Synthesis of in situ purified iron nanoparticles in an electrochemical and vibrating microreactor: study of ramified branch fragmentation by oscillating bubbles

Abstract

We report on a novel concept of microreactor for the synthesis of metal nanoparticles (nP). The principle is to make grow by electrocrystallization long metallic ramified branches, made of metal nanocrystals, in a Hele-Shaw cell. The activation of an integrated vibrating element (PZT disk) induces the fragmentation of these fragile branches and so the nanocrystal release. One advantage is the possibility to flush the branches (and so the nP), prior to fragmentation, to avoid purification step. This principle is applied here to the synthesis of iron nP and focus is put on the branch fragmentation. High speed visualisations highlight the key role of H$_2$ bubbles (co-formed during the branch growth). An effective fragmentation is obtained only using a square signal with which the initial bubbles coalesce. The resulting large bubbles exhibit shape oscillations and they induce microstreaming. This latter brings the branches close to bubble surface, where they are fragmented into fine particles. The required initial coalescence events are explained by a “dancing bubbles” effect made easier to achieve because initial small bubbles are excited at their resonance frequency by the sufficiently stiff signal steps. Transmission electron microscopy (TEM) reveals that dendritic particles, $\sim$ 2 $\upmu$m long and $\sim$ 1 $\upmu$m wide (broken secondary dendrites) with very high specific surface, and needle-like particles, $\sim$ 200 nm long and $\sim$ 20 nm in diameter (broken tertiary dendrites), are produced. A force balance, between the mechanical constraint applied by the fast flow near bubble surface and the material resistance, allows highlighting the key role of shape oscillations in the breakage of the dendrites.

Appendix: Estimation of the rise time of the depth fluctuations \(\epsilon (t)\)

The current I(t) flowing through the PZT is the sum of a capacitive current $I_c(t)=c_0{\text {d}}V/{\text {d}}t$ ($c_0$ being the capacitance and V the voltage at the PZT terminals) and a current $I_p(t)$, relative to the (direct and reverse) piezoelectric effect $I_p(t) \propto {\text {d}}S/{\text {d}}t$, S being the strain of the PZT material (Arnau 2008) considered as an average value here. The derivation of the proportionality factor requires the integration of the fundamental piezoelectric relations considering the bending of the PZT disk and the metallic support as well as the location of the clamps. Here, we restrict the analysis and simply consider that the PZT surface bending, measured by $\epsilon (t)=e(t)-e_0$, is proportional to S, $I_p(t)$ is thus $\propto {\text {d}}\epsilon /{\text {d}}t$. Consequently, from the simultaneous measurements of V(t) and I(t) and knowing (or measuring) the value of $c_0$ = 20 nF, the form of the idealized depth fluctuations signal can be determined: $e(t)-e0 \propto \int _{0}^{t} (I(\tau )-I_c(\tau )){\text {d}}\tau$.

For the square signals, these measurements (with a filled cell) have shown that the PZT surface displacement is limited by the PZT response and a rise time $t_{\text {r}}=0.1 V_{\text {pp}}$ (with $t_{\text {r}}$ given in $\upmu$s and $V_{\text {pp}}$ in volts) has been determined (the rise time of the amplifier = 0.0005 $\upmu$s/V in open loop).