Microscopy and Microanalysis (2018), 24, 647–656

doi:10.1017/S1431927618015465

Original Article

Exploring Photothermal Pathways via in Situ Laser Heating in the

Transmission Electron Microscope: Recrystallization, Grain Growth,
Phase Separation, and Dewetting in Ag0.5Ni0.5 Thin Films
Yueying Wu1,a, Chenze Liu2,a, Thomas M. Moore3, Gregory A. Magel3, David A. Garfinkel2, Jon P. Camden1,
Michael G. Stanford2, Gerd Duscher2 and Philip D. Rack2,4
1
Department of Chemistry and Biochemistry, University of Notre Dame, Notre Dame, IN 46556, USA, 2Department of Materials Science and Engineering,
University of Tennessee, Knoxville, TN 37996, USA, 3Waviks Inc., 10330 Markison Road, Dallas, TX 75238, USA and 4Center for Nanophase Materials Sciences,
Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA

Abstract
A new optical delivery system has been developed for the (scanning) transmission electron microscope. Here we describe the in situ and
“rapid ex situ” photothermal heating modality of the system, which delivers >200 mW of optical power from a fiber-coupled laser diode to
a 3.7 μm radius spot on the sample. Selected thermal pathways can be accessed via judicious choices of the laser power, pulse width, number
of pulses, and radial position. The long optical working distance mitigates any charging artifacts and tremendous thermal stability is
observed in both pulsed and continuous wave conditions, notably, no drift correction is applied in any experiment. To demonstrate the
optical delivery system’s capability, we explore the recrystallization, grain growth, phase separation, and solid state dewetting of a Ag0.5Ni0.5
film. Finally, we demonstrate that the structural and chemical aspects of the resulting dewetted films was assessed.
Key words: in situ TEM, laser heating, TEM instrumentation, optical delivery system
(Received 8 June 2018; revised 19 October 2018; accepted 28 October 2018)

Introduction
Rapid materials synthesis and characterization of materials
systems is critical to advances in a wide variety of applications
from electronic devices to more fuel efficient automobiles. To
accomplish this, one critical pathway is the development of
combinatorial approaches to rapidly synthesize multicomponent
material libraries (Xiang et al., 1995; Klein et al., 2005; Wu et al.,
2011). Another critical need is the development of in situ and
rapid ex situ characterization approaches, so that materials can
be exposed to external stimuli and the subsequent responses can
be measured to elucidate, for instance, the free-energy landscape
of various phase transformations. Among the various char-
acterization techniques (scanning) transmission electron
microscopy [(S)TEM], and its associated spectroscopic com-
plements, is one of the few techniques in which this information
can be gathered with atomic scale resolution. Thus, there has
been a concerted effort to develop various in situ accoutrements
for the (S)TEM (Zheng et al., 2015; Taheri et al., 2016; Zheng &
Zhu, 2017), examples include: in situ heating stages, gas (Jiang
et al., 2018) and liquid cells (Chen et al., 2015), optical delivery
Author for correspondence: Philip D. Rack, E-mail: prack@utk.edu
a
Yueying Wu and Chenze Liu are co-first authors.
Cite this article: Wu Y, Liu C, Moore TM, Magel GA, Garfinkel DA, Camden JP,
Stanford MG, Duscher G and Rack PD (2018) Exploring Photothermal Pathways via
in Situ Laser Heating in the Transmission Electron Microscope: Recrystallization, Grain
Growth, Phase Separation, and Dewetting in Ag0.5Ni0.5 Thin Films. Microsc Microanal 24(6),
647–656. doi: 10.1017/S1431927618015465
© Microscopy Society of America 2018
and collection (Miller & Crozier, 2013; Allen et al., 2017; Joseph
et al., 2017). Furthermore, over the past few decades, a few
groups across the world have developed very complex and
specialized ultrafast electron microscope or dynamic TEM sys-
tems using photo-cathodes that are exposed to short laser pulses
to generate electron beamlets (and single electrons), that syn-
chronously arrive at the sample in time relative to another
pulsed laser directed at the same sample (Kim et al., 2008;
Barwick et al., 2009; Liu et al., 2014; Barwick & Zewail, 2015; Li
et al., 2016; Shorokhov & Zewail, 2016; Kaplan et al., 2017).
In Situ Photon Delivery System Overview
Motivated by the above and the desire to develop a commercially
available and more ubiquitous system useful for a variety of
optical in situ studies, Waviks Inc. has developed a new photon
delivery system which can be mounted on any (S)TEM system.
Figures 1a and 1b are photographs of the system mounted on the
Libra 200 (S)TEM. Figure 1c is a computer-aided design sche-
matic overview of the system with magnified images of the:
Figure 1d the lens assembly subsystem which houses the focusing
optics that images the fiber optic ends; Figure 1e the flange
adaptor, in vacuo shaft carrying the fiber optics, and part of the
x–y–z nanomanipulator; Figure 1f the protective shield for the
individual fiber optics (3 shown in the diagram in purple) which
are cabled and carried via a ~10 m fiber optics to the control box
which houses the laser/light sources and electronic drivers (not

")!"
$"
&&#%))) $"$"$

$%%
   
"!



&



%'&
&"
&
 $
"$
&$ %
"
'%
(
&
&&#%))) $"$"$&$ %
&&#%""$  


 648 Yueying Wu et al.

Figure 1. a, b: Are photographs of the optical delivery system mounted onto the Libra 200 (scanning) transmission electron microscopy. c: Is an overview computer-aided
design schematic of the system with magnified views of the (d) lens assembly end piece, (e) vacuum mounting flange and in vacuo shaft which carries the optical fiber and part
of the x–y–z nanomanipulator, and (f) protective shielding for the fiber optics which are cabled and carried to the control box which houses the laser/light sources and
electronic driving units (not shown). g: Is a screen shot of the Waviks Inc. software which controls the nanomanipulator and laser drivers.

shown). Figure 1g is a screen shot of the Waviks Inc. software
which interfaces with the nanomanipulator and the laser drivers.
This system has been adapted from a system previously developed
for the dual scanning electron and ion microscope (Roberts et al.,
2012), and used for in situ laser-assisted nanoscale electron
(Roberts et al., 2013; Stanford et al., 2015; Lewis et al., 2017) and
ion (Stanford et al., 2016a, 2016b; Stanford et al., 2017) beam
synthesis.
The alpha prototype used in these experiments contains two
optical delivery channels/fibers. Peak powers up to >200 mW
are delivered to the sample from a 785 nm wavelength laser
diode system coupled through a 5 µm mode field diameter
single-mode fiber. The laser is gated by a software-controlled
pulse generator that can vary the laser pulse width from a few
nanoseconds to continuous wave (cw) at repetition rates up to
16 MHz. A second optical channel, containing a 100 µm core
diameter broad spectrum multimode fiber, is available for
coupling to any excitation source in the wavelength range from
200 to 2,100 nm using a standard subminiature assembly fiber
connector. The system is mounted to a 3 axis (± x, y, z)
nanomanipulator for easy focusing to the electron/sample
coincident point (with sample tilted at ~45°) and uses a lens
system to re-image the fiber optics (1 × magnification) at a
working distance of ~10 mm. The working distance is suffi-
ciently long so as to not introduce any charging artifacts when
the optical probe is inserted and aligned, and minimizes re-
deposition of material onto the lens system. The current system
is installed on the energy-dispersive X-ray analysis port, which
is perpendicular to the sample entry and provides convenient
access to the sample via simply tilting the stage. Future options
under development include co-mounting the system with the
aperture strip or integrating the system with the sample entry/
manipulation system. Figure 2a schematically illustrates the end
of the optical probe in proximity (~1 cm) to the TEM substrate
and Figure 2b is a schematic illustrating that a vast array of
thermal pathways can be achieved via different combinations of
laser power, pulse width, number of pulses, and radial position.
For instance, from left to right we can run the laser cw, and
various pulse widths (>1 ns), and various power densities (up to
~500 kW/cm2). Figure 2c are simulated (see methods for
details) time-temperature profiles of the center of the laser spot
for a 200 μs laser pulse at various laser powers irradiating the
Ag0.5Ni0.5 film.
While in a future publication we propose to demonstrate the
system’s ability for studying photoexcitation, the goal here is to
demonstrate the photothermal heating modality via in situ ima-
ging of the recrystallization, grain growth, phase separation, and
solid state dewetting of a Ag0.5Ni0.5 film (see supplemental
information for a table summary of all laser conditions and
associated figures). Furthermore, while our microscope does not
have a pulsed cathode, and is thus limited to longer acquisition
times, we will show that valuable information can be obtained via
a so-called rapid ex situ data acquisition mode, which can reveal
sub-microsecond information. Notably, in all the experiments

")!"
$"
&&#%))) $"$"$

$%%
   
"!



&



%'&
&"
&
 $
"$
&$ %
"
'%
(
&
&&#%))) $"$"$&$ %
&&#%""$  


 Microscopy and Microanalysis
Figure 2. a: Schematic illustrating the end of the optical delivery system which re-images a laser-coupled single mode fiber optical fiber onto the transmission electron
microscopy sample. b: Schematic illustrating that various laser conditions can generate numerous in situ photothermal pathways via different combinations of laser power (up
to 215 mW), pulse width (1 ns to CW), number of pulses and radial position. c: Time–temperature plots from a finite element photothermal simulation of 20-nm thick Ag0.5Ni0.5
film on a 20 nm SiO2 membrane exposed to various laser powers explored in this study.
shown here, no drift correction was done during or after the laser
exposures.
Material and Methods
The AgxNi1 − x system
The Ag–Ni equilibrium phase diagram is interesting as it contains
regions where both the liquid and solid state have limited solu-
bility. At the Ag0.5Ni0.5 composition, above ~2,950 K liquid Ag
and Ni are soluble, whereas between ~2,950 and 1,700 K the two
liquids separate into Ag-rich and Ni-rich phases. Below 1,700 K a
Ni-rich solid solution with <2.5 at% Ag precipitates and is in
equilibrium with the Ag-rich liquid with <5 at% Ni. Finally, near
the Ag melting temperature at 1,234 K, a silver rich solid solution
with <1 at% Ni is in equilibrium with a Ni-rich solid solution of
<1 at% Ag. Below this temperature, the solubility of each phase
gradually decreases for the minor constituent. AgxNi1 − x films
have been studied previously with most of the thin film investi-
gations catalyzed by Berkowitz et al.’s work on heterogeneous
bimetallic giant magnetoresistance observed in immiscible Cu–Co
alloys (Berkowitz et al., 1992). Several studies have investigated
the magnetic (Kitada, 1985) and magnetoresistance properties of
AgxNi1 − x alloys (Kubinski & Holloway, 1995) where a magne-
toresistance value of ~7% was observed in gas flow condensation
grown nanoparticles (Dimesso & Hahn, 1998). Furthermore,
microstructural and magnetic properties of dilute nickel in silver
alloys have been investigated, which revealed that clusters having
fewer than six atoms have a reduced associated magnetic moment
due to charge transfer between the Ni sp and d orbitals (Garcia
Prieto et al., 2015). Of particular relevance to this study, the
evolution of silver rich AgxNi1 − x films at various compositions
and annealing temperatures have previously been studied (Proux
et al., 1999, 2000a, 2000b). For x > 0.85, as-deposited films were
single phase, where all the nickel atoms are substitutional in the
silver lattice forming a supersaturated solution. For
0.65 < x < 0.80, the films consisted of the supersaturated solution
and nanometer-sized nickel grains. Upon annealing, all the silver-
rich regions of the films experienced nickel out-diffusion from the
supersaturated phase, with nickel nucleation and growth for
x > 0.85 and nickel growth for 0.65 < x < 0.80. Finally, the mag-
netic properties of these films were also measured and correlated
to the nickel precipitate size (Nagamine et al., 1999) and a
")!"
$"
&&#%))) $"$"$

$%%
 
649
maximum magnetoresistance value of 2% was realized, which
corresponded to nickel particle diameters of 1.28 nm.
Thin Film Sputtering
Radio frequency (RF) magnetron co-sputtering was used to
deposit an ~20 nm Ag–Ni film of Ag0.5Ni0.5 on three 20-nm thick
SiO2 substrates. The 50 mm diameter Ag and Ni targets had RF
powers of 20 and 70 W, respectively, to induce corresponding
deposition rates of 3 and 2 nm/min. Sputtering was conducted at
a pressure of 5 mTorr with Ar flowing at a rate of 25 sccm. In
order to ensure a uniform film, the samples were rotated for the
duration of sputtering. Energy-dispersive X-ray spectroscopy
(7 keV) using a Zeiss Merlin scanning electron microscope was
used to confirm the film stoichiometry which was estimated to be
Ag0.47Ni0.53.
Electron Energy Loss Spectra (EELS) Experiments
EELS experiments were conducted in a Carl Zeiss LIBRA
(Oberkochen, Germany) 200MC (S)TEM operated at 200 kV. The
STEM images were performed with a convergence semi-angle of
10 mrad, and a collection semi-angle of 15 mrad. All core-loss
EELS spectra were quantitatively analyzed using the Quantifit
software (Mohsin et al., 2018). To quantitatively estimate the
silver fraction in the film, the reference cross section model from
X-ray photo-absorption spectra was applied to fit the core-loss
edge after background subtraction.
Laser Heating Simulations
The temperatures presented in what follows are estimated by
finite element simulations as described below. The simulations
assume a continuous uniform film with the relevant parameters
and assumptions listed below. Importantly, the simulations are
valid for the continuous film and will change when dewetting
occurs, as the incident angle and thickness will change, as will the
thermal conduction, which for the continuous film dominates via
the Ag0.5Ni0.5 film and for isolated islands and particles is limited
by conduction through the SiO2 membrane. High-resolution
pyrometry and/or patterned thermistors are planned in the
future. Importantly, while the laser power is sufficient to sig-
nificantly photothermally heat on thin membranes due to the
limited thermal conduction in effectively the radial direction, the
 
"!




&



%'&
&"
&
 $
"$
&$ %
"
'%
(
&
&&#%))) $"$"$&$ %
&&#%""$  


 650 Yueying Wu et al.

overall power is low and thus it is easily dissipated in the sur- Table 1. Material and Laser Parameters for the Thermal Simulations.
rounding silicon substrate.
Simulation
Simulations of laser-induced heating of the Ag0.5Ni0.5/SiO2 Parameters Value Description
membrane were performed using COMSOL Multiphysics 5.0 a
commercial finite element method software package. The simu- wx 3.7 (µm) Laser 1/e 2 irradiance radius—x
lation used a thickness of 20 nm for AgNi and the SiO2 mem-
wy 5.2 (µm) Laser 1/e 2 irradiance radius—y
branes. The membranes were 50 × 50 μm in length and anchored
onto a Si heat sink, to emulate the geometry of TEM membranes. R 0.69 Reflection coefficient of AgNi at
The expression for heat delivered to the substrate from the laser is 785 nm
derived from the Beer–Lambert law: A 5.52E5 (1/cm) Absorption coefficient of AgNi at
2Ac 785 nm
Qin ðx; y; zÞ = Po ð1Rc Þ Gðx; yÞexpðAc z Þ; (1)
πwx wy P0 12.8–22.2 (mW) Laser power
where P0 is the optical power of the laser, R the reflection coef- Pulse 200 (µs) Laser pulse width
ficient, A the linear attenuation or absorption coefficient, G(x,y)
the Gaussian laser irradiance profile, wx and wy the 1/e 2 irra- Cp(AgNi) 340 [J/(kg*K)] AgNi heat capacity
diance radii of the Gaussian laser profile in the x and y directions,
ρ(AgNi) 9,700 (kg/m3) AgNi density
respectively, and z the depth from the substrate’s surface. wx and
wy were experimentally determined to be 3.7 and 5.2 µm, κ(AgNi) 88 [W/(m × K)] AgNi thermal conductivity
respectively, as described above. A linear 10 ns laser ramp time
Cp(SiO2) 710 [J/(kg × K)] SiO2 heat capacity
was assumed for this simulation and the laser pulse width was
200 μs. All absorption was assumed to occur in the Ag0.5Ni0.5 film, ρ(SiO2) 3,440 (kg/m ) 3
SiO2 density
since the extinction coefficient for 785 nm photons in SiO2 is ~0.
κ(SiO2) 1.0 [W/(m × K)] SiO2 thermal conductivity
The following time dependent heat equation was used to simulate
the heat transfer throughout the Ag0.5Ni0.5 and SiO2 membranes,
as well as the silicon heat sink:
dT range of 100k pulses, where bright-field TEM and selected area
ρCp + ρCp u  ∇T = ∇  ðκ∇T Þ + Qðx; y; z Þ; (2) electron diffractograms (SAED) were periodically collected. The
dt
second experiment uses the same laser conditions (22.2 mW
where ρ is the material density, Cp the heat capacity at constant power and 200 μs pulse width) on a different film area, where this
pressure, u the velocity vector for thermal transport, and κ the time complementary STEM images and EELS were collected and
thermal conductivity. Convective heat transfer to the surrounding correlated to first experiment. The third experiment compares
atmosphere was neglected since irradiation conditions were under STEM images taken at various regions along the radius of films
high vacuum and T0 was defined as 293.15 K. Notably we have exposed to (1) 200 μs–12k pulses and (2) 20 μs–2.5 M pulses both
ignored radiative heat loss in the simulation where including at a 36.5 mW laser power. The fourth experiment illustrates the
radiative heat loss would lower the simulated temperatures and cw modality by capturing high-angle annular dark-field
have a more pronounced effect on higher temperature simula- (HAADF) STEM movies of the dewetting dynamics/timescales of
tions. A backward differentiation formula time-stepping method the Ag0.5Ni0.5 film at four different laser powers from 12.8 to
with strict time steps was used to generate the temporal tem- 22.2 mW. The fifth experiment illustrates the rapid ex situ mod-
perature evolution during laser irradiation. Table 1 reports rele- ality and the system stability as a high magnification pulse study
vant simulation and material parameters used to simulate the from 1–35k pulses was performed at 16.6 mW and 200 μs pulse
temperature temporal evolution induced by the 785 nm laser width. Finally, we illustrate how complementary core-loss and
pulse. The Ag0.5Ni0.5 and SiO2 heat capacitance and thermal low-loss EELS can give useful information about the phase
conductivity were approximated for thin films in accordance to separation and plasmonic activity during various stages of the
the following references (Langer et al., 1997; Lee & Cahill, 1997; solid state dewetting process.
Chen & Hui, 1999). For the Ag0.5Ni0.5 film, heat capacities and Figures 3a and 3b show TEM and STEM images, respectively,
densities were averaged for each element to estimate the values for of the as-deposited ~Ag0.5Ni0.5 film. At room temperature, the
the alloy. The thermal conductivity was estimated based on the Ag0.5Ni0.5 region of the Ag–Ni phase diagram suggests a mixture
literature, which suggests that Ni thermal conductivity is inde- of Ag- and Ni-rich phases with very limited solubility (<1%). The
pendent of thickness (Langer et al., 1997), and at 20 nm, Au as-deposited images (and radially averaged selected area electron
thermal conductivity will be ~20% of the bulk value (Chen & Hui, diffraction pattern in Fig. 3) confirm that the films are nano-
1999), which is applied to the Ag bulk value. granular phase-separated Ag-rich and Ni-rich grains. The bright
spots in the HAADF STEM images are silver-rich regions as
confirmed by EELS (see Supplemental Information). The diffuse
Results and Discussion diffraction rings are attributed to the nanoscale grain size; fur-
Presented below is a series of experiments designed to illustrate thermore, the rapid quenching during the Ag–Ni condensation
some of the photothermal heating capabilities of the in situ laser results in supersaturation of Ni in Ag and Ag in Ni and likely
delivery system. Table 1 in the Supplementary Information file micro-strain. An estimate using Vegards law and based on the
overviews the details of all the figures and laser conditions, but we peak positions of the as-deposited films suggests a composition of
briefly introduce the experiments here. The first experiment Ag0.91Ni0.08 for the Ag-rich phase and Ag0.17Ni0.83 for the Ni-rich
illustrates the laser spot size and the evolution of the Ag0.5Ni0.5 phase. Figures 3c to 3f present a low-magnification series of
film irradiated at 22.2 mW power and 200 μs pulse width over a bright-field TEM images with SAED insets taken as a function of

")!"
$"
&&#%))) $"$"$

$%%
   
"!



&



%'&
&"
&
 $
"$
&$ %
"
'%
(
&
&&#%))) $"$"$&$ %
&&#%""$  


 Microscopy and Microanalysis 651

Figure 3. a: HAADF scanning transmission electron microscopy and (b) transmission electron microscopy (TEM) images of the as-deposited Ag0.5Ni0.5 film. Low-magnification
TEM images of the Ag0.5Ni0.5 film exposed to (c) 1, (d) 1.5k, (e) 15k, and (f) 100k 22.2 mW and 200 μs laser pulses (see more exhaustive image set in supplemental information). g,
h: Radially averaged and normalized selected area electron diffraction patterns taken of the Ag0.5Ni0.5 film as a function of the number of laser pulses (noted in the legend);
indexed peaks are blue for Ni, and green for Ag. Inset in (g) is the lattice constant determined from the Ag <111 > and Ni <002 > peak positions of the selected area electron
diffractograms patterns.

the number of 785 nm laser pulses delivered to the Ag0.5Ni0.5 film
(22.2 mW power and 200 μs, see supplement for more exhaustive
series). SAED patterns were taken using a 20 μm condenser
aperture and the selected area is about 500 nm in diameter.
Figures 3g and 3h are the radially averaged and normalized
intensity, as a function of inverse wavenumber, for the SAED
patterns taken as a function of the number of laser pulses noted in
the legend. Figure 2c illustrates that at this power and pulse width,
a steady-state maximum temperature of ~890oC is reached in just
~25 μs. At the end of the pulse, the temperature decays rapidly to
less than 50oC in ~25 μs. Thus each pulse constitutes an
approximately 175 μs 890oC anneal. After the first pulse, the
images and SAED patterns have evidence of recrystallization as
distinct Ag-rich and Ni-rich regions emerge, and the broad
as-deposited rings form discrete spots and sharpen. Progressing to
500 pulses, the Ag <111 > and Ni <111 > lattice parameters
undergo a slight lattice expansion and contraction, respectively, as
evidenced by the slight SAED peak shifts, which is accompanied
by a gradual peak narrowing. The figure inset Figure 3i illustrates
the lattice parameter for the Ag-rich phase determined from the
<111 > peak position and the Ni-rich phase determined from the
<002 > peak position as a function of the number of laser pulses.
The peak shifts are consistent with out-diffusion of super-
saturated solute atoms in the Ag-rich and Ni-rich solid solutions,
which both have very limited solubility for the other component
and thus approach the pure Ag and Ni lattice parameter,
respectively. This is consistent with previous work on AgxNi1 − x
alloys that show significant Ni supersaturation of up to 13.5% for

")!"
$"
&&#%))) $"$"$

$%%
   
"!



&



%'&
&"
&
 $
"$
&$ %
"
'%
(
&
&&#%))) $"$"$&$ %
&&#%""$  


 652
Figure 4. HAADF scanning transmission electron microscopy (STEM) images of the same region of an Ag0.5Ni0.5 film (a) as deposited and (b–h) as a function of different number
of 22.2 mW and 200 μs laser pulses. b,c: Were initially offset by 5 μm from the laser center and (d–h) are images after the region was shifted to the laser center and subsequently
pulsed again. i: Is the corresponding electron energy loss spectra and the calculated Ag content.
Ag0.65Ni0.35 in the silver-rich as-deposited material (Proux et al.,
1999). Furthermore, they showed a progressive increase in the Ni-
rich phase, and a decrease in the Ni supersaturation in the silver-
rich phase with increased annealing temperature. Pulsed laser
deposition of AgxNi1 − x alloys also demonstrated a mixed phase
alloy consisting of silver- and nickel-rich phases as well as a
supersaturated phase with total solid solubility (van Ingen et al.,
1994). Again, modest annealing up to 400oC caused the meta-
stable phase to decompose into the equilibrium phases with
negligible solubility. Progressing from 1.5k pulses the normalized
silver peak intensities uniformly decrease until they essentially
vanish at 30k pulses. Simultaneously, the nickel peaks uniformly
increase over this pulse range, indicating a net silver loss con-
sistent with preferential silver evaporation due to silver’s almost
five order of magnitude higher vapor pressure relative to nickel.
Figures 4a to 4h illustrate higher magnification HAADF
images of a new series taken in a new field of view using the same
laser conditions (22.2 μW and 200 μs) at an electron beam posi-
tion initially aligned ~5 μm from the laser center. Figure 4i are
complementary EELS of the same region shown in Figures 4a to
4h with the resultant estimated silver fraction. Comparing
Figures 4a and 4b the re-crystallization and phase separation is
clear after the first pulse. The silver and nickel grains undergo
significant coarsening during the entire pulse progression. The
circles noted in Figure 3e are silver-rich rings, which the pro-
gression reveals lead to dewet holes, a network of rivulets,
")!"
$"
&&#%))) $"$"$

$%%
 
Yueying Wu et al.
bimetallic Ag–Ni nanoparticles and eventually individual nano-
particles consisting of primarily Ni. The net silver loss observed in
the images is confirmed via the EEL spectra in Figure 4i. For the
first 100k pulses the EEL spectra were taken in the offset position
consistent with the STEM images in Figures 4a to 4c. Subsequent
to the first 100k pulses the laser center was aligned with the
electron beam and pulsed for another 50k pulses. As noted, when
at the centered position, the silver peak dramatically decreases in
the 20k pulse spectra and is negligible in the 50k pulse spectra,
which is in good agreement with the SAED patterns where the
silver peak vanishes at 30k pulses.
As the laser irradiation profile on the sample is approximately
Gaussian and the confined thickness limits thermal diffusion to
the radial direction, the time-temperature profile varies with
radius. Figure 5a is a plot of the simulated temperature as a
function of radius at various times and Figure 5b is a two-
dimensional temperature map of the surface temperature at
200 μs for a 200 μs and 36.5 mW pulse. The temperature map
exhibits the elliptical irradiation profile on the sample surface
caused by the ~45° sample tilt with respect to the incident laser
beam (see Fig. 2a). Figure 5c is a low-magnification overview and
Figures 5d to 5h are HAADF STEM images taken at different
radial positions after 12k pulses. Clearly, the radial thermal gra-
dient results in various dewetting morphologies from single
particles to coalesced rivulets to continuous films with varying
degrees of grain growth. From Figure 5a the center temperature
 
"!




&



%'&
&"
&
 $
"$
&$ %
"
'%
(
&
&&#%))) $"$"$&$ %
&&#%""$  


 Microscopy and Microanalysis 653

reaches ~1,445oC, whereas at an 8 μm radius, for instance, the
temperature only reaches approximately only 940oC.
To demonstrate the effect of laser pulse width, Figure 5i is a
low-magnification overview and Figures 5j to 5n are HAADF
STEM images as a function of radius for an area exposed to 2.5 M
pulses at 20 μs pulse width and 36.5 mW power. Though the
cumulative laser exposure time for the 20 μs pulse width exposure
(50 s) is much longer than that for the 200 μs pulse width (2.4 s),
the temperature is continuously rising during the 20 μs pulse and
only reaches a maximum temperature of ~1,310oC. Comparing
Figures 5d to 5h with Figures 5j to 5n, it is clear that the inte-
grated thermal budget is much lower for the shorter pulse width;
the 20 μs and 2.5 M pulsed sample reveals very early stages of re-
crystallization and grain growth observed in Figures 5j and 5k,
and virtually no significant changes relative to the as-deposited
material in Figures 5l to 5n.
As shown in Figure 2b, various powers, pulse widths, pulse
numbers, and radii can generate different temperature pathways.
Figures 6a to 6d are image captures taken at different times and
similar dewetting profiles for cw experiments where laser power
was varied from 12.8 to 22.2 mW (see Supplementary Material for
full movies). The movies illustrate the tremendous stability of the
system even at the highest temperature where rapid nanoparticle
diffusion occurs. From Figure 2c, the steady-state temperatures
range from 520 to 890oC and Figure 6e illustrates that the time to
reach the nanoparticle morphology evolution has the expected
exponential temperature dependence consistent with the thermally
activated diffusive solid state dewetting (Thompson, 2012). The
results are in agreement with Niekiel et al. (2015) who exhaustively
studied the solid-state dewetting of thin gold films in situ using an
electrically heated TEM system. The estimated temperatures are
consistent with dewetting morphology and timescales as compared
to the comparably thick Au films (22 nm). For the gold films a
similar dewetting profile as shown in Figure 6d was achieved at
1,000 s at a homologous temperature of 0.58. The simulated tem-
perature of the Ag0.5Ni0.5 film for the 1,005 s image in Figure 6d has
a homologous temperature range from 0.46 to 0.64 and an average
of 0.54, in reasonable agreement with the Au film. Comparing the
cw 22.2 mW and 200 μs pulsed evolution in Figure 3 (and in the
Supplemental Material), the morphologies are fairly consistent, as
expected, since the cumulative time of 3 s is accomplished at ~17k
pulses (assuming the 175 μs per pulse annealing). In Figure 3g the
dewetting stage is initiated at ~15k pulses and individual nano-
particles are formed by 20k pulses (Supplemental Material). This
illustrates the power of the pulsed mode in that thermal time
stamps shorter than the (S)TEM acquisition time can be accessed
via the rapid ex situ imaging, which is enabled by the fact that no
discernible drift artifacts are introduced by the laser system. For

Figure 5. a: Simulated surface temperature versus position for various times illustrating the spatial and temporal temperature evolution for a 36.5 mW power and 200 μs pulse
width (b) two-dimensional plot of the surface temperature at 200 μs for the 36.5 mW power. c: Low-magnification HAADF scanning transmission electron microscopy (STEM)
image after 12k pulses of 36.5 mW power and 200 μs pulse width and (d–h) are the associated high-magnification HAADF STEM images as a function of radius [noted by the
dashed red arrow in (c)]. i: Low-magnification HAADF STEM image after 2.5 M pulses of 36.5 mW power and 20 μs pulse width and (j–n) are the associated high-magnification
HAADF STEM images as a function of radius [noted by the dashed red arrow in (i)].

")!"
$"
&&#%))) $"$"$

$%%
   
"!



&



%'&
&"
&
 $
"$
&$ %
"
'%
(
&
&&#%))) $"$"$&$ %
&&#%""$  


 654 Yueying Wu et al.

Figure 6. a–d: Image captures of continuous wave experiments exposed to various powers and collected at different times with comparable dewetting morphologies (see the
Supplemental Material for full movies). e: Plot of the image capture time for each image in (a–d) versus simulated temperature, which illustrates the expected exponential time
dependence of the solid state dewetting process.

Figure 7. High-magnification bright field transmission electron microscopy (TEM) images of an (a) as-deposited Ag0.5Ni0.5 and after (b) 1, (c) 100, (d) 10k, and (e) 35k laser
pulses of 16.6 mW power and 200 μs pulse width (see Supplemental Information for exhaustive image set). f: Are radially averaged and normalized fast Fourier transforms of
this image series.

example, a drift rate for the 12.8 mW cw experiment was estimated
to be 2.2 nm/min.
To illustrate the rapid ex situ mode, a high magnification pulse
study was performed at 16.6 mW and 200 μs pulse width. Figure 7a
reproduces the as-deposited image in Figure 3c and shows a series
of bright-field TEM images taken in the same field of view after
various numbers of pulses (see the Supplemental Material for
exhaustive series). Radially averaged fast Fourier transforms of the
high-magnification images are shown in Figure 7f. It is again clear
from comparing the as-deposited and 1 pulse image and fast
Fourier transform (FFT) that the film is recrystallized after the
single pulse, which according to Figure 2c represents a ~175 μs
630oC anneal. While the FFT images do not change significantly,
there is obvious grain growth that occurs. This series illustrates the
tremendous stability of the system as again no drift-correction was
applied during the experiment.

")!"
$"
&&#%))) $"$"$

$%%
   
"!



&



%'&
&"
&
 $
"$
&$ %
"
'%
(
&
&&#%))) $"$"$&$ %
&&#%""$  


 Microscopy and Microanalysis
Conclusion
We introduce a new optical delivery system that can be attached
to a (S)TEM tool for in situ and rapid ex situ imaging of
photothermally induced phenomena at the nanoscale. Various
thermal pathways are achieved by a combination of laser power
(0–215 mW), pulse width (1 ns to cw), number of pulses (user
defined) and radial position. To demonstrate the functionality of
the tool, we studied the recrystallization, grain growth, phase
separation, and solid-state dewetting of a Ag0.5Ni0.5 film under
various laser conditions. Using only a fraction of the laser power
(~20%), a maximum temperature of ~1,445oC is realized with
~25 μs rise times. The focused 3.7 μm radius spot and ~10 μm
heat affected zone allows for many individual experiments to be
performed on a single standard TEM membrane. Furthermore,
the radial temperature gradient in the heat affected zone yields
multiple simultaneous thermal profiles, thus providing a vast
array of thermal profiles in a single experiment.
Supplementary Materials. To view supplementary material for this article,
please visit https://doi.org/10.1017/S1431927618015465
Acknowledgments. P.D.R., T.M., and G.A.M. acknowledge that the laser
delivery system was supported by NSF via 1721719 (Ben Schrag program
manager). P.D.R., J.P.C., and Y.W. acknowledge the EELS plasmon studies of
the bimetallic nanoparticles is supported by NSF DMR-1709275 (Miriam
Deutsch program manager). D.A.G. and M.G.S. acknowledges support from
NSF CBET-1603780.
References
Allen FI, Kim E, Andresen NC, Grigoropoulos CP and Minor AM (2017) In
situ TEM Raman spectroscopy and laser-based materials modification.
Ultramicroscopy 178, 33–37.
Barwick B, Flannigan DJ and Zewail AH (2009) Photon-induced near-field
electron microscopy. Nature 462, 902–906.
Barwick B and Zewail AH (2015) Photonics and plasmonics in 4D ultrafast
electron microscopy. ACS Photonics 2, 1391–1402.
Berkowitz AE, Mitchell JR, Carey MJ, Young AP, Zhang S, Spada FE,
Parker FT, Hutten A and Thomas G (1992) Giant magnetoresistance in
heterogeneous Cu-Co alloys. Phys Rev Lett 68, 3745–3748.
Chen G and Hui P (1999) Thermal conductivities of evaporated gold films on
silicon and glass. Appl Phys Lett 74, 2942–2944.
Chen X, Li C and Cao H (2015) Recent developments of the in situ wet cell
technology for transmission electron microscopies. Nanoscale 7, 4811–4819.
Dimesso L and Hahn H (1998) Structure and magnetoresistance effect in
granular Ag–Ni alloys prepared by gas flow condensation technique.
Applied Physics 84, 953–957.
Garcia Prieto A, Arteche A, Aguilera-Granja F and Luisa Fdez-Gubieda M
(2015) Breakdown of magnetism in sub-nanometric Ni clusters
embedded in Ag. Nanotechnology 26, 455703.
Jiang Y, Zhang Z, Yuan W, Zhang X, Wang Y and Zhang Ze (2018) Recent
advances in gas-involved in situ studies via transmission electron
microscopy. Nano Res 11, 42–67.
Joseph F, Zeng C, Firestein K and Goldberg D (2017) Optical and
optoelectronic property analysis of nanomaterials inside transmission
electron microscope. Small 13, 1701564.
Kaplan M, Yoo B-K, Tang J, Karam TE, Liao B, Majumdar D, Baltimore D,
Jensen GJ and Zewail AH (2017) Photon-induced near-field electron
microscopy of eukaryotic cells. Angew Chem Int Ed Engl 56, 11498–11501.
Kim JS, LaGrange T, Reed BW, Taheri ML, Armstrong MR, King WE,
Browning ND and Campbell GH (2008) Imaging of transient structures
using nanosecond in situ TEM. Science 321, 1472–1475.
Kitada M (1985) Magnetic properties of immiscible Co-Ag and Ni-Ag thin
films prepared by co-sputtering. J Mater Sci 20, 269–273.
")!"
$"
&&#%))) $"$"$

$%%
 
655
Klein KL, Melechko AV, Rack PD, Fowlkes JD, Meyer HM and Simpson
ML (2005) Cu–Ni composition gradient for the catalytic synthesis of
vertically aligned carbon nanofibers. Carbon 43, 1857–1863.
Kubinski D and Holloway H (1995) Structure and giant magnetoresistance of
granular Ag‐Co and Ag‐Ni alloys grown epitaxially on MgO{100}. Applied
Physics 77, 782–789.
Langer G, Hartmann J and Reichling M (1997) Thermal conductivity of thin
metallic films measured by photothermal profile analysis. Rev Sci Instrum
68, 1510–1513.
Lee S-M and Cahill DG (1997) Heat transport in thin dielectric films. J Appl
Phys 81, 2590–2595.
Lewis BB, Winkler R, Sang X, Pudasaini PR, Stanford MG, Plank H, Unocic
RR, Fowlkes JD and Rack PD (2017) 3D Nanoprinting via laser-assisted
electron beam induced deposition: Growth kinetics, enhanced purity, and
electrical resistivity. Beilstein J Nanotechnol 8, 801–812.
Li J, Yin W-G, Wu L, Zhu P, Konstantinova T, Tao J, Yang J, Cheong S-W,
Carbone F, Misewich JA, Hill JP, Wang X, Cava RJ and Zhu Y (2016)
Dichotomy in ultrafast atomic dynamics as direct evidence of polaron
formation in manganites. npj Quantum Mater 1, 16026.
Liu H, Kwon O-H, Tang J and Zewail AH (2014) 4D imaging and diffraction
dynamics of single-particle phase transition in heterogeneous ensembles.
Nano Lett 14, 946–954.
Miller BK and Crozier PA (2013) System for in situ UV-visible illumination
of environmental transmission electron microscopy samples. Microsc
Microanal 19, 461–469.
Mohsin A, Cross NG, Liu L, Watanabe K, Taniguchi T, Duscher G and Gu
G (2018) Mapping the layer count of few-layer hexagonal boron nitride at
high lateral spatial resolutions. 2D Mater 5, 015007.
Nagamine LCCM, Mevel B, Dieny B, Rodmacq B, Regnard JR, Revenant-
Brizard C and Manzini I (1999) Magnetic properties and magnetoresis-
tance of as-deposited and annealed CoxAg1 − x and NixAg1 − x
(x = 0.2,0.37) heterogeneous alloys. J Magn Magn Mater 195, 437–451.
Niekiel F, Schweizer PM, Kraschewski S, Butz B and Spiecker E (2015) The
process of solid-state dewetting of Au thin films studied by in situ scanning
transmission electron microscopy. Acta Mater 90, 118–132.
Proux O, Mimault J, Regnard JR, Revenant-Brizard C, Mevel B and Dieny
B (2000a) Structural coherence between phases in Ni0.35Ag0.65 thin films.
J Phys Condens Matter 12, 3939–3953.
Proux O, Mimault J, Revenant-Brizard C, Regnard JR and Mevel B (1999)
Structural evolution of NiAg heterogeneous alloys upon annealing. J Phys
Condens Matter 11, 147.
Proux O, Regnard JR, Manzini I, Revenant C, Rodmacq B and Mimault J
(2000b) In situ X-ray absorption spectroscopy study of the thermal
behaviour of giant magnetoresistance CoxAg1 − x and NixAg1 − x
heterogeneous alloys. Eur Phys J Appl Phys 9, 115–124.
Roberts N, González C, Fowlkes JD and Rack PD (2013) Enhanced
by-product desorption via laser assisted electron beam induced deposition
of W(CO)6 with improved conductivity and resolution - IOPscience.
Nanotechnology 24, 415301.
Roberts NA, Fowlkes JD, Magel GA and Rack PD (2013) Enhanced
material purity and resolution via synchronized laser assisted electron beam
induced deposition of platinum. Nanoscale 5, 408–415.
Roberts NA, Magel GA, Hartfield CD, Moore TM, Fowlkes JD and Rack PD
(2012) In situ laser processing in a scanning electron microscope. J Vac Sci
Technol A 30, 041404.
Shorokhov D and Zewail AH (2016) Perspective: 4D ultrafast electron
microscopy—Evolutions and revolutions. J Chem Phys 144, 080901.
Stanford MG, Lewis BB, Iberi V, Fowlkes JD, Tan S, Livengood R and
Rack PD (2016a) In situ mitigation of subsurface and peripheral focused ion
beam damage via simultaneous pulsed laser heating. Small 12, 1779–1787.
Stanford MG, Lewis BB, Mahady K, Fowlkes JD and Rack PD (2017)
Review article: Advanced nanoscale patterning and material synthesis
with gas field helium and neon ion beams. J Vac Sci Technol B 35, 030802.
Stanford MG, Lewis BB, Noh JH, Fowlkes JD and Rack PD (2015) Inert gas
enhanced laser-assisted purification of platinum electron-beam-induced
deposits. ACS Appl Mater Interfaces 7, 19579–19588.
Stanford MG, Mahady K, Lewis BB, Fowlkes JD, Tan S, Livengood R, Magel
GA, Moore TM and Rack PD (2016b) Laser-assisted focused He + ion
 
"!




&



%'&
&"
&
 $
"$
&$ %
"
'%
(
&
&&#%))) $"$"$&$ %
&&#%""$  


 656
beam induced etching with and without XeF2 gas assist. ACS Appl Mater
Interfaces 8, 29155–29162.
Taheri ML, Stach EA, Arslan I, Crozier PA, Kabius BC, LaGrange T, Minor
AM, Takeda S, Tanase M, Wagner JB and Sharma R (2016) Current
status and future directions for in situ transmission electron microscopy.
Ultramicroscopy 170, 86–95.
Thompson CV (2012) Solid-state dewetting of thin films. Annu Rev Mater Res
42, 399–434.
van Ingen RP, Fastenau RHJ and Mittemeijer EJ (1994) Formation of
crystalline AgXNi1-X solid solutions of unusually high supersaturation by
laser ablation deposition. Phys Rev Lett 72, 3116–3119.
")!"
$"
&&#%))) $"$"$

$%%
 
Yueying Wu et al.
Wu Y, Fowlkes JD and Rack PD (2011) The optical properties of Cu-Ni
nanoparticles produced via pulsed laser dewetting of ultrathin films: The
effect of nanoparticle size and composition on the plasmon response.
J Mater Res 26, 277–287.
Xiang X-D, Sun X, Briceño G, Lou Y, Wang K-A, Chang H, Wallace-
Freedman WG, Chen S-W and Schultz PG (1995) A combinatorial
approach to materials discovery. Science 268, 1738–1740.
Zheng H, Meng YS and Zhu Y (2015) Frontiers of in situ electron
microscopy. Mater Res Soc 40, 12–18.
Zheng H and Zhu Y (2017) Perspectives on in situ electron microscopy.
Ultramicroscopy 180, 188–196.
 
"!




&



%'&
&"
&
 $
"$
&$ %
"
'%
(
&
&&#%))) $"$"$&$ %
&&#%""$