with the OI-HJ and the photoconductor is 3.4 ± 0.5% at

615 nm. The

additional EQE in the OI-HJ device is attributed to DBP exciton dissociation via HCTEF. Further, since 22.8 ± 2.5% of the photogenerated excitons reach the heterointerface where 3.4 ± 0.5% successfully dissociate, we infer an exciton-to-charge conversion efficiency of 14 ± 3%. This suggests a high recombination rate of either the HCTE or its excitonic precursor at the OI-HJ. To understand the effects of the HCTEQW binding energy on its dissociation, we normalize the EQE due to DBP (

615 nm) exciton dissociation at a given voltage to the

relative change in InGaN QW EQE at that voltage from its value at 0 V. This factors out the change in charge collection through the DBP layer and emission over ∆Ec due to the applied field. Excitons generated in the InGaN layer dissociate in the QW because their binding energy is lower than the thermal energy at room temperature. We then subtract the contribution due to DBP photoconductivity determined by biasing the photoconductor to match the field in the DBP layer in the QW. The electric field in the QW is determined using the uniform field approximation.

12

ACS Paragon Plus Environment

Page 13 of 29 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

Nano Letters

The increase in DBP EQE vs the electric field, F, in the QW is shown in Fig. 6. The observed voltage dependence is a result of the field-dependent dissociation of HCTEQW via Poole-Frenkel emission, which follows1,22:

(1)

Here,

is the dissociation rate of the HCTE,

elementary charge,

is the

.

(2)

is a constant, and: =

Also,

is the Boltzmann’s constant,

is the vacuum permittivity and

is the dielectric constant of the inorganic

semiconductor given in Table 1. The data are fit using Eq. 1, with slopes

= 1.0 ± 0.2x10-3

(cm/V)1/2 for x = 0.11, and 1.2 ± 0.2 x10-3 (cm/V)1/2 for the x = 0.21 QW, which are both similar to

= 9.0x10-3 (cm/V)1/2 predicted by Eq. 2. The change in intercept at F = 0 gives a

difference in binding energies of the HCTEQW between the two In composition devices of = 42 ± 12 meV, that matches the

= 38 meV calculated due to the change in ∆Ec.

The observation of the bound electron in the HCTEQW, the high recombination rate of the HCTE at the OI-HJ and the predicted electron wavefunction penetration into the organic side of the OI-HJ suggests that HCTEQW should be observable by both PL and EL. However, no HCTEQW PL is observed in the GaN/InGaN/CBP QW samples, indicating a high nonradiative recombination rate of the state. In contrast, broad and voltage independent HCTE EL is observed from the GaN/InGaN/CBP QW. To confirm if the HCTE emission spectrum is due to recombination at the OI-HJ, we replace CBP with N,N′-Di(1-naphthyl)-N,N′-diphenyl-

13

ACS Paragon Plus Environment

Nano Letters 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

(1,1′-biphenyl)-4,4′-diamine (NPD) and observe that the HCTE spectral peak red shifts to λ 675 nm (1.8 eV), in quantitative agreement with the shallower HOMO of NPD (5.5 eV). Further, at higher Va, the inorganic and organic semiconductor emission intensities increase while that of the HCTE decreases due to charge emission over the barriers at the OI-HJ. The HCTE EL spectral peak is likely due to HCTET since it does not shift with Va, and has an abrupt blue shift at T = 10 K. The blue shift results from carrier freeze-out that pins the Ef,n near to the band edge.23 The voltage independence of the spectral speak suggests that a large density of trap states are concentrated close to the InGaN conduction band, pinning Ef,n at the trap energy. The trap states are likely generated during CVD processing of SiO2 on the nitride surface, which is consistent with reports of the formation of trap states during oxide deposition.24,25 The behavior of the dark J-V further confirms that a high density of trapped charge carriers with discharge time constant > 17 ms (the measurement sample time) exist at the OI-HJ. Further, the HCTE EL spectral peak at 2.1 eV is higher than the 1.7 eV predicted for HCTEQW, implicating interface states in determining the OI-HJ alignment. From these observations, we conclude that the EL spectra from the GaN/InGaN/CBP QWs are due to transitions from HCTET to the ground state. The high non-radiative recombination rate of the HCTEQW results from rapid phonon-assisted thermalization through the midgap states on time scales faster than the radiative recombination rate. This is further confirmed by the lack of PL from the HCTEQW. Assuming that the HCTE or exciton quenching rates at the GaN and (In)InGaN surfaces are similar, we can estimate the minimum surface recombination site density. Since 86% of the excitons that reach the OI-HJ recombine, and since the HCTEF radius is ~ 10 nm with a binding energy of only 10 meV (see Fig. 3), they are unlikely to diffuse along the

14

ACS Paragon Plus Environment

Page 14 of 29

Page 15 of 29 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

Nano Letters

interface before dissociation. From this, we estimate a minimum surface recombination site density of 3x1011 cm-2. This is lower than the interface trap density of 1x1012 cm-2 reported at GaN/oxide interfaces.24,25 Using chemical passivation a six-fold increase in the GaN surface recombination velocity has been reported.26 Assuming the surface recombination sites are passivated in our case, we expect expect a similar increase in the dissociation yield of the HCTEQW. Our quantum mechanical model opens the possibility of the use of OI QWs to modify and control the organic exciton dissociation process. For example, controlling dissociation by changing the resonance condition between the electron energy in the QW with the organic exciton is now possible in such hybrid QW systems. Conclusion We demonstrated that the dissociation of organic semiconductor excitons at heterojunctions with (In)GaN is mediated through hybrid charge transfer excitons (HCTE). Further, we observe quantum confinement of the HCTE state (HCTEQW) within hybrid InGaN QWs bounded on one side by GaN, and on the other by an organic semiconductor. The HCTEQW binding energy increases with quantum well depth, making it stable at room temperature. The existence of the HCTEQW and its confinement are inferred from the electric field dependent dissociation via Poole-Frenkel emission. This study demonstrates the tunability of the optoelectronic properties of HCTEs by confinement within QWs comprising a new class of heterogeneous materials systems.

15

ACS Paragon Plus Environment

Nano Letters 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

Acknowledgements: This work was supported in part by the Army Research Office Contract W911NF-15-1-0554. We acknowledge the Lurie Nanofabrication Facility at the University of Michigan for infrastructure used in device fabrication. We thank David Laleyan for helpful discussions.

16

ACS Paragon Plus Environment

Page 16 of 29

Page 17 of 29 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

Nano Letters

References (1)

Pope, M.; Swenberg, C. E. Electronic Processes in Organic Crystals and Polymers, 2 edition.; Oxford University Press: New York, 1999.

(2)

Sze, S. M.; Ng, K. K. Physics of Semiconductor Devices, 3 Ed.; Wiley-Interscience: Hoboken, N.J, 2006.

(3)

O’Regan, B.; Gratzel, M. Nature 1991, 353 (6346), 737–740.

(4)

Sargent, E. H. Nat. Photonics 2012, 6 (3), 133–135.

(5)

Qi, X.; Li, N.; Forrest, S. R. J. Appl. Phys. 2010, 107 (1), 014514.

(6)

Slootsky, M.; Liu, X.; Menon, V. M.; Forrest, S. R. Phys. Rev. Lett. 2014, 112 (7), 076401.

(7)

Liu, X.; Gu, J.; Ding, K.; Fan, D.; Hu, X.; Tseng, Y.-W.; Lee, Y.-H.; Menon, V.; Forrest, S. R. Nano Lett. 2017, 17 (5), 3176–3181.

(8)

Renshaw, C. K.; Forrest, S. R. Phys. Rev. B 2014, 90 (4), 045302.

(9)

Panda, A.; Renshaw, C. K.; Oskooi, A.; Lee, K.; Forrest, S. R. Phys. Rev. B 2014, 90 (4), 045303.

(10) Panda, A.; Ding, K.; Liu, X.; Forrest, S. R. Phys. Rev. B 2016, 94 (12), 125429. (11) Slawinski, M.; Weingarten, M.; Axmann, S.; Urbain, F.; Fahle, D.; Heuken, M.; Vescan, A.; Kalisch, H. Appl. Phys. Lett. 2013, 103 (15), 153305. (12) Levinshteĭn, M. E.; Rumyantsev, S. L.; Shur, M. Properties of advanced semiconductor materials: GaN, AlN, InN, BN, SiC, SiGe; Wiley: New York, 2001. (13) Gil, B. Group III nitride semiconductor compounds: physics and applications; Clarendon Press ; Oxford University Press: Oxford; New York, 1998. (14) Bergemann, K. J.; Forrest, S. R. Appl. Phys. Lett. 2011, 99 (24), 243303.

17

ACS Paragon Plus Environment

Nano Letters 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

(15) Lunt, R. R.; Giebink, N. C.; Belak, A. A.; Benziger, J. B.; Forrest, S. R. J. Appl. Phys. 2009, 105 (5), 053711. (16) Forrest, S. R. Chem. Rev. 1997, 97 (6), 1793–1896. (17) Chuang, S. L. Physics of optoelectronic devices; Wiley: New York, 1995. (18) Hirade, M.; Adachi, C. Appl. Phys. Lett. 2011, 99 (15), 153302. (19) Chandran, H. T.; Ng, T.-W.; Foo, Y.; Li, H.-W.; Qing, J.; Liu, X.-K.; Chan, C.-Y.; Wong, F.-L.; Zapien, J. A.; Tsang, S.-W.; Lo, M.-F.; Lee, C.-S. Adv. Mater. 2017, 29 (22), 1606909. (20) Pettersson, L. A. A.; Roman, L. S.; Inganäs, O. J. Appl. Phys. 1999, 86 (1), 487–496. (21) Peumans, P.; Yakimov, A.; Forrest, S. R. J. Appl. Phys. 2003, 93 (7), 3693–3723. (22) Killesreiter, H.; Baessler, H. Phys. Status Solidi B 1972, 53 (1), 193–199. (23) Morkoç, H. Handbook of Nitride Semiconductors and Devices, Materials Properties, Physics and Growth; John Wiley & Sons, 2009. (24) Ostermaier, C.; Lee, H.-C.; Hyun, S.-Y.; Ahn, S.-I.; Kim, K.-W.; Cho, H.-I.; Ha, J.-B.; Lee, J.-H. Phys. Status Solidi C 2008, 5 (6), 1992–1994. (25) Gregušová, D.; Stoklas, R.; Mizue, C.; Hori, Y.; Novák, J.; Hashizume, T.; Kordoš, P. J. Appl. Phys. 2010, 107 (10), 106104. (26) Martinez, G. L.; Curiel, M. R.; Skromme, B. J.; Molnar, R. J. J. Electron. Mater. 2000, 29 (3), 325–331.

18

ACS Paragon Plus Environment

Page 18 of 29

Page 19 of 29 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

Nano Letters

Table 1: Parameters used to calculate the hybrid charge transfer exciton properties Parameter#

Value

mO* / me m*I / me (x = 0, 0.11, 0.21) ε rO

1.0 0.2, 0.19, 0.18 4 8.9, 9.6, 10.2

ε rI (x = 0, 0.11, 0.21) #

mO* and m*I are the effective masses of the hole in the organic and the electron in the

inorganic semiconductor, respectively, where me is the electron rest mass. ε rO and dielectric constants of the organic and inorganic semiconductors, respectively.

19

ACS Paragon Plus Environment

ε rI are the

Nano Letters 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 20 of 29

Figure Captions: Figure 1: (a) Experimental and calculated spectrally resolved tetraphenyldibenzoperiflanthene (DBP) photoluminescence intensity ratio vs. wavelength at a quenching and blocking interface (structures shown in inset). The match between the experimental and calculated ratios indicates that 100% of the excitons that reach the organic-inorganic heterojunction (OIHJ) either dissociate or recombine. (b) External quantum efficiency (EQE) of a GaN/DBP OIHJ and a DBP photoconductor biased at +0.66 V to match the built-in field in the DBP layer in the OI-HJ device at 0V. Both device structures are shown in inset. The EQE spectrum is due to dissociation of DBP excitons. The percentage of light absorbed by the DBP layer and the percentage of photogenerated excitons in the layer that diffuse to the OI-HJ are also shown.

Figure 2: Spatial probability density of the lowest eigenvalue solution of the free and confined singlet hybrid charge transfer excitons (1HCTEF and 1HCTEQW) at the GaN/DBP heterojunction, and the GaN/1.5 nm In0.21Ga0.79N/DBP quantum well. The binding energy of HCTEQW increases from 165 meV to 10 meV for HCTEF, and the electron wavefunction penetration into the organic semiconductor increases from 6%, to 0.1%.

Figure

3:

Energy

level

diagram

of

GaN/1.5

nm

InxGa1-xN

(x

=

0.11

or

0.21)/tetraphenyldibenzoperiflanthene (DBP) organic-inorganic quantum wells (QWs) devices at flat-band determined by vacuum level alignment. Also shown are the energy levels of 4,4′bis(N-carbazolyl)-1,1′-biphenyl (CBP). The ∆EOI is the difference between GaN conduction

20

ACS Paragon Plus Environment

Page 21 of 29 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

Nano Letters

band minimum energy and CBP highest occupied molecular orbital energy, ∆Ec is the offset between the conduction band minima of GaN and InGaN, and EE1 is the energy of the first quantized electron state in the QW relative to the InGaN conduction band minimum energy.

Figure 4: (a) Dark and illuminated current density vs. voltage characteristics of GaN (60 nm)/1.5 InxGa1-xN (x = 0.11 or 0.21)/DBP/MoO3 (15 nm)/Al (100 nm) organic-inorganic QW devices. (b) External quantum efficiencies of x = 0.11 and 0.21 QWs from 0 V to -5 V in 1 V steps. The x = 0.21 device has lower EQE as compared to the device with x = 0.11 due to a larger conduction band offset of the InGaN with GaN, resulting in increased electron confinement in the QW. The InGaN spectral peaks are observed between wavelength – 475 nm, while the photoresponse between

400

475 – 650 nm is due to DBP.

Figure 5: Electroluminescence spectra of GaN (60 nm)/1.5 nm InxGa1-xN (x = 0.11 or 0.21) /4,4′-bis(N-carbazolyl)-1,1′-biphenyl (CBP)/MoO3 (15 nm)/Al (100 nm) OI QW devices at temperature T = 294 K and applied voltage, Va = 6 V, and at T = 10 K and Va = 12 V. Spectrally resolved emission from bulk CBP and InGaN are observed at 600 nm (~ 2.1 eV) due to the trapped HCTEs, that are bound states between electrons in surface trap states on the nitride surface and delocalized holes in the CBP. The HCTE spectra (shown in grey) are averaged to remove the Fabry–Pérot microcavity modes. The HCTE spectral shapes and position are independent of temperature at T >10 K, and voltage. At T = 10 K the spectral peak has an abrupt blue shift.

21

ACS Paragon Plus Environment

Nano Letters 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

Figure 6: Voltage dependence of HCTEQW dissociation yield in GaN (60 nm)/1.5 nm InxGa1xN

(x = 0.11 or 0.21)/DBP/MoO3 (15 nm)/Al (100 nm) QW devices vs. electric field, F in the

QW. The exciton dissociation efficiency is described by Poole-Frenkel emission (lines show fits to the data). The fits give β = 1.0x10-3 ± 0.2x10-3 (V/cm)-1 (x = 0.11) and 1.2x10-3 ± 0.2x10-3 (V/cm)-1 (x = 0.21). The change in HCTEQW binding energy for the two In compositions,

= 42 ± 12 meV, determined from the change in the intercept at F = 0

matches the calculated

= 38 meV.

22

ACS Paragon Plus Environment

Page 22 of 29

Page 23 of 29

+

(

%""

!"#

+E=/?(@4JB

$""

$#"

Nano Letters 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

Page 24 of 29

6JL'&K

!"# $%7J?'&K

?J>'&K

(J='&K

!"#* O'7JR'&K

E"FC$ LJ('&K

7J>

""& 345

)*+ 9$NG/%*/&

P ;45 (JI'&K

!"()%%O'(JL'&K H-M)*(.M+ M'O'7J(( M'O'7J6(

>J>'&K

@AB?FCD

,(J>'-82 >JQ'&K =J?'&K

7J6'&K

!"' $%7J('&K

!"#$%&'6 ACS Paragon Plus Environment

Page 25 of 29 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

Nano Letters

H-A%#*-"U ,)*+2

(V;EW

!

(V;EW

H-7J6()*7J=Q+ ,(J>'-82

(7'-8'S*T"$G

B%#*-"U ,3452

!"#$%&'? ACS Paragon Plus Environment

XY

Nano Letters

+

)

*

.>'K

H-)*+ ,+ O'7J6(2

7'K

) %

)* *

Z"#:/

$

!)

+

)*

!(

)*

3*%[

!'

)*

# !

XY

(

!&

)*

345

'

)*!% )*!$ !"

H-)*+ ,+ O'7J((2

+

)*

+A0B?C+6!+D+*E)); ;45 +A0B?C+6!+D+*E();

,-,+./0

)*

(

+

&"

)* '

,-../01+2/03415+6789:7 (;

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

Page 26 of 29

&

!"#

!#

!$

!%

!&

!' !( !) 1?@/+6