Sub-nanosecond Infrared Photodetection using III-V Colloidal Quantum Dots

: Colloidal quantum dots (CQDs) are promising materials for IR light detection due to 15 their tunable bandgap and solution processing; but to date, the time response of CQD IR 16 photodiodes has been inferior to that provided by Si and InGaAs. We reasoned that the high 17 permittivity of II-VI CQDs leads to slow charge extraction due to screening and capacitance; 18 whereas III-Vs – if their surface chemistry could be mastered – offer a strong covalent character for 19 low permittivity and fast operation. In initial studies, we found that existing covalent character led 20 to imbalanced charge transport in InAs, the result of unpassivated surfaces and uncontrolled heavy 21 doping. We report surface management using amphoteric ligand coordination and find that it 22 addresses simultaneously the In and As surface dangling bonds. The new InAs CQD solids combine 23 high mobility (0.04 cm 2 V -1 s -1

Infrared photodetection underpins applications in medicine and bioimaging, information technology, machine vision, and security 1,2 .Emerging technologies such as autonomous driving and augmented reality rely on Light Detection and Ranging (LiDAR) based on time of flight (ToF) 3 .This requires sensitive and ultrafast photodetection of infrared light with sub-ns resolution 4 .Today, this is achieved in the nearinfrared (NIR) using indirect bandgap silicon detectors-limited by silicon's low absorption coefficientand, at longer wavelengths, using epitaxially-grown semiconductors such as III-Vs and Hg 1-x Cd x Te 5,6 .
Colloidal quantum dots (CQDs) are of interest by their low-temperature solution processing, which allows them to be integrated in post-processing with a silicon electronics front-end [7][8][9][10] .Their bandgap is size-tuned over a wide range of wavelengths.PbS, for example, has a widely-programmable absorption onset covering the visible and short-wavelength infrared (SWIR) 11,12 ; however, its high permittivity, stemming from its ionic character -ɛ r =180 for bulk PbS 13 -slows charge extraction both for bulk 14 and CQD photodiodes 15 due to screening and capacitance effects.
Indium arsenide (InAs) CQDs can be tuned in a similar spectral range as PbS CQDs, and offer the prospective advantage of a covalent lattice and hence lower permittivity 16,17 .This, however, comes with a challenge: the surface in InAs CQDs is charge-imbalanced, leading to poor passivation and heavy doping, as surface states pin the Fermi level near the conduction band minimum.Much effort has been paid to improving mobility by decreasing interdot distance via ligand exchanges and surface treatments in CQD solids 18,19 , but III-V CQDs require a new approach in order to neutralize CQD charge surface states and reduce trap density.
Here we present a new surface passivation strategy that addresses charge imbalance and passivation in InAs CQD solids for infrared photodetectors.We introduced InBr 3 passivants to replace native insulating oleic acid ligands, thereby providing surface passivation and charge transport simultaneously.We found that InBr 3 is amphoteric, dissociating into an X-type ligand (Br -) that passivates In dangling bonds; and into a Z-type (InX 2 + ) ligand 25 that passivates As dangling bonds.We incorporate N,N-dimethylformamide (DMF) as a coordinating agent to stabilize otherwise unstable Br -and InX 2 + passivants.We assess the copassivated InAs surfaces using a suite of spectroscopies and density functional theory (DFT) calculations.
The resulting InBr 3 -InAs CQD solids achieve a mobility value of 0.04 cm 2 V -1 s -1 , >10 times higher than in halide-exchanged InAs solids 18 ; and a low dielectric constant of 6.3 -a near 4x advance compared to PbS CQD solid counterparts.The corresponding photodiode devices achieve a 30% external quantum efficiency (EQE), a responsivity of 0.22 A W -1, and measured detectivity of 10 11 cmHz 1/2 W -1 at the excitonic peak (940 nm).Ultrafast transient photocurrent (TPC) experiments reveal a fall time of 300 ps, equivalent of >1 GHz bandwidth.This is the first demonstration of efficient III-V CQD photodiodes and the fastest solution-processed infrared photodiode reported, with over a 100-fold improvement compared to the best CQDs photodiodes 15,26 .We first synthesized InAs CQDs (semisphere, Supplementary Fig. 1) using a modified approach via the continuous injection 27,28 .As-synthesized InAs CQDs are capped by insulating oleic acid ligands, which need to be removed for electrical coupling.The oleic acid capped InAs (OA-InAs) in octane show low photoluminescence quantum yield (PLQY < 1 %) 27 , which we attribute to the presence of surface defects as negatively charged As dangling bonds 29 that are not passivated using oleic acid.
To increase understanding of the nature and role of the dangling bonds (DBs) and guide in the design of a surface exchange strategy to passivate InAs surfaces, we studied the zinc-blende InAs structure using DFT calculations and estimated the thermodynamic stability of their surfaces (see Supplementary text).
The crystal facets exposed, including ( 111) and (100), are polar and terminated with positively charged In and negatively charged As.
The nature of zinc-blende InAs structure demands the simultaneous passivation of both In and As dangling bonds for efficient ambipolar optoelectronics -i.e. for high performance in devices whose operation relies on electrons and holes simultaneously, as distinct from unipolar devices such as fieldeffect transistors (FETs).The success of X-type on positively charged surfaces suggests 30 that halides could be promising candidates for In passivation.Z-type ligands binding as a neutral two-electron acceptor (a Lewis acid) have been demonstrated to passivate phosphide deep traps leading to enhanced photoluminescence efficiency 31 .This prompted us to design a dual passivation strategy that would seek to address In and As dangling bonds using X-type and Z-type ligands, respectively.
We hypothesized that metal halide salts such as InBr 3 could be candidates for this as when dissolved in certain polar solvents such as DMF, they would dissociate into the needed X-type (Br -) and Z-type (InBr 2 + )) ligands 25 .To assess the potential of these ions as co-passivants, we performed DFT calculations on different InAs surfaces (see Suppl.Info for details).The calculation results indicate that Br -passivates In-terminated surfaces whereas InBr 2 + addresses As-terminated surfaces (Supplementary Fig. 2, 3, and 4).
Based on these calculations, we designed a one-step, two-phase solution exchange method to replace OA long ligands with InBr 3 salts assisted by ammonium acetate (AA), which enables the removal of the original long organic ligands (OA) 32 .Before the ligand exchange, InAs CQDs are dispersed in octane and InBr 3 (0.1 mol L -1 , 0.18g in 5 mL) and AA (0.04 mol L -1 , 0.023g in 5 mL) are pre-dissolved in DMF.
During the exchange, InAs CQDs transfer from nonpolar octane layer to polar DMF solution, which indicates that long OA ligands are replaced by InBr x ligands.To track the removal of long ligands in the exchanged film, we carried out Fourier transform infrared spectroscopy (FTIR) measurements (Fig. 2a).OA-InAs CQDs show typical CH resonances at ~2900 cm -1 and C=O vibrations owing to OA on CQD surface at ~1540 cm -1 .After the ligand exchange, the OA signals disappear.The C=O signal at ~1640 cm -1 is attributed to DMF complexes, which differs from the peak position of free DMF at 1675 cm -1 .
To assess the role of DMF complex, we performed synchrotron (In L 3 -edge) X-ray absorption near edge structures (XANES) (Fig. 2b).XANES features of In are not very sensitive to anion substitutions but to the coordination number 33 .InBr 3 -InAs films show a similar peak position as pure InBr 3 films deposited from DMF solution, but a shift compared to that of OA-InAs films.The peak position difference suggests that In in InBr 3 -InAs and InBr 3 have sixfold coordination (In connects to 6 atoms) whereas OA-InAs only has only fourfold coordination 33 .
Considering the presence of DMF, its potential interactions with the passivants and the sixfold coordination of In, we calculated the stabilization energy of Br -/DMF and In-Br/DMF co-passivants.In the case of zinc-blende InAs structure, there are 1.5 DBs (electrons from As or holes from In) at (100) surfaces and 0.75 DBs at (111) surfaces 20 .These sites are hard to passivate as the ligands can only provide an integer number of carriers.These results suggest a larger stabilization energy by adding DMF coordination for both In-rich and As-rich surfaces (Fig. 1, Supplementary Fig. 2-4).The lower surface energies through Br/In-Br + DMF co-passivation suggest that the dangling bonds on InAs CQD surface are passivated by DMF complexes.
We carried out thermogravimetric analysis (TGA) (Fig. 2c) to study the stability of InAs CQDs and gain insights into the composition of the final CQD solids.We analyzed the decomposition of ligands and solvent at different temperatures for the exchanged CQDs comparing TGA traces of pure InBr 3 , AA-InBr 3 , OA-InAs, and InBr 3 -InAs: the weight loss of 1% in the range from 50℃ to 158℃ is attributed to the free DMF solvent (boiling point 153℃) in InBr3-InAs solids; the weight loss of ≈2.5% in the range from 162℃ to 240℃ is attributed to the DMF from the decomposition of DMF-InBrx complex; the weight loss of ≈3.5% in the range from 240℃ to 340℃ may be the AA/InBr 3 complex decomposition (similar to the decomposition of AA-InBr 3 ); the final weight loss of ≈36.5% from 340℃ to 420℃ is attributed to the decomposition of InBr 3 and residual OA.The decomposition of DMF-InBr x complex agrees with the InBr 3 -DMF co-passivation mechanism 34 .
We performed InBr 3 -DMF ligand exchange on OA-capped InAs (OA-InAs) CQDs.After ligand exchange, the In-to-As ratio is close to 1:1 (Fig. 2f) vs. 3:1 before exchange.This high In-to-As ratio of OA-InAs is attributed to the free In-oleate in the solution that was not removed during CQD purification, evidenced by the decreased In signal peak width after ligand exchange (Fig. 2d) and the high C-to-As ratio before exchange.After InBr 3 -DMF ligand exchange, the C:In ratio decreased to ≈2:1.Fig. 2e shows the As signal after ligand exchange without any As signal oxidized.The appearance of N and Br signal suggests the existence of DMF and Br-in InBr 3 -InAs films, supporting the results above.
Based on the successful ligand exchange, we characterized the absorption of CQDs before and after the ligand exchange in solution (Fig. 3a).OA-capped CQDs show an excitonic peak at 916 nm with a peak-to-valley ratio of ≈2.6.After ligand exchange, the excitonic peak redshifts to 928 nm and the peakto-valley ratio decreases to ≈2.2.This indicates that quantum confinement is preserved after exchange.To assess the interparticle distance and the necking produced after ligand exchange, we carried out grazing-incidence small-angle X-ray scattering (GISAXS) measurements (Fig. 3b).After InBr 3 ligand exchange, the interdot distance decreased from 4.6 nm to 3.6 nm, as extracted from the azimuthally integrated coherence peak.This is consistent with OA replacement by inorganic ligands.
To characterize charge transport in the exchange solid, we measured mobility using FET devices 35 .
FET output characteristics reveal the n-type transport enhancement mode for InBr 3 -InAs CQD films (Supplementary Fig. 5).The carrier mobility is 0.032±0.003cm 2 V -1 s -1 in the linear regime (μ lin ) and 0.040±0.005cm 2 V -1 s -1 in the saturation regime (μ sat ) is, (Fig. 3c), over 10 times higher than the two-step ligand exchange 18 .The current on/off ratio is only about 10, indicating that the thickness of the CQD layer exceeded the depth of the accumulation channel formed in the CQD solid upon applied gate bias and that the electron concentration is high in InBr 3 -InAs CQD solids.
We extracted the carrier concentration through mobility and conductivity, obtaining a value of 1.11x10 17 cm -3 and 1.39x10 17 cm -3 from μ lin and μ sat , respectively.This suggests that the InBr 3 -InAs CQD solids are n-type doped.We confirmed the n-type transport polarity of InBr 3 -InAs CQD films using ultraviolet photoelectron spectroscopy (UPS), which revealed a conduction band minimum (CBM) at -4.62 eV, a valance band maximum (VBM) at -5.92 eV, and a Fermi level (E F ) at -4.79 eV (Supplementary Fig. 8a).This agrees with the carrier concentration obtained from FET results.
To assess the optical properties of exchanged InAs films, we measured the complex refractive index and compared it with PbS CQD films for reference.We found that the real part of the refractive index is higher in PbS CQD films (Fig. 3d, 3e), which could be attributed to its higher permittivity.To characterize the dielectric constant at electrical frequencies, we measured the capacitance in a diode device configuration (Supplementary Fig. 6 and Fig. 3f).We obtained ɛ r = 34.8 for PbS CQDs and a much lower value (ɛ r = 6.3) for InAs solids.Considering the energy levels of InAs CQD solids, we designed a device architecture consisting of ITO / ZnO / InAs / Au.Unfortunately, this arrangement resulted in poor diode behavior and high dark currents (Supplementary Fig. 7a).Reasoning that this structure may have produced back injection of electrons from InAs into Au, we added a layer of 10 nm of MoO 3 as an electron blocking layer between InAs and Au (Fig. 4a, Supplementary Fig. 8b).We observed a decrease in dark current by 3 orders magnitude to ≈70 nA cm -2 at 1.0 V reverse bias (Fig. 4b) accompanied by a significant increase in EQE from 9% to 30% at 0 V bias (Fig. 4c), a responsivity of 0.22 A W -1 (Supplementary Fig. 9) and measured detectivity of 10 11 cmHz 1/2 W -1 at the excitonic peak (940 nm, Fig. 4d).This is the highest EQE reported for an InAs CQD photodetector.
We then characterized the time response of the InAs CQD photodiodes after ultrafast photoexcitation using 100 fs laser pulses.We designed an electrical footprint that enables us to measure EQE/ JV curves and incorporate small pixels with diameters from 1400 µm to 100 µm (Supplementary Fig. 10).
To assess the impact of device capacitance, we studied the response time for different pixel areas.
The fall time decreases from 50 ns for 1.5 mm 2 to 300 ps for 0.01 mm 2 pixels (Fig. 4e).The magnification of transient photocurrent spectra for the lowest pixel area is shown in the inset.PbS CQDs devices show a similar trend but ≈2 orders of magnitude longer fall time compared to InAs diodes (Fig. 4f).This can be ascribed to the higher capacitance in PbS compared to InAs QDs.These results also suggest that we have not reached the mobility-limited transport regime even at the smallest pixel area, as fall time keeps decreasing with decreasing device area and does not reach a plateau (Fig. 4f).
We then measured the -3 dB cut-off frequency using a vertical-cavity surface-emitting laser (VCSEL) light source modulated at different frequencies (Supplementary Fig. 11).The response stays constant until 25 MHz and crosses -3 dB at about 150 MHz (Fig. 4f).TPC results for the same diameter pixel (0.2 mm 2 ) reveal a fall time of ≈2 ns, which would correspond to a 3 dB cut-off frequency of ≈175 MHz.We note that due to the low power of the VCSEL and focusing limitations, we could not obtain cutoff frequency for smaller pixels.Based on the TPC decay of the fastest pixels, we can expect a ≈1 GHz bandwidth for 0.01 mm 2 devices.
In sum, we have introduced an amphoteric passivation strategy to co-passivate both In and As on InAs CQD surfaces providing passivation and charge transport for low permittivity CQD solids.We revealed the important role of DMF as a coordinating agent that stabilizes both X-and Z-type passivants.
Consequently, we were able to achieve for the first time efficient InAs CQDs photodiodes showcasing an EQE of ~30% at 0 V and a specific detectivity of 10 11 cmHz 1/2 W -1 at the near-infrared.The low permittivity and passivation of the InAs CQD solid result in a fall time of 300 ps, the fastest CQD photodiode.Our findings open the door to the realization of high-performance optoelectronic devices based on III-V CQDs.

Supplementary Files
This is a list of supplementary les associated with this preprint.Click to download.

Fig. 1 |
Fig. 1 | Stabilization of InAs CQDs.Scheme of InBr 3 ligand exchange and InAs quantum dot surface configuration according to DFT calculations: a larger stabilization energy was observed by adding DMF coordination in the system for both In-rich and As-rich surfaces.

Fig. 2 |
Fig. 2 | Investigation of passivation using InBr3-DMF complexes.a, Attenuated total reflection ( AT R ) FTIR of InAs CQD films before and after InBr 3 ligand exchange on glass.b, In L 3 -edge XANES spectra of InAs CQD films and InBr 3 films (spin-coated from DMF solution) on Si wafer.c, Thermogravimetric analysis (TGA) of InAs CQDs and InBr 3 salt with/without AA samples.d, In 3d X-ray photoelectron spectroscopy (XPS) signal.e, As 3d XPS signal.f, Elemental ratio of InAs CQD films before and after InBr 3 ligand exchange.

Fig. 3 |
Fig. 3 | Improved transport properties and low dielectric constant of InBr3-InAs CQD films.a, Absorption of InAs CQD solution before and after ligand exchange.b, Azimuthal integration of (GISAXS) patterns of InAs CQD films with GISAXS 2D pattern (inset) of InBr3-InAs CQD film.c, Transfer curves of InBr 3 -InAs CQD film in linear and saturation regimes with the field-effect transistor (FET) device structure (inset).d, The real and e, Imaginary refractive index of InAs compared to PbS QDs.f, Dielectric constant of InBr 3 -InAs and PbI 2 -PbS CQD films.

Figures Figure 1
Figures

Figure 2 Investigation
Figure 2

Figure 3 Improved
Figure 3

Figure 4 Fast
Figure 4