Rev. Sci. Instrum. 92, 025115 (2021); https://doi.org/10.1063/5.0037844 92, 025115
© 2021 Author(s).
Toolbox for atomic layer deposition process
development on high surface area powders
Cite as: Rev. Sci. Instrum. 92, 025115 (2021); https://doi.org/10.1063/5.0037844
Submitted: 16 November 2020 . Accepted: 29 January 2021 . Published Online: 18 February 2021
K. Knemeyer, R. Baumgarten, P. Ingale, R. Naumann d’Alnoncourt, M. Driess, and F. Rosowski
COLLECTIONS
This paper was selected as an Editor’s Pick
ARTICLES YOU MAY BE INTERESTED IN
Conformality in atomic layer deposition: Current status overview of analysis and modelling
Applied Physics Reviews 6, 021302 (2019); https://doi.org/10.1063/1.5060967
Surface chemistry of atomic layer deposition: A case study for the trimethylaluminum/
water process
Journal of Applied Physics 97, 121301 (2005); https://doi.org/10.1063/1.1940727
Understanding chemical and physical mechanisms in atomic layer deposition
The Journal of Chemical Physics 152, 040902 (2020); https://doi.org/10.1063/1.5133390
Review of
Scientific Instruments ARTICLE scitation.org/journal/rsi
Toolbox for atomic layer deposition process
development on high surface area powders
Cite as: Rev. Sci. Instrum. 92, 025115 (2021); doi: 10.1063/5.0037844
Submitted: 16 November 2020 •Accepted: 29 January 2021 •
Published Online: 18 February 2021
K. Knemeyer,1R. Baumgarten,1P. Ingale,1R. Naumann d’Alnoncourt,1,a) M. Driess,1,2
and F. Rosowski1,3
AFFILIATIONS
1BasCat—UniCat BASF JointLab, Technische Universität Berlin, Hardenbergstraße 36, 10623 Berlin, Germany
2Institut für Chemie, Technische Universität Berlin, Straße des 17. Juni 135, 10623 Berlin, Germany
3Process Research and Chemical Engineering, BASF SE, Carl-Bosch-Straße 38, 67056 Ludwigshafen, Germany
a)Author to whom correspondence should be addressed: [email protected]
ABSTRACT
Atomic layer deposition (ALD) is an industrially applied technique for thin film deposition. The vast majority of processes target flat substrates
rather than powders. For ALD on powders, new processes are needed, as different reaction conditions are required. Here, two setups are
described in detail, which enhance the ALD process development for powders. The first setup described is capable of directly measuring the
vapor pressure of a given precursor by a capacitance diaphragm gauge. Promising precursors can be pre-selected, and suitable precursor
saturation temperatures can be determined. The second setup consists of four parallel reactors with individual temperature zones to screen
the optimal ALD temperature window in a time efficient way. Identifying the precursor saturation temperature beforehand and subsequently
performing the first ALD half cycle in the parallel setup at four different reactor temperatures simultaneously will drastically reduce process
development times. Validation of both setups is shown for the well-known ALD precursors, trimethylaluminum to deposit aluminum oxide
and diethyl zinc to deposit zinc oxide, both on amorphous silica powder.
Published under license by AIP Publishing. https://doi.org/10.1063/5.0037844
., s
I. INTRODUCTION
Atomic layer deposition (ALD) is a thin film deposition tech-
nique capable of coating almost any given material with inorganic
layers.1–4 The gas solid reaction proceeds in a cyclic fashion in which
a precursor is added to the surface and reacts with exposed reac-
tive surface sites until saturation. Subsequent purging of residue
precursors and by-products is followed by dosing a reactant to the
chemisorbed precursor to form the desired layer and recreating sim-
ilar reactive surface sites. This procedure can be repeated indefi-
nitely to grow the deposited material layer by layer to the desired
thickness. The self-limitation in each so-called half-cycle leads to
the precise control of the deposited thickness and excellent repro-
ducibility. ALD has its origin in the microelectronic industry in
which most often Si-wafers are coated with elements to tune elec-
tric properties.5Over the course of the past 50 years, a wide variety
of elements and combinations, thereof, were deposited via ALD.1
Recently, ALD attracted interest in the field of surface modifica-
tion of powders, e.g., for drugs,6–8 batteries,9–11 or catalysts.12–16
Several reactor concepts for powders were already developed,17 such
as fixed bed,18 rotary,19 or pulsed bed reactors.20 Unfortunately, the
already developed processes for flat substrates cannot be straight-
forwardly transferred to powders as additional challenges have to
be overcome. Generally, fluidized beds and fixed beds work at a
relative high pressure compared to ultra-high vacuum (UHV) pro-
cesses on flat substrates leading to longer diffusion times. Addi-
tionally, the use of plasma or ozone is limited as both tend to
recombine readily at contact with a surface,21,22 making it impos-
sible to be effective on the bottom fraction of a fixed bed or in
highly porous substrates. Furthermore, powders have a larger spe-
cific surface area in the order of up to several magnitudes, demand-
ing excellent precursors. The precursor must be reactive toward
surface groups while being inert to evolving by-products and must
exhibit thermal stability up to the desired process conditions. Most
importantly, the precursor must be highly volatile to achieve satu-
ration in reasonable time frames.23 Lowly volatile precursors lead
to long dosing times and are, therefore, unfeasible for industrial
applications.
Rev. Sci. Instrum. 92, 025115 (2021); doi: 10.1063/5.0037844 92, 025115-1
Published under license by AIP Publishing
Review of
Scientific Instruments ARTICLE scitation.org/journal/rsi
Vapor pressure can be determined by thermogravimetric analy-
sis and subsequent calculations based on the Langmuir equation24,25
or by directly measuring it with a capacitance manometer.26–28 If the
necessary vapor pressure cannot be reached below the decomposi-
tion temperature, the precursor has to be replaced by a more volatile
one. Therefore, pre-selection of suitable precursors for ALD should
always be the first step in process development.25,29,30 Once suit-
able vapor pressures and evaporation temperatures are obtained, the
reactivity of a precursor should be investigated by quick testing facil-
ities to further reduce process development times. For ALD on flat
substrates, a quartz crystal microbalance31 or ellipsometer32 leads to
quick investigations of the growth behavior, but for powders, both
techniques are obsolete. Therefore, finding the right combination of
precursor, substrate, temperature, pressure, and dosing times is time
consuming without proper equipment.
This paper shows two self-designed setups, which tackle the
essential challenges of process development on powders.
Vapor pressure determination:
•Direct determination of vapor pressure up to 200○C for any
given precursor
•Dismiss lowly volatile precursors (vapor pressure <10 mbar)
•Identify the ideal precursor saturation temperature for maxi-
mum molar flow
Parallel deposition:
•Identify precursor reactivity at four different temperatures in
parallel
•Determine the ALD window
In combination, both setups lead to time efficient ALD pro-
cess development on powders in fixed bed reactors. The setups are
evaluated by two of the most prominent ALD precursors, trimethy-
laluminum (TMA) and diethyl zinc (DEZ), of which the vapor pres-
sure curves are measured and reactivity tests on amorphous SiO2are
conducted.
II. SETUP DESIGN
Both setups are operated separately and are described in detail
in Secs. II A and II B with their respective flow charts (Figs. 1 and
2). However, both setups are located within a closed cabinet of poly-
carbonate windows embedded in aluminum profiles. They share a
common ventilation to evacuate possible evolving gases.
A. Vapor pressure determination
1. Precursor
The precursor chamber consists of a quartz reservoir (5 cm
height ×2 cm inner diameter) with a KF16 flange connected to a
¼” bellow valve (Swagelok®). The precursor is filled inside a glove-
box or in air, respectively, depending on its air stability. Prior to
the vapor pressure determination the residual gas atmosphere has
to be completely removed so that the resulting pressure is the result
of the precursor’s vapor pressure. This is realized by three consec-
utive cycles of freezing the precursor in liquid nitrogen, evacuating
the precursor chamber, and then thawing the precursor. This can
be done conveniently by the Schlenk technique. Subsequently, the
precursor chamber is connected to the setup by VCR connections
(Swagelok®).
2. Heating
Heating of the precursor is realized by an oven (Salvislab Ther-
mocenter TC100) in which the precursor and all affected dosing
lines (stainless steel) are located. Having one heating zone prevents
cold spot formation and with that undesired condensation of the
precursor. The oven can be operated up to 200 ○C, and step wise
temperature programs can be carried out.
3. Measurement cell
The measurement cell consists of a capacitance manometer
(MKS Baratron®Type 631) with a control device (MKS Instru-
ments, PDR2000 dual capacitance diaphragm gauge controller),
which displays the pressure measured by the capacitance diaphragm
gauge. The measuring principle is based on the distance between a
diaphragm and a reference electrode, together serving as a capacitor.
The diaphragm is exposed on one side to vacuum and on the other
side to the sample pressure. It deflects depending on the present
pressure changing the distance to the electrode and with that the
capacitance. The capacitance is, therefore, correlated with the pres-
sure in the system. The control device precisely measures the 0 V
–10 V signal from the capacitance manometer to cover the full range
from 0.1 Torr (0.13 mbar) to 1000 Torr (1300 mbar). The data are
logged by reading the analog output with a data logger (Picolog
TC-08 Thermocouple logger) and are then processed with a per-
sonal computer. To prevent any condensation on the diaphragm, the
capacitance manometer is internally heated up to 200○C.
4. Downstream
Evacuation of the system is realized by two pumps. One is a
rotary vane pump (Pfeiffer Vacuum, DUO 5) for a vacuum down
to 3 ×10−3mbar, which is used for evacuating the system. The sec-
ond pump is a turbo molecular pump (Pfeiffer Vacuum, HighPace®
80) and is used whenever the baratron needs to be zeroed before
the measurement, as the turbo molecular pump reaches pressures
below the full range of the baratron. To prevent the exposure of
precursor to the pumps, a liquid nitrogen cooling trap is installed
between the pumps and the rest of the system. The system can be
flushed with nitrogen (99.999%) to purge residues away. To prevent
damage to the baratron by pressures exceeding 1 bar, an overflow
valve is installed in the nitrogen line, which releases pressures above
1 bar.86,87
B. Parallel deposition setup
1. Reactors
The reactors consist of borosilicate glass (Duran®) tubes with
an inner diameter of 3 mm, a wall thickness of 2.5 mm, and a length
of 7 cm. At the lower third, a narrowing serves as a ledge for quartz
wool on which the powder can be filled (0.1 ml). The bottom and
top parts of the tube each contain a KF16 flange connected on the
top side to a bellow valve and on the bottom end to the precursor
reservoir and subsequently to an additional bellow valve. The valves
allow the assembly of substrate and precursor under inert conditions
Rev. Sci. Instrum. 92, 025115 (2021); doi: 10.1063/5.0037844 92, 025115-2
Published under license by AIP Publishing
Review of
Scientific Instruments ARTICLE scitation.org/journal/rsi
inside a glovebox. The precursor reservoir is made of borosilicate
glass to visualize the precursor level. The precursor dosing is realized
by overflowing the precursor with nitrogen.
2. Heating
Heating is realized in two heating zones. Heating zone 1 is reg-
ulated by an oven (Salvislab Thermocenter TC100), which is respon-
sible for heating all precursors simultaneously. Furthermore, it heats
all stainless steel tubes and walls, which are in contact with the pre-
cursor to prevent cold spots and, therefore, condensation. Heating
zone 1 has generally the same or lower temperature than heating
zone 2. Heating zone 2 consists of individual heating jackets (Win-
kler, custom made) for each reactor. It can be adjusted by a tem-
perature controller (Winkler, W-200 series) from room temperature
to 400○C to allow different temperatures to be screened in paral-
lel. The heating jacket itself resists up to 200○C from outer heating,
which allows the heating zone 1 to be operated up to 200○C. Note
that the set point of the heating jacket is considered as substrate tem-
perature, and blank experiments showed a negligible temperature
offset.
3. Gas supply
The gas supply consists of synthetic air (99.999%) and nitrogen
(99.999%) of which either is dosed to the reactor selected by a three-
way valve. The selected gas is distributed into four lines, leading to
a manual needle valve (Swagelok, 100 ml/min) for each reactor. An
overflow valve prevents damage to the glass equipment and needle
valves, as pressures over 1 bar are released.
III. EXPERIMENTAL
TMA and DEZ are both intensively studied ALD precursors
and will, therefore, be used as reference systems to evaluate the
setups. Their vapor pressures were measured and compared to the
literature for validation of the vapor pressure setup. Additionally,
vapor pressures of metal acetylacetonates were measured and pro-
vided. Deposition experiments of TMA and DEZ on amorphous
SiO2were conducted at different temperatures to validate the par-
allel deposition setup.
A. Chemicals
Silica powder [SiO2, high-purity grade ≥99% (Davisil
Grade 636), average pore size 60 Å, 35–60 mesh particle size,
Sigma-Aldrich, specific surface area 505 m2g−1] was used
as a substrate. Trimethylaluminum [Al(CH3)3, TMA, elec. gr.,
99.999% Al], diethyl zinc [Zn(C2H5)2, DEZ, elec. gr., 99.999%
Zn, Strem Chemicals GmbH], manganese(III) acetylacetonate
(Sigma-Aldrich, technical grade), chromium(III) acetylacetonate
(Sigma-Aldrich, 99.99%), Iron(III) acetylacetonate (abcr, 95%),
nickel(II) acetylacetonate (Sigma-Aldrich, 95%), copper(II) acety-
lacetonate (Sigma-Aldrich, 97%), cobalt(II) acetylacetonate (Sigma-
Aldrich, 97%), vanadium(III) acetylacetonate (Sigma-Aldrich, 97%),
cobalt(III) acetylacetonate (Sigma-Aldrich, 99.99%), and water
(H2O, CHROMASOLV®, for HPLC, Riedel-de Haën) served as pre-
cursors and were used without further purification. High purity N2
and synthetic air (99.999%) were used as carrier and purging gases.
B. Experimental—Vapor pressure determination
The metal organic precursor (1 ml) was inserted in the precur-
sor chamber under inert conditions inside a glovebox. The precursor
chamber was then transferred from the glovebox to a Schlenk line by
connecting it via the bellow valve. The precursor was frozen in liq-
uid nitrogen, and the excessive atmosphere was removed by a rotary
pump and subsequently thawed at room temperature. This cycle was
repeated three times to remove the residual glovebox atmosphere.
The precursor chamber was then assembled to the setup inside the
oven. Then, the setup was evacuated by the rotary vane pump fol-
lowed by flushing of nitrogen. This was repeated up to three times
to make sure no residue air or adsorbates remain inside the lines.
The system was then evacuated to pressures below the full range
of the capacitance manometer by using the turbo molecular pump
followed by zeroing the manometer. The precursor and setup walls
were heated to the desired starting temperature by the oven. After
temperature stabilization, the bellow valve of the precursor chamber
was opened and the precursor was released into the system where
the pressure was constantly measured and recorded. This pressure
equals directly to the vapor pressure of the precursor. The tem-
perature was then increased stepwise until the desired maximum
temperature was reached. For each temperature step, sufficient long
stabilization time should be realized to reach thermal equilibrium.
After reaching the maximum temperature and measuring the result-
ing vapor pressure, the bellow valve of the precursor chamber was
closed and the system was then evacuated. The last step was flush-
ing with N2to remove the precursor. The cooling trap was then
disassembled, and the frozen precursor was quenched.
C. Experimental—Parallel deposition
The reactors were assembled in the glovebox under inert con-
ditions, allowing the substrate and the precursor to be free of
unwanted adsorbates such as water. The glass tube was filled with
a thin layer of quartz wool on which the substrate (SiO2, 0.1 ml) was
deposited. The top part of the reactor was then connected to a closed
bellow valve. The precursor vessel was filled with the respective pre-
cursor and connected to the bottom end of the reactor, which con-
tains a closed bellow valve. This was done individually for up to four
reactors, which were then transferred from the glovebox to the setup.
Reactors were assembled to the setup by connecting the bottom
valve to the gas supply and the top valve to the exhaust line. Heat-
ing jackets were attached and set to the desired temperatures. For
DEZ, the set temperatures were 50 ○C, 80 ○C, 100 ○C, and 120 ○C,
and for TMA, the set temperatures were 30○C, 80○C, 150○C, and
200○C, respectively. Nitrogen served as carrier gas for the precursor,
and the four needle valves were set to 20 ml/min. The bellow valve
on the top of the reactor was then opened followed by the bellow
valve on the bottom. The precursor was flowing through the sub-
strate from the bottom to the top without fluidizing it. After a certain
reaction time, the precursor was fully evaporated and transported
through the substrate bed. Once this was observed and sufficient
long N2dosing time passed, synthetic air was dosed by switching the
three-port valve. This allows air-sensitive chemisorbed precursors
to react in a controlled environment before exposing it to air. After-
ward, all heaters were turned off, the bellow valves were closed, and
the reactors were disassembled. The substrates were then chemically
analyzed either by x-ray fluorescence spectroscopy (XRF, Bruker S4
Rev. Sci. Instrum. 92, 025115 (2021); doi: 10.1063/5.0037844 92, 025115-3
Published under license by AIP Publishing
Loading more pages...