Dedoping of Intraband Silver Selenide Colloidal Quantum Dots through Strong Electronic Coupling at Organic/Inorganic Hybrid Interfaces

Colloidal quantum dot (CQD) infrared (IR) photodetectors can be fabricated and operated with larger spectral tunability, fewer limitations in terms of cooling requirements and substrate lattice matching, and at a potentially lower cost than detectors based on traditional bulk materials. Silver selenide (Ag2Se) has emerged as a promising sustainable alternative to current state-of-the-art toxic semiconductors based on lead, cadmium, and mercury operating in the IR. However, an impeding gap in available absorption bandwidth for Ag2Se CQDs exists in the short-wave infrared (SWIR) region due to degenerate doping by the environment, switching the CQDs from intrinsic interband semiconductors in the near-infrared (NIR) to intraband absorbing CQDs in the mid-wave infrared (MWIR). Herein, we show that the small molecular p-type dopant 2,3,5,6-tetrafluoro-7,7,8,8-tetracyanoquinodimethane (F4-TCNQ) can be used to extract electrons from the 1Se state of MWIR active Ag2Se CQDs to activate their intrinsic energy gap in the SWIR window. We demonstrate quenching of the MWIR Ag2Se absorbance peak, shifting of nitrile vibrational peaks characteristic of charge-neutral F4-TCNQ, as well as enhanced CQD absorption around ∼2500 nm after doping both in ambient and under air-free conditions. We elucidate the doping mechanism to be one that involves an integer charge transfer akin to doping in semiconducting polymers. These indications of charge transfer are promising milestones on the path to achieving sustainable SWIR Ag2Se CQD photodetectors.

The two-band k.p model originally demonstrated for HgTe CQDs by Lhuillier et al. 1 was recently adapted to Ag2Se CQDs by Scimeca et al., 2  where Einter is the interband energy gap between the 1Sh and 1Se states, Eintra is the intraband gap between the 1Se and 1Pe states, EG is the bulk bandgap, Ep is the Kane parameter, ħ is the reduced Planck's constant, m0 is the free electron mass,  1 =   ,  1 = 4.49  , and R is the radius of the particle.Utilizing a bulk bandgap of 0.07 eV and Kane parameter 8.7 eV, 3 the latter based on fitting of the model to experimental size and absorbance data, 2 the interband gap for Ag2Se CQDs with a certain intraband gap can be predicted.Specifically, for our MWIR active Ag2Se CQDs with an intraband energy gap of ~5 μm/0.248eV, the interband gap would be in the SWIR (~2 μm/0.62 eV).

Section 2: F4-TCNQ concentration in doped CQD films
The elemental ratios were determined based on SEM-EDS measurements using a TM3000 Hitachi tabletop SEM with a Bruker EDS System and QUANTAX 70 software.
The Ag2Se CQD thin-films were drop cast on sapphire substrates, and ligand exchanged and doped with F4-TCNQ in an identical procedure to the samples used for optical characterization, as described above.The absorbance spectra in the range 2500 -9000 nm were taken in a Thermo Scientific Nicolet iS50 FT-IR spectrometer with Omnic software to ensure comparable development of the spectra to Figure 4a) and Figure 5a).
Using SEM-EDS, elemental ratios were determined at three different spots on the films and averaged for further calculations of dopant concentration.Converting from at% elemental fluorine (F) and Ag to F4-TCNQ/Ag2Se CQD ratios, the following assumptions were made: 1.The radius of Ag2Se CQDs with peak intraband absorption at 5.1 μm is 2.6 nm (in accordance with a two-band k.p model). 1,2 The Ag2Se CQDs exist in a tetragonal crystal structure at room temperature with lattice parameters a=b=0.706nm and c=0.498 nm. 4 3.Each F4-TCNQ molecule contains four F atoms. 5 Based on the above assumptions, the volumes of the tetragonal unit cell and the Ag2Se CQD can be compared, yielding 297 unit cells per CQD and a total of 1188 Ag atoms per CQD.This enables the calculation of F4-TCNQ/Ag2Se molecular and CQD ratios for 0, 1, 2, and 3 drops of added 1 mg/mL F4-TCNQ in IPA, as shown in

Section 3: Determination of Fermi level of Ag2Se CQDs
The Fermi level, EF, of our NIR Ag2Se:EDT CQDs was determined to -4.286 eV relative to vacuum based on KPFM measurements, Figure S.4.This is a good match with the reported EF of 5 nm Ag2Se CQDs (-4.3 eV). 6Our obtained EF is also in the middle of the reported band gap of NIR Ag2Se CQDs, 7     dodecanethiol (DDT) ligands on a calcium fluoride (CaF2) window before and after sequential doping with F4-TCNQ dissolved in chloroform (1 mg/mL).Monoanion peak pair formation can be observed around 800 nm, although only at extremely high doping concentrations, causing aggregation of particles.XRD experiments confirmed beginning aggregation and transition to the orthorhombic phase, which has been reported to occur at particle sizes approaching 40 nm. 4 It was hypothesized that the long DDT carbon chain ligands prevented efficient charge transfer until high concentrations of dopants were reached, which then resulted in particle aggregation.Shorter ligands are expected to reduce this issue, thus we employed 1,2-ethanedithiol (EDT) in the following experiments.
Additionally, the chloroform solvent diluted the film, as illustrated in  In the remainder of this section, additional absorbance spectra for MWIR-active Ag2Se CQD thin-films are presented, illustrating the doping effect on ligand exchanged (EDT) thin-films, utilizing alternate solvents for F4-TCNQ, varying F4-TCNQ concentration, and F4-TCNQ doping through soaking instead of dripping.In all cases, quenching of the MWIR peak was observed, however, the interband gap feature at ~1900 nm was only observed with the original long-chain ligands, as shown in Figure 4 and

Section 5: Doping of bulk Ag2Se
Doping of bulk Ag2Se was investigated in order to decouple any nanoscale effects affecting the molecular doping of this semiconductor system by F4-TCNQ.Bulk Ag2Se thin-films were fabricated based on a method inspired by Webber and Brutchey, 13,14 through a cation exchange process from copper (I) selenide (Cu2Se) direct thin films.
Briefly, 100 mg Cu2Se was dissolved in 2 mL ethylenediamine and 0.2 mL 1,2ethanedithiol, spin coated (1800 rpm, 60 sec) onto 1x1 cm clean glass substrates and annealed at 350 °C for 1 hour.The cation exchange process consisted of soaking the Cu2-xSe films in 0.01M silver nitrate (AgNO3) in methanol for 10 min, followed by 45 sec rinse in neat methanol and drying for 2 min on a hot plate at 50 °C.The cation exchanged films were annealed at 350 °C for 30 min.Finally, 60 nm gold was thermally evaporated through a shadow mask onto all four corners of the substrates as electrical contacts.All fabrication steps took place in a N2-filled glovebox.
The bulk film samples were exposed to 1 mg/mL F4-TCNQ in IPA or chloroform dropwise, and room temperature sheet resistance, Rsheet, and Seebeck coefficient, ⍺, were measured between each new exposure.Rsheet was measured in a standard 4-probe van-der-Pauw setup, while the ⍺ was determined in a homemade setup using small Peltier units (CUI Inc.) to provide temperature gradients and Keithley 2400 source meters and Keithley 2000 multimeters to provide currents sources and measure voltages.
Custom Labview programs were used to control the instruments and to process and save the data.
The Seebeck coefficient for the as-fabricated Ag2Se samples was on the order of -40 μV/K and the Rsheet was on the order of 76 Ohm/sq.This is as expected for an unoptimized sample fabricated through this method, 14 the negative ⍺ coefficient confirming the n-type carrier transport in this semiconductor.After exposure to a single drop of 1 mg/mL F4-TCNQ in IPA, the Rsheet increased by 19% and the absolute value of ⍺ increased by 111%.This can be explained through the decrease in available charge carriers, n, and demonstrates the doping impact of F4-TCNQ for bulk Ag2Se.Upon exposure to another drop of 1 mg/mL F4-TCNQ in IPA, the Rsheet decreased slightly, indicating that the limited surface area of bulk Ag2Se had already been saturated with F4-TCNQ, and that the additional IPA only served to redissolve some of the F4-TCNQ -  The polarity of the solvent has been suggested to play an important role in the interaction between F4-TCNQ and the host material. 11Therefore, chloroform, which has a slightly lower polarity than IPA, was utilized as a solvent in a 1 mg/mL F4-TCNQ solution.Similarly to for IPA, exposure of the Ag2Se film to a single drop of F4-TCNQ in chloroform yields an increase in both the Rsheet and the Seebeck coefficient, however slightly lower than with IPA (10% and 20% for Rsheet and Seebeck, respectively, compared to 19% and 111% using IPA) The Rsheet stayed fairly stable during storage for one day in a N2 filled glovebox, and increased further with exposure to another drop of F4-TCNQ in Overall, these results demonstrate that doping of Ag2Se by F4-TCNQ is possible, not only for quantum confined nanoparticles, but also for bulk thin-films, although the doping effect might be limited by available surface area in bulk films.

Section 6: Other doping strategies
HgS CQDs are known for their stable n-type doping in the ambient, 18 with strong absorbance features in the MWIR and NIR.Jeong et al. demonstrated that the spectral response of these CQDs was sensitive to surface treatment.Exposing the CQD film to S 2-ions, the MWIR absorption was removed and the NIR edge red shifted, while with subsequent treatment with Hg 2+ , the MWIR absorption was reintroduced and the NIR edge blue shifted.The authors claim that this was the first demonstration of stabile surface doping controlled in ambient, and suggest that the doping can be attributed to rigid shifts of the energy bands with respect to the environment.Dopant concentration during the solution phase doping step was ~7 mM for both S 2-and Hg 2+ .
In the report by Yarema et al. 19 the authors claim to utilize a donor-acceptor heterojunction to fabricate a lateral (near) IR photodetector.Mixing Ag2Se CQD with [6,6]phenyl-C61-butyric acid methyl ester (PCBM) in a 1:4 weight ratio and depositing a blended film on a glass substrate with interdigitated ITO electrodes, the authors measured a responsivity of 200 mA/W at 900 nm, with photoresponse up to 1300 nm for a device with 13.5 mm 2 active area.The authors insinuate that an electron transfer is taking place between the CQDs and PCBM, but do not report any change in the spectral response of the CQDs as a result of the blending.With a reported work function of -4.2-4.3 eV for PCBM, 20 the reported valence band edge of NIR Ag2Se of -4.9 eV by Graddage et al., 7 and our measured EF of Ag2Se of -4.3 eV, the driving force for electron transfer from the CQD to PCBM seems low.Rather, it could be the improved mobility due to excess PCBM in the blend, as well as the high applied bias (100 V) that yields a working Ag2Se sensitized photodetector with such high responsivity.
shown in Figure S.1 and expressed mathematically in Eq.S.1 and S.2:

Figure S. 5 ,
resulting in a theoretical Type II band alignment with respect to F4-TCNQ.

Figure S. 5 .
Figure S.5.Type II band edge energy alignment for NIR Ag2Se CQDs 7 and F4-TCNQ. 8The dotted line represents the Fermi level, EF, determined by KPFM, as shown in Figure S.4.

Section 4 :
Figure S.6 shows absorbance spectra for MWIR Ag2Se CQDs with 1- Figure S.7, thus the need to switch to an orthogonal solvent such as isopropyl alcohol (IPA) or acetone.

Figure S. 6 .Figure S. 7 .
Figure S.6.UV-Vis-NIR (a) and FTIR (b) absorbance spectra of MWIR Ag2Se:DDT before and after doping with F4-TCNQ in chloroform (1 mg/mL).The addition of F4-TCNQ creates a monoanion radical pair as seen in (a), but only at extremely high concentrations that aggregate the particles.

Figure 5
Figure5in the main text.Soaking the film in the same solution as used for dripping did not yield the same absorbance response.It was also observed that the F4-TCNQ concentration and choice of solvent seem to play an important role.F4-TCNQ has a relatively low solubility in many solvents,9,10 and from the organic semiconductor literature, it has been shown that the choice of solvent can affect the doping mechanism.11

Figure S. 8
Figure S.8 shows the full absorbance spectra in the Vis/NIR/SWIR range for MWIR Ag2Se:TOP CQDs before and after exposure to 1 mg/mL F4-TCNQ in IPA under air-free (Figure S.8a,c) and ambient (Figure S.8b,d) conditions.

Figure S. 8 .
Figure S.8.Absorbance spectra in the visible/NIR/SWIR range for MWIR Ag2Se:TOP CQDs before and after exposure to 1 mg/mL F4-TCNQ in IPA under air-free (a) and ambient (b) conditions.c) Near UV range under air-free conditions showing indication of dianion (F4-TCNQ 2- ) peak at low doping concentrations.d) Near UV, visible and NIR range under ambient conditions.

Figure S. 9 12 Figure S. 9 .
Figure S.9 shows the effect of brief air-exposure for a Ag2Se:TOP CQD thin-film doped under air-free conditions.No significant change was observed due to air-exposure, contrary to what has been observed for CQDs exposed to another molecular dopant.12

Figure S. 11
Figure S.11 shows the effect of F4-TCNQ doping on ligand exchanged MWIRactive Ag2Se:EDT CQD films under ambient conditions.In the visible/NIR/SWIR range (Figure S.11a), no evidence of charge-neutral F4-TCNQ or F4-TCNQ -monoanion peaks are observed.A new feature at ~1100 nm appears after ligand exchange from TOP to EDT, which red-shifts along with a feature at ~500 nm with increasing F4-TCNQ doping.In the MWIR range (Figure S.11b), complete quenching of the MWIR peak ~5 μm is observed at a dopant concentration of ~128 F4-TCNQ molecules per Ag2Se CQD, with shifting of the F4-TCNQ nitrile stretch with increasing dopant concentration, indicating partial ICT.

Figure S. 12
Figure S.12 shows the effect of F4-TCNQ doping on ligand exchanged MWIRactive Ag2Se:EDT CQD films through soaking in 0.0036M F4-TCNQ in IPA under air-free conditions.In the visible/NIR/SWIR range (Figure S.12a), evidence of excess F4-TCNQ is observed after doping, but no F4-TCNQ -monoanion peaks.In the MWIR range (Figure S.12b), complete quenching of the MWIR peak ~5 μm is observed after 20 sec soaking in F4-TCNQ solution, with minimal shifting of the F4-TCNQ nitrile stretch, indicating a different doping mechanism than ICT.

Figure S. 13
Figure S.13 and S.14 show the doping effect on MWIR-active Ag2Se:TOP CQD and ligand exchanged MWIR-active Ag2Se:EDT CQD films, respectively, through dropwise addition of 0.0036M F4-TCNQ in acetone under air-free conditions.In the visible/NIR/SWIR range (Figure S.13a and S.14a), evidence of excess F4-TCNQ as well as a slight F4-TCNQ -monoanion peak ~876 nm are observed after doping.In the MWIR range (Figure S.13b and S.14b), complete quenching of the MWIR peak ~5 μm is observed at ~5 F4-TCNQ molecules per Ag2Se CQD, with minimal shifting of the F4-TCNQ nitrile stretch, indicating a different doping mechanism than ICT.Acetone appears to have a higher solubility towards F4-TCNQ than IPA.

Figure S. 13 .
Figure S.13.Absorbance data for Ag2Se:TOP with dropwise addition of 0.0036M F4-TCNQ in acetone under air-free conditions.Spectra for undoped Ag2Se:TOP CQDs are shown as blue curves, and spectra for CQDs with a doping level of ~5 ± 3 F4-TCNQ molecules per Ag2Se CQD are shown as green curves.The doping level was estimated based on the trend in doping concentration in Figure S.3.Solid and dotted red curves illustrate reference spectra for F4-TCNQ and F4-TCNQ -monoanion, respectively.(a) Vis/NIR/SWIR: Slight anion peak at ~876 nm.Excess Slight "dent" ~2000 nm.(b) MWIR-LWIR: Complete quenching of ~5 μm peak at ~5 F4-TCNQ molecules per Ag2Se CQDs.No shifting of nitrile stretch indicates no anion formation.Peaks ~3500 nm are due to C-H stretch in ligands.Curves have been adjusted in the y-direction to better visualize quenching.

Figure S. 14 .
Figure S.14.Absorbance data for Ag2Se:EDT with dropwise addition of 0.0036M F4-TCNQ in acetone under air-free conditions.Spectra for undoped Ag2Se:EDT CQDs are shown as blue curves, and spectra for CQDs with a doping level of ~5 ± 3 F4-TCNQ molecules per Ag2Se CQD are shown as green curves.The doping level was estimated based on the trend in doping concentration in Figure S.3.Solid and dotted red curves illustrate reference spectra for F4-TCNQ and F4-TCNQ -monoanion, respectively.(a) Vis/NIR/SWIR: Slight anion peak ~876 nm.Excess F4-TCNQ.(b) MWIR-LWIR: Complete quenching of ~5 μm peak at ~5 F4-TCNQ molecules per Ag2Se CQD.No shifting of nitrile stretch indicates no anion formation.Peaks ~3500 nm are due to C-H stretch in ligands.Curves have been adjusted in the y-direction to better visualize quenching.

Figure S. 15
Figure S.15 shows the doping effect on ligand exchanged MWIR-active Ag2Se:EDT CQD films through dropwise addition of 0.00058M F4-TCNQ in acetone under air-free conditions.In the visible/NIR/SWIR range (Figure S.15a), evidence of excess F4-TCNQ as well as a slight F4-TCNQ -monoanion peak ~876 nm are observed after doping.In the MWIR range (Figure S.15b), complete quenching of the MWIR peak ~5 μm is observed at ~1 F4-TCNQ molecules per Ag2Se CQD, with minimal shifting of the F4-TCNQ nitrile stretch, indicating a different doping mechanism than ICT.

Figure S. 15 .
Figure S.15.Absorbance data for Ag2Se:EDT with dropwise addition of 0.00058M F4-TCNQ in acetone under air-free conditions.Spectra for undoped Ag2Se:EDT CQDs are shown as blue curves, spectra for CQDs with a doping level of ~1 ± 0.5 F4-TCNQ molecules per Ag2Se CQD are shown as green curves, and spectra for doping level ~5 ± 2.5 F4-TCNQ molecules per Ag2Se CQD is shown as black curve The doping levels were estimated based on the trend in doping concentration in Figure S.3.Solid and dotted red curves illustrate reference spectra for F4-TCNQ and F4-TCNQ -monoanion, respectively.(a) Vis/NIR/SWIR: Slight anion peak ~876 nm at lowest doping concentration.Excess F4-TCNQ.(b) MWIR-LWIR: Complete quenching of ~5 μm peak at ~1 F4-TCNQ molecule per Ag2Se CQD.No shifting of nitrile stretch indicates no anion formation.Peaks ~3500 nm are due to C-H stretch in ligands.Curves have been adjusted in the y-direction to better visualize quenching.
monoanions.The Rsheet stayed fairly constant during storage for two days in a N2-filled glovebox, and the Rsheet did not change much upon 10 sec soaking in neat IPA.Annealing at 300 °C for 30 min reversed the initial change in Rsheet.The relative change in Rsheet is shown in Figure S.16.F4-TCNQ has a melting point of 291 °C, 15 but has been reportedto sublimate even below 200 °C,16,17 thus annealing at 300 °C was expected to vaporize the F4-TCNQ.The bulk Ag2Se thin-films had already been exposed to a temperature 350 °C prior to doping.

Figure S. 16 .
Figure S.16.Relative change in Rsheet for a bulk Ag2Se thin-film on glass substrate exposed to 1 mg/mL F4-TCNQ in IPA.

Figure S. 17 .
Figure S.17.Relative change in Rsheet for a bulk Ag2Se thin-film on glass substrate exposed to 1 mg/mL F4-TCNQ in chloroform.