cases, such as hydrogen imaging on graphene
(24) or hydrogen-induced phase changes (25).
Such TEM methods fail to provide robust results
when facing more complex systems, such as hydrogen within an iron carbide (26), in which projection and diffraction effects may dominate.
However, one possible alternative technique is
atom probe tomography (APT). This method is
a three-dimensional (3D) microscopy technique
with near-atomic resolution (27) that can resolve
the chemical identities of individual atoms within
a target material, even for light elements (28).
Unfortunately, although in theory APT can resolve hydrogen without any instrumental modification, naturally occurring hydrogen is present
as a major element in most analyses. Hydrogen
in these cases is primarily a contaminant from the
high vacuum chamber of these systems (29, 30).
This means that interpretation of the 1H signal
is severely limited by the ambiguity of its origins,
and for the most part it is simply ignored in typical analyses.
To address this limitation, several researchers
have used deuterium (2H) to distinguish environmentally derived hydrogen from that deliberately
introduced for analysis. Deuterium’s natural scarcity (1.2 × 10−4 fraction of natural hydrogen)
guarantees that nearly all deuterium detected originates from the sample’s microstructure. However, other challenges remain that limit quantitative
spatial information on the hydrogen distribution
in engineering microstructures at the atomic scale.
To circumvent the remaining challenges for
high-quality measurements of the spatial distribution of hydrogen, we combined two approaches
to APT hydrogen-charging experiments: electrolytic charging (31) and cryogenic transfer (32). By
using the two approaches in concert, we provide a
quantitative measure of the 3D hydrogen content
within a proposed steel that contains VMoNbC.
In hydride-forming systems, APT can image
deuterium introduced to the material through
gaseous charging (33). However, the low solubility of deuterium within both ferrite and austenite (34) means that limited data for deuterium in
steel are available. A notable exception to this is
the work of Takahashi et al., in which deuterium
was qualitatively shown to correlate to the positions of vanadium and titanium carbides (32, 35)
through a combination of annealing, gaseous
charging, and rapid quenching. Recently, work
has been undertaken that uses a D2O charging
approach. Theoretically, a D2O charging approach
should achieve a higher deuterium content (31).
In recent work using this approach, no spatial data
were shown for steel samples, primarily because
of diffusion losses.
Examining a ferritic steel that contains VC
(Fig. 1), we used a high-fugacity electrolytic method
to charge hydrogen to high concentrations, in
conjunction with a cryogenic transfer, to suppress
loss of hydrogen via diffusion during transfer to
the atom probe. We used an existing cryo-transfer
system designed to also accommodate biological
or liquid specimen transfers (36) with similar ex-
perimental protocols. This approach provides a
complete and stringent cold chain, after hydrogen
charging, from laboratory benchtop to the atom
probe ultrahigh vacuum chamber (fig. S1). The
procedure requires a cold chain; however, a consid-
erably less advanced system than used here may
also prove effective in hydrogen detection.
To perform the deuterium charging, there are
several straightforward requirements. First, a spec-
imen in the needle geometry required for APT is
prepared from the steel sample by using standard
electropolishing techniques (28). The specimen
is then subjected to a short analysis in the atom
probe system. The result of this step is that after
the experiment, the field evaporation process that
underpins APT effectively provides an atomically
clean surface (37). The electrolytic charging pro-
cess is then conducted in a charging solution, con-
sisting of a deuterated electrolyte (NaOD, 0.1 M)
in heavy water (D2O). The in-vacuum cold chain is
incorporated into the process to suppress hydro-
gen diffusion, while entirely avoiding ice forma-
tion. The full details of the experimental component
of this approach are given in (38).
We electrolytically deuterated a sample of fer-
ritic steel and transferred at ~100 K for an atom
probe experiment (Fig. 2). The highlighted deu-
terium atoms (red) correspond to the mass-to-
charge peak observed at 2 Da. To verify that the
atoms shown are actually deuterium, and not H2,
an additional control sample was charged by using
light water (H2O) charged and analyzed in the
exact same manner; unlike the heavy-water charged
data set, no peak at 2 Da was observed (fig. S3).
The positions of the deuterium atoms show a
strong correlation to the proposed VC-trapping
sites in this material, which is consistent with
earlier work (32). Visual inspection of the data
set does not allow for identification of the exact
location of the hydrogen, either within each indi-
vidual carbide or at the carbide-matrix interface.
This is partly due to the limited number of atoms
involved. Indeed, an ongoing debate exists about
whether the hydrogen is at the surface (39) or
penetrates into the carbide itself (17). Previous
qualitative APT analysis suggests that this may
be a surface effect (32), but no quantitative data
confirm this result. In order to maximize the
hydrogen-trapping potential, the location of the
hydrogen relative to the trapping inclusions has
To further investigate the spatial location of
hydrogen, relative to that of the carbide, we analyzed
our data using an advanced statistical procedure
SCIENCE sciencemag.org 17 MARCH 2017 • VOL 355 ISSUE 6330 1197
Fig. 2. Introduced deuterium colocated with VCs. 3D view of deuterated ferritic steel, showing
individual carbides. As can be seen from the top-down and side slices, deuterium atoms (2H) are
correlated to the carbide positions.
RESEARCH | REPORTS