The Crystal Hotel: A Microfluidic Approach to Biomimetic Crystallization

A "crystal hotel" microfluidic device that allows crystal growth in confined volumes to be studied in situ is used to produce large calcite single crystals with predefined crystallographic orientation, microstructure, and shape by control of the detailed physical environment, flow, and surface chemistry. This general approach can be extended to form technologically important, nanopatterned single crystals.

prepared lithographically fabricated silicon master and cured in a ventilated oven at 60 °C for 120 min. Inlets and outlets (1 mm in diameter) were punched in the cured PDMS substrate, which was then plasma cleaned 1 min, Harrick Plasma Cleaner RD-002), along with piranhacleaned glass slides. The Crystal Hotel was then assembled by gently pressing the plasma exposed surfaces (glass and PDMS) together and finally transferring the assembled device to the oven for a further 5 min to strengthen the PDMS-glass bonding. Devices were used immediately after fabrication, both without further functionalization and after functionalization with self-assembled monolayers, as described below.

Functionalization of the Glass Substrate in the Crystal Hotel:
The glass slide forming the base of the crystal hotel was also functionalized with a self-assembled monolayer (SAM) to change the surface chemistry and thus direct the orientation of the precipitating crystals. This was carried out by depositing alkoxyaminosilane monolayers on the interior of the device and then chemically modifying the end-groups. In order to achieve this, the chips must be exceptionally clean and the surfaces fully hydroxylated via plasma cleaning. 1% APES was diluted with 95% methanol and 4% acetic acid and then injected into the freshly prepared PDMS chips. The surface reaction was stopped after 20 min by rinsing the chips with methanol 3 times. The end group modification was achieved as follows. In order to generate carboxylate end groups, APES was coupled with docosanedioc acid (DCDA) (APES-DCDA).
0.27 mM of DCDA (1 equiv, MW = 370.6) and 2.8 equivalents of DIEA (132 µL, MW = 129.3) was added to 104 mL of 2:1 chloroform/DMF and stirred for 3 h to dissolve. Then 0.1 equiv of HOBT (MW = 135) and 0.1 equivalent PyBOP (MW = 520) were added as coupling agents. The carboxylate activation reaction was allowed to proceed for 30 min prior to injecting into the PDMS chips. After soaking overnight in the coupling solution, the chips were rinsed twice with 1:1 chloroform/DMF, twice with a 1:1 chloroform/ methanol mixture, and once with methanol. [2] Crystallization within Crystal Hotel Devices: Calcium carbonate was precipitated within the confines of the hotel rooms by delivering the reagents (CaCl 2 solution and NH 3(g) + CO 2(g) gasresulting from the decomposition of (NH 4 ) 2 CO 3(s) ) to the device using two Harvard Apparatus PHD 2000 syringe pumps, one loaded with CaCl 2 solution and the other with (NH 4 ) 2 CO 3 powder (Figure 1a). The device was initially flushed with CaCl 2 solution (5 mM) through the first feed channel for 5 min at a flow rate 10 µL min -1 to remove any impurities in the channels, after which time the flow rate was reduced to 1 µL hr -1 . The system was then left undisturbed for 10 min to establish a stable liquid flow. After this, the gas (NH 3(g) + CO 2(g) ) was pumped through the second feed channel at a rate of 1-10 µL min -1 . The NH 3(g) is of particular importance as it raises the solution pH, thus facilitating the precipitation of CaCO 3 . [3] The reactions were allowed to continue for 16-264 h, after which time pumping was discontinued and the delivery tubes removed. Experiments were also performed where additives were added to the 5 mM calcium solution to concentrations of 5 mM MgCl 2 or 1 µg mL -1 poly(acrylic acid) (PAA).
The use of multiple reaction chambers placed successively in a single design was chosen as it provides us with the opportunity to study CaCO 3 nucleation and crystal growth under a decreasing range of supersaturation across the separate chambers in a single experiment. This is achievable as the gas (NH 3(g) and CO 2(g) ) feeding pressure drops (∆p) with increasing channel length as given by Equation 1.
Where L is the length of channel, D h is the dimension of channels, µ m is the mean fluid velocity, ρ is the fluid density and f the Darcy friction factor. [4] In the current case we have a distance of 200 µm between adjacent chambers (center to center). This results in a ~7 fold drop in gas pressure from R1 (closest to the inlet) to R8, Figure 1e. The pressure drop was calculated by measuring the arc length of a liquid in the feeding structure when applying a constant gas pressure ( Figure S1). The existing pressure drop directly governs the velocity of gas (NH 3(g) and CO 2(g) ) permeation though the thin PDMS layer to each reaction chamber as is evident from Equation 2. Here (P) is permeability factor of PDMS for a specific gas, p 2 is the gas feeding pressure, p 1 is the gas permeate pressure, l is the PDMS membrane thickness and N is the steady-state gas flux through the PDMS "membrane". [5] = ( − ) Now assuming that the gas delivery to the chambers consist of 3 sequential steps. The rapid saturation of the gas phase with CO 2(g) and NH 3(g) in the syringe, subsequent transportation of CO 2(g) and NH 3(g) across the PDMS membrane, and lastly the diffusion of "CO 2 and NH 3 " across the gas-liquid interface into the solution, with the intermediate step being the slowest given the diffusion coefficients of ammonia and carbon dioxide in water are 1.64×10 −5 cm 2 s -1 and 1.92×10 −5 cm 2 s -1 , [4] while 7.08×10 −10 cm 2 s -1 and 2.1×10 −10 cm 2 s -1 in PDMS respectively we can study crystallization under different supersaturation regimes by simply adjusting the gas feeding pressure (by varying the gas flow rate) while maintaining the "Ca 2+ " liquid flow rate, here set to 1 µL h -1 .
Characterization: Crystal morphologies were determined in situ in the crystal hotel devices using polarized light microscopy using a Nicon Eclipse LV100 equipped with a circular rotating specimen stage to facilitate orientation studies and using scanning electron microscopy (SEM). Samples for SEM were prepared by immersing the microfluidic devices in ethanol for 3 h to weaken the PDMS to glass bonding, before peeling the PDMS from the glass. This procedure leaves the CaCO 3 crystals adhered to the glass substrate. The glass substrate was then rinsed with ethanol to remove residual PDMS and reagent solutions before drying in air. The dried glass was mounted on an SEM stub using adhesive conducting pads, and images were acquired using a FEI Nova NanoSEM 650 after gold coating. Crystal polymorphs were determined by Raman microscopy, using a Renishaw 2000 inVia-Raman microscope equipped with a 785 nm diode laser. Electron backscatter diffraction (EBSD) was also performed to determine the orientation of the calcite crystals on the glass substrate.
The crystals were examined using an FEI Quanta 650 environmental SEM with a Centaurus  solution and the other with (NH 4 ) 2 CO 3 powder which releases NH 3(g) + CO 2(g) gas on decomposition. b) Transparent "Crystal hotel" devices were fabricated from PDMS (polydimethylsiloxane), were bonded to glass slides and were studied under light microscopy. Figure S2. Illustration of the gas feeding pressure drop with increasing channel length and chamber number (R1-R8). The arrow indicates the gas flow direction. The relative feeding pressure drop is calculated by measuring the arc length (dashed red arc) across the subsequent chambers. To do this, we firstly filled all the channels with water and then pumped gas at flow rate of 10 µL min -1 to squeeze the water out of the chip. The gas will also squeeze the water in the arc area due to the compressibility and permeability of PDMS.