Nanosecond electron pulses in the analytical electron microscopy of a fast irreversible chemical reaction

We show how the kinetics of a fast and irreversible chemical reaction in a nanocrystalline material at high temperature can be studied using nanosecond electron pulses in an electron microscope. Infrared laser pulses first heat a nanocrystalline oxide layer on a carbon film, then single nanosecond electron pulses allow imaging, electron diffraction and electron energy-loss spectroscopy. This enables us to study the evolution of the morphology, crystallography, and elemental composition of the system with nanosecond resolution. Here, NiO nanocrystals are reduced to elemental nickel within 5 µs after the laser pulse. At high temperatures induced by laser heating, reduction results first in a liquid nickel phase that crystallizes on microsecond timescales. We show that the reaction kinetics in the reduction of nanocrystalline NiO differ from those in bulk materials. The observation of liquid nickel as a transition phase explains why the reaction is first order and occurs at high rates.


Nanosecond electron pulses in the analytical electron microscopy of a fast irreversible chemical reaction
Shyam K. Sinha 1,a , Amir Khammari 1,a , Matthieu Picher 1 , Francois Roulland 1 , Nathalie Viart 1 , Thomas LaGrange 2 , Florian Banhart 1 * To compare the activation energies of the fast reduction reaction after laser excitation with isothermal quasi-static conditions, the same system (i.e., a NiO layer on a carbon membrane) was heated in a standard TEM heating stage. Imaging, electron diffraction, and EELS were carried out with a continuous electron beam. From EEL spectra at different temperatures in the range 570 -630 K, the quantitative composition of the system was determined as a function of time. Supplementary  Figure 1 (see below) shows the NiO layer before and the Ni crystals after slow reduction at 630 K. Supplementary Figure 2 shows the temporal evolution of the composition as derived from a quantitative EELS analysis of the oxygen edge at different temperatures. An activation energy for the isothermal reaction of 1.4 eV is obtained. Extrapolating the line to 2000 K (Suppl. Fig. 2b), which corresponds to the reaction temperature after laser pulses, the measured reaction rate of the fast reduction (red dot) is in good agreement with the measurements under isothermal conditions (black dots).

Heating of a NiO layer on silicon nitride
The heating experiment was also carried out with a similar NiO layer of the same thickness (20 nm) and crystallite size on a Si 3 N 4 film. No carbothermal reaction is possible in this case. Supplementary Figure 3 shows TEM images and the according diffraction patterns before and after heating for several minutes at different temperatures. Supplementary Figure 4 shows an EELS before and after heating at 1070 K. Although a certain decomposition of NiO is visible, there is still a considerable percentage of oxygen up to 1070 K.

Supplementary Note 2: Loss of carbon in the IR laser pulses
To find out whether carbon loss due to ablation in the intense laser pulse has occurred, EEL spectra from a bare carbon membrane before and after a laser shot were taken. The loss of counts in the inelastic part of the spectrum relative to the zero-loss peak indicates a carbon loss of approximately 10%. Since the carbon membrane in the actual experiment is covered with the NiO layer on one side, half of the amount, i.e., a loss of 5% of the carbon layer by ablation can be assumed.

Heating
Quantitative estimates: The absorption of IR in NiO is very small, therefore heating of the NiO layer occurs by heat conduction from the hot carbon film during and after the laser pulse. The temperature change dT/dt at the surface of the NiO layer with thickness d is , where T is the temperature difference between the carbon film and the surface of the NiO layer, λ the thermal conductivity, C the heat capacity, and the density of the NiO layer. The temperature as a function of time is then . ( Assuming a temperature T carbon = 2000 K of the carbon film, the specimen has heated by 1800 K within 1 ns. After heating, heat losses occur by the contribution of reaction enthalpy (see main text), heat radiation at high temperature, and lateral heat conduction through the layers.
Radiative heat losses follow dQ/dt =   A T 4 and are the dominant mode of heat transfer at high temperatures ( is the Stefan-Boltzmann constant,  the emissivity and A the specimen surface). For the radiative cooling of the layer, we obtain a time span t between the initial and final temperatures T 1 and T 2 : Between the melting temperatures of NiO (2257 K) and Ni (1728 K) and assuming an emissivity of 0.8, cooling of the system by radiation would need approximately 200 µs. Assuming a slightly lower starting temperature, this is in accordance with the measured timescale for crystallization of the Ni particles of approximately 100 µs.
Heat losses by lateral heat conduction are less important due to the large heated area (150 µm in diameter) and the small thickness of the layers (altogether 50 nm). Due to the complicated geometry of the layer system on a Cu grid, a precise calculation is almost impossible; however, under some simplifying assumptions (no Cu grid, laser heats a cylindrical volume of the specimen), it is found that tens of milliseconds would be needed to cool the center of the specimen by 500 K. Therefore, the cooling after the reaction should be dominated by radiation.