Continuous injection synthesis of indium arsenide quantum dots emissive in the short-wavelength infrared

With the emergence of applications based on short-wavelength infrared light, indium arsenide quantum dots are promising candidates to address existing shortcomings of other infrared-emissive nanomaterials. However, III–V quantum dots have historically struggled to match the high-quality optical properties of II–VI quantum dots. Here we present an extensive investigation of the kinetics that govern indium arsenide nanocrystal growth. Based on these insights, we design a synthesis of large indium arsenide quantum dots with narrow emission linewidths. We further synthesize indium arsenide-based core-shell-shell nanocrystals with quantum yields up to 82% and improved photo- and long-term storage stability. We then demonstrate non-invasive through-skull fluorescence imaging of the brain vasculature of murine models, and show that our probes exhibit 2–3 orders of magnitude higher quantum yields than commonly employed infrared emitters across the entire infrared camera sensitivity range. We anticipate that these probes will not only enable new biomedical imaging applications, but also improved infrared nanocrystal-LEDs and photon-upconversion technology.

The QD size is calculated based on the absorbance peak using a sizing curve. The curves initially overlap quite well, suggesting that material added to the growth solution is added to existing particles. After about 45 minutes, the particle growth rate appears to slow down and precursor material added to the growth solution does not seem to be exclusively added to existing particles. Supplementary Figure 4| Synthesis of InAs QDs from (iPrDMSi) 3 As. The synthesis of InAs QDs from tris(isopropyldimethylsilyl)arsine ((iPrDMSi) 3 As) was adapted from the continuous injection synthesis described in the main article. (iPrDMSi) 3 As was synthesized as previously described. [1] 1 mmole In(Acetate) 3 was added to 4 mmoles oleic acid and 5 mL ODE and the dispersion was degassed at 115°C under vacuum to form In(Oleate) 3 . The final pressure after 1 hour was measured to be lower than 10 mtorr. The reaction atmosphere was switched to nitrogen and heated to 295°C. Subsequently, 0.05 mmoles of (iPrDMSi) 3 As in 1 mL TOP were injected. After 10 minutes a continuous injection of a 0.168 M solution of (iPrDMSi) 3 As in ODE was started. The injection speed was set to 2 mL h −1 for the first 20 minutes and then switched to 0.15 mL h −1 for another 575 min, such that the final amount of arsenic precursor injected into solution equals 0.404 mmoles. The PL peak continuously shifted from 765 nm and a FWHM of 172 meV (10 minutes) to 1233 nm and a FWHM of 139 meV (605 minutes). Left: Absorption and PL spectra taken during the growth of InAs QDs from (iPrDMSi) 3 As. Right: Position and FWHM of the PL peak as a function of the amount of arsenic precursor added to the solution.  3 As. The synthesis of InAs QDs from tris(trimethylsilyl)arsine) ((TMSi) 3 As) was adapted from the continuous injection synthesis described in the main article. (TMSi) 3 As was synthesized as previously described. [1] 1 mmole In(Acetate) 3 was added to 4 mmoles oleic acid and 5 mL ODE and the dispersion was degassed at 115°C under vacuum to form In(Oleate) 3 . The final pressure after 1 hour was measured to be lower than 10 mtorr. The reaction atmosphere was switched to nitrogen and heated to 295°C. Subsequently, 0.05 mmoles of (TMSi) 3 As in 1 mL TOP were injected. After 10 minutes a continuous injection of a 0.168 M solution of (TMSi) 3 As in ODE was started. The injection speed was set to 2 mL h −1 for the first 30 minutes and then switched to 0.15 mL h −1 for another 480 min, such that the final amount of arsenic precursor injected into solution equals 0.301 mmoles. The PL peak continuously shifted from 751 nm and a FWHM of 172 meV (10 minutes) to 1149 nm and a FWHM of 131 meV (510 minutes). During the reaction, the formation of a small metallic solid on the side of the flask was noted and the growth solution turned slightly turbid. The resulting aliquots after the injection of 0.238 mmoles arsenic precursor were thus filtered (200 nm syringe filter) before the PL spectrum was measured. We attribute the formation of precipitate in the solution to a decomposition reaction of (TMSi) 3 As at these elevated temperatures. Left: PL spectra acquired during the growth of InAs QDs from (TMSi) 3 As. Right: Position and FWHM of the PL peak as a function of the amount of arsenic precursor added to the solution.  Ref [2] Bawendi 2016

Supplementary Note 1| Inhomogeneous Linewidth Broadening through Shape Anisotropy
The results from this paragraph and Supplementary Fig. 2-3 were first discussed and published in the Ph.D. thesis of D.K. Harris in 2014 and were included for completion. [5] To investigate the slowdown in growth rate after the particles reach sizes of about 4.5-5 nm (absorbance peaks approximately around 1000-1100 nm), we compared the particle growth rate with the rate of material added to the reaction solution. During the CI synthesis described in Fig. 2a in the main text (1 mL h −1 ) aliquots were removed periodically and characterized by absorbance and photoluminescence spectroscopy to monitor the evolution of QD size and size distribution. As described in the main article, avoiding a secondary nucleation step is key to obtaining a narrow size distribution. To measure whether all precursor material added to the solution is incorporated into existing crystals, we used: • UV absorbance: Absorbance of short wavelength light scales with InAs material in solution and should be independent of particle size. [6,7] • QD size from sizing curve. To obtain the QD size from the first excitonic feature in the absorption spectrum, the sizing curve of Yu et al. was employed. [8] Particle size and total InAs concentration are normalized to the aliquot taken immediately before the beginning of the continuous injection and plotted in Supplementary Fig. 2. These data clearly show that particle growth is commensurate with precursor addition initially. During this time, the PL linewidth drops rapidly before becoming very broad after about 65 minutes. The minimum FWHM achieved during this synthesis occurs at about 45 minutes, and corresponds to the point in Supplementary Fig. 2 where the material added and the QD volume diverge.
The most obvious interpretation of this result is that material added after 45 minutes results in the formation of new particles. To investigate the nature of the broadening of the linewidth during growth, TEM was used to characterize the aliquots taken at 10 minutes, 45 minutes, and 140 minutes. Supplementary Fig. 3 shows that the initial 10 minute aliquot and the 45 minute aliquot have approximately spherical shapes with a mean size measured to be 4.5 nm from TEM images. However, the particles in the 140 minute aliquot have asymmetric shapes. These shapes seem to have lobes that do not share the symmetry of the zincblende crystal structure. The average lobe diameter was measured to be approximately 3.5 nm. This is consistent with a growth trajectory in which the particles of 4.5 nm diameter coalesce with smaller particles that are not fully absorbed into the parent to form a spherical shape. Therefore, it appears that the broadening in the optical spectra is driven in part by shape inhomogeneity.

Supplementary Note 2| Continuous Injection Synthesis of InP QDs
The results from this paragraph and Supplementary Fig. 6 were first discussed and published in the Ph.D. thesis of D.K. Harris in 2014 and were included for completion. [5] Although the focus of our work so far has been on InAs QDs, we have also attempted to make InP QDs using a continuous injection strategy. However, due to the chemistry of InP synthesis, we had to slightly adjust our strategy. In contrast to the growth of InAs QDs, [9] InP QD growth is known to be negatively affected by the presence of excess carboxylic acid, [10,11] but we believe that the presence of some acid is beneficial to the continuous injection process as it may preferentially digest smaller particles. Therefore, we used a two-pot approach, where we synthesized and purified InP QDs with an absorbance peak at 525 nm for use as seeds. We then redispersed these seeds in an In(Myristate) 3 solution for continued growth.
For seed growth, 1 mmole of In(Acetate) 3 was mixed with 3 mmoles of myristic acid and 5 mL of ODE. This solution was heated under vacuum at 115°C for 60 minutes to remove acetic acid displaced during the formation of In(Myristate) 3 . The resulting clear solution was heated to 150°C under argon, and 0.5 mmoles of (TMSi) 3 P dissolved in 1 mL of TOP were injected. The temperature controller was immediately switched to 275°C for growth. The solution was cooled 7 minutes after injection. 0.5 mL of the growth solution was removed to atmosphere, and the residual particles were purified by adding acetone and centrifuging the turbid solution. Hexane was used to redisperse the pellet of InP QDs at the bottom of the centrifuge tube. This was repeated once more, and the resulting stock solution was diluted for measurement of its absorbance spectrum and found to have an OD of 100 at 300 nm. A solution of 1 mmole of In(Myristate) 3 in ODE was prepared by again adding 1 mmole of In(Acetate) 3 , this time with 3.1 mmoles of myristic acid. Acetic acid was removed by evacuating at 115°C. The reaction vessel was allowed to cool and 630 mg of the solution of InP cores was added to the reaction vessel and vacuum was applied briefly to remove the hexanes as the reaction solution was heated to 250°C. The reaction timer was started as the solution reached 150°C. Aliquots were removed at t = 0 min, t = 10 min, and t = 18 min, prior to starting the continuous injection at t = 18 min. Absorption spectra were taken of these aliquots, and they reveal that the InP cores were somewhat unstable in the reaction solution. The absorption peak blueshifted from 525 nm to 510 nm at t = 18 min and the half width at half max (HWHM) of the absorption peak increased from 111 meV for the initial seeds to 132 meV for the aliquot removed at t = 18 min. Supplementary Fig. 6 shows that as soon as the continuous injection is started, we note an onset of QD growth (the absorption maximum redshifts about 80 nm) and a narrowing of size distribution (the HWHM of the absorption maximum narrows by about 20 meV). However, it is apparent that as the reaction proceeds, the particle growth slows and the features become less defined as it did with InAs QDs at faster injection speeds. TEM images suggest that both shape inhomogeneity as well as size inhomogeneity contribute to the spectral broadening observed.

Supplementary Note 3| Determination of intrinsic QYs in Water.
As phase transfer techniques are known to negatively affect QYs for QD systems, for example by altering QD ligand chemistry and surface passivation, QDs typically exhibit lower QYs in aqueous media, compared to organic solvents. However with regards to SWIR emissive QDs, the QY in aqueous media may be further decreased by both near-field interactions, such as non-radiative energy transfer to nearby water molecules and far-field effects such as reabsorption of emitted photons through the solvent. While enhanced near-field interactions, such as FRET, can be considered intrinsic properties of the fluorophore-solvent system, reabsorption processes strongly depend on the measurement setup, sample concentration, cuvette length, or geometry of the integrating sphere, which impedes a comparison across literature.
To extract the intrinsic, setup-independent QY of SWIR QDs in aqueous solution corrected for reabsorption events, we measured PL lifetimes and the extrinsic QYs of four QD samples in water (saline solution, H 2 O) and heavy water (D 2 O) (see Supplementary Fig. 11 a- To exchange H 2 O for D 2 O, 1 mL of a 1 mg mL −1 QD solution in H 2 O was transferred to a centrifugal filter (Amicon Ultra, Ultracel 30K MWCO) and centrifuged for 8 min at 7 krpm so that almost all H 2 O was removed. 1 mL of D 2 O was added and the sample was centrifuged for another 8 minutes at 7 krpm. This procedure was repeated 3 times. Afterwards the sample was redispersed in 1 mL of D 2 O.
Supplementary Fig. 11 Fig. 11 c,d). Comparing the QY in D 2 O with the calculated, intrinsic QY in H 2 O we find a relative QY reduction due the near-field interaction between QDs and H 2 O of 3%, 15%, 44% and 50% for samples emitting at 970 nm, 1110 nm, 1300 nm, and 1430 nm, respectively. The increase in non-radiative energy transfer with increasing emission wavelength is in good qualitative agreement with the increase in overlap between QD emission and the water absorption band at 1450 nm ( Supplementary Fig. 11 e).