Architected Metal Selenides via Sequential Cation and Anion Exchange on Self-Organizing Nanocomposites

Shape-preserving conversion reactions have the potential to unlock new routes for self-organization of complex three-dimensional (3D) nanomaterials with advanced functionalities. Specifically, developing such conversion routes toward shape-controlled metal selenides is of interest due to their photocatalytic properties and because these metal selenides can undergo further conversion reactions toward a wide range of other functional chemical compositions. Here, we present a strategy toward metal selenides with controllable 3D architectures using a two-step self-organization/conversion approach. First, we steer the coprecipitation of barium carbonate nanocrystals and silica into nanocomposites with controllable 3D shapes. Second, using a sequential exchange of cations and anions, we completely convert the chemical composition of the nanocrystals into cadmium selenide (CdSe) while preserving the initial shape of the nanocomposites. These architected CdSe structures can undergo further conversion reactions toward other metal selenides, which we demonstrate by developing a shape-preserving cation exchange toward silver selenide. Moreover, our conversion strategy can readily be extended to convert calcium carbonate biominerals into metal selenide semiconductors. Hence, the here-presented self-assembly/conversion strategy opens exciting possibilities toward customizable metal selenides with complex user-defined 3D shapes.


S2. Scaled up growth of BaCO 3 /SiO 2 nanocomposites
A solution of BaCl 2 (7.4 g, 30 mM) in 300 mL of water was added to a solution of Na 2 SiO 3 (1.6 g, 13 mM) in 1200 mL of water. This solution was shortly stirred and poured in a metal tray (30x50x10 cm) to maximize the surface area in contact with air, while keeping at least 1 cm of depth to the solution. The solution was left for 1.5 hours with the tray covered polycarbonate lid which is perforated with 0.3 mm holes. The resulting nanocomposites floating on the meniscus were separated from the solution via vacuum filtration. For further exchange reactions, the microstructures were instantly removed from the filtration paper with a spatula and transferred directly into the exchange solution ( Figure S2).

S3 Conversion to CdCO 3 nanocomposites
Cadmium chloride (458 mg, 50 mM) was dissolved in 50 mL demineralized water. A substrate containing fresh BaCO 3 nanocomposites was placed in the solution for 12 minutes. The resulting CdCO 3 nanocomposites were washed in two demineralized water baths followed by an acetone bath. The structures were analyzed using SEM, EDS and XRD ( Figure S3). EDS analysis showed a Cd:Si ratio of 75.8:24.2 and XRD Scherrer analysis reveals a slight increase in crystal domain size as compared to the BaCO 3 from 18 nm to 21 nm.

S4 Determining the barium content in CdCO 3 nanocomposites
CdCO 3 nanocomposites were placed in a single zone tube furnace. The pressure in the furnace was reduced to <1 mbar. The temperature was raised to 500°C for 12 hours after which the oven was allowed to cool. The resulting nanocomposites were deprived of cadmium due to the sublimation of CdO. The composition of the structures was analyzed using EDS and they were found to consist of 98-99% of silica with 1-2% barium. Using this ratio, we can calculate that the original barium content was approximately 0.2-0.4%.

S5 Conversion to CdO nanocomposites
CdCO 3 nanocomposites were placed in a single zone tube furnace. The furnace was purged of oxygen and filled with nitrogen gas till a pressure of 500 mbar was reached. This pressure was maintained with a 50 sccm flow of N 2 as the furnace was heated to 290°C for 4 hours. The resulting nanocomposites were characterized by SEM, EDS and XRD ( Figure S4).

S6 Conversion to CdSe nanocomposites
CdCO 3 nanocomposites were placed in a single zone tube furnace. An alumina boat was added to the single zone tube furnace. The furnace was purged of oxygen and filled with nitrogen gas until a pressure between 20 and 50 mbar was reached. This pressure was maintained with a 6 sccm flow of N 2 as the furnace was heated to 500°C at a rate between 0.5 and 50°C/min and maintained at this temperature for 1 to 12 hours. The structures used for the conversion to Ag 2 Se were created at 20 mbar with a heating rate of 50°C/min and the temperature was maintained for 1 hour. The

S7 Conversion to Ag 2 Se nanocomposites
Silver nitrate (AgNO 3 , 4 g, 0.5 M) was dissolved in 50 mL methanol at 55°C and filtered to remove any undissolved AgNO 3 . A substrate containing CdSe nanocomposites was placed in the solution for 120 minutes. The resulting Ag 2 Se nanocomposites were washed in methanol and allowed to dry by air. The structures were analyzed by SEM, EDS, and XRD ( Figure S6). Using Oxford Instrument's True-Q ® deconvolution algorithm the atomic ratio of elements was determined to be 59%Ag, 33.9%Se, 5.6%Si, and 1.5% Cd with cadmium falling outside the certainty range. An additional deconvoluted linescan was preformed to show the difference in ratio between Ag and Cd ( Figure S7 The same reaction was attempted using small amounts of acetonitrile in methanol to increase the solubility of AgNO 3 . Unfortunately, the addition of acetonitrile prevented the formation of orthorhombic Ag 2 Se and an amorphous result was obtained. The color for this amorphous material was gray instead of black for the orthorhombic Ag 2 Se.

S8 Conversion of bio sample to CdCO 3
A sand dollar was bleached using potassium hydroxide (KOH, 1M). It was placed in a solution of CdCl 2 (1.4 g, 50 mM) in water (150 mL) for 120 minutes. Afterwards, the sand dollar was placed in a water bath to remove any remaining cadmium ions for 12 hours. Finally, it was dried using an acetone bath. The sample was analyzed by SEM and EDS.

S9 conversion of CdCO3 bio sample to CdSe
The CdCO 3 sand dollar was placed in a single zone tube furnace. An alumina boat was added to the single zone tube furnace. The furnace was purged of oxygen and filled with nitrogen gas until a pressure of 20 mbar was reached. This pressure was maintained with a 6 sccm flow of N 2 as the furnace was heated to 500°C at a rate of 50°C/min and maintained at this temperature for 1 hour. The resulting sample was analyzed using SEM and EDS.

S10 Bandgap determination
CdSe powders produced using varied reaction conditions were analyzed using UV-VIS on a PerkinElmer UV/VIS/NIR spectrometer Lambda750 with a 150 mm InGaAs integrating sphere. To calculate the optical bandgap, Tauc plots were constructed. Since the bandgaps were in the visible spectrum, the formula for a direct gap (n=⅟₂) was used ( Figure S8). The linear fit is defined as: And the corresponding intersection with the horizontal axis, e.g. the bandgap energy, is then defined as:

= -
The influence of the pressure and heating rate on the bandgap was determined to be dominated by the heating rate with only a minor contribution caused by the pressure and the cooling rate.

S11 Scherrer equation
CdCO 3 powders were produced in two batches (B1 and B2) and subsequently converted into CdSe under different reaction conditions (Table S1). Using the Scherrer equation the crystal domain size was calculated from the XRD patterns. This analysis shows that for heating rates above 10 °C/min the crystal domain size was mainly controlled by the pressure.

Sample
Grain  Table S1: Influence of reaction conditions on the grain domain size and the bandgap.

S12 investigation of the bandgap shift
To further investigate the cause of the measured bandgap shift based on the heating rate we performed XRD analysis in the range of 2θ = 10-80° for sample batch B1, CdSe 1 mbar, 5 °C/min and sample batch B1, CdSe 20 mbar, 50 °C/min (See Figure S9). We found insignificant XRD peak shifts (See Table S2) ruling out strain and we did not find significant peak broadening at 2θ = 42°. To identify the presence of surface defects we measured photoluminescence (PL) (See Figure S10). PL spectra were taken by exciting a sample with a 405 nm Thorlabs S1FC405 diode laser through a WiTEC Alpha300 SR confocal microscope. The laser is focused on the sample with a P5-305A-PCAPC1 optical fiber. The photoluminescence was measured with a UHTC 300VIS WiTEC spectrometer. We detected a PL signal at 720 nm, consistent with literature, for fast heating CdSe and we observe near complete quenching for the slow heating CdSe.  Table S2: XRD shift of CdSe formed with a slow heating rate vs CdSe with a fast heating rate, showing minimal peak shifts. Figure S10: Photoluminescence spectra of CdSe with a slow heating rate (Blue) and CdSe with a fast heating rate (Orange). S13 Calculating the expected microscopic volume change [1] The expected volume change of the nanocrystals can be expressed as a change in crystal lattice volume: and as the unit cell volumes of the starting material and final material, and Z S and Z F as their respective number of formula units per cell.
Part of our structures consists of inert silica. To compensate for this matrix we need to calculate the volume fraction (λ) of the silica matrix. Λ is calculated using the following formula: We compute the and using the following procedure: First we determine the atomic 2 3 ratio between the carbonate phase and the silica phase via EDS. This is approximately 0.8:0.2. The atomic ratio is multiplied by the molecular weight and divided by the density. The density of the silica phases from similar sol-gel reactions found in literature is 1.28±0.23 g/cm³. [2] With this we calculate λ as follows = 0.2 * 60/(1.28 ± 0.23) 0.2 * 60/(1.28 ± 0.23) + 0.8 * 197.34/4.29 = 0.20 ± 0.03 Finally, we calculate the expected microscopic volume change (ε) as a function of the expected volume change of the nanocrystals following: . ε = (1 -λ)θ + λ