Highly customisable scanning droplet cell microscopes using

Transcription

Highly customisable scanning droplet cell microscopes using
Journal of Electroanalytical Chemistry 740 (2015) 53–60
Contents lists available at ScienceDirect
Journal of Electroanalytical Chemistry
journal homepage: www.elsevier.com/locate/jelechem
Highly customisable scanning droplet cell microscopes
using 3D-printing
Jan Philipp Kollender a,b, Michael Voith b, Simon Schneiderbauer c, Andrei Ionut Mardare a,b,
Achim Walter Hassel a,b,⇑
a
Christian Doppler Laboratory for Combinatorial Oxide Chemistry Institute for Chemical Technology of Inorganic Materials, Johannes Kepler University Linz, Altenberger Straße
69, 4040 Linz, Austria
b
Institute for Chemical Technology of Inorganic Materials (ICTAS), Johannes Kepler University Linz, Altenberger Straße 69, 4040 Linz, Austria
c
Department of Particulate Flow Modelling, Johannes Kepler University Linz, Altenberger Straße 69, 4040 Linz, Austria
a r t i c l e
i n f o
Article history:
Received 5 November 2013
Received in revised form 15 December 2014
Accepted 29 December 2014
Available online 7 January 2015
Keywords:
Scanning droplet cell microscopy
Rapid prototyping
3D printing
Flow cell
a b s t r a c t
3D printing was applied for the first time to produce highly customised flow type scanning droplet cell
microscope heads which combine electrochemical measurements with downstream analytics of the
electrolyte. The main advantages are the optimised fluid dynamics, the homogeneous and laminar mass
transport along with the simplicity of the production at low costs. An improved design is presented that is
hard to be machined in a classical way. This flow-type scanning droplet cell microscope (FT-SDCM) combines features from older versions of the techniques, the classical theta capillary based version and Vshaped microscopes. Different versions are compared and fluid dynamic simulations were performed
to reveal their particularities in terms of electrolyte flow and surface wetting. Fabricating of the complex
design of the flow cell was realised using a rapid prototyping approach. The newly proposed prototype is
tested under various experimental conditions for assessing its stability, wetting and sealing performances. Both chemical and electrochemical dissolution experiments have shown a perfect electrolyte
confinement within the cell and a complete wetting of the addressed area together with high throughput
experimentation capabilities due to the robust design and ease of use in combination with a gantry robot.
Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction
The working principle of the Scanning Droplet Cell Microscopy
(SDCM) was firstly developed in 1997 by Hassel et al. for electrochemical investigations of nanostructured oxide films on aluminium [1]. The central idea behind SDCM is to address only a small
area of the investigated surface (working electrode) which results
in a strong localisation of the electrochemical reactions. This strong
localisation has been achieved by bringing an electrolyte droplet at
the tip of a glass capillary in contact with the working electrode
(free droplet mode) or by pressing a soft plastic capillary or silicone-sealing terminated glass capillary against the working electrode (contact mode). Spot sizes down to a few micrometers can
be achieved with excellent reproducibility in contact mode [2].
With SDCM all common electrochemical measurements such as
⇑ Corresponding author at: Christian Doppler Laboratory for Combinatorial Oxide
Chemistry, Institute for Chemical Technology of Inorganic Materials, Johannes
Kepler University Linz, Altenberger Straße 69, 4040 Linz, Austria. Tel.: +43 732 2468
8700; fax: +43 732 2468 8905.
E-mail address: achimwalter.hassel@jku.at (A.W. Hassel).
http://dx.doi.org/10.1016/j.jelechem.2014.12.043
1572-6657/Ó 2015 Elsevier B.V. All rights reserved.
cyclic voltammetry, chronoamperometry, potentiometry and electrochemical impedance spectroscopy can be performed [3]. Within
the last 15 years various types of SDCMs such as flow-type SDCM
(FT-SDCM) [4] and V-shaped SDCM [5] for electrochemical flowthrough experiments or photoelectrochemical SDCM (PE-SDCM)
[6] for local photoelectrochemistry have been developed. Until
now SDCM has been applied for example for high-throughput
screening of thin film material libraries [7,8], microstructuring of
surfaces [9], local anodisations for plastic electronics [10], direct
writing of anodic oxides [11], localised measurements on bulk
samples [12–15], impedance measurements on the microscale
[16] and online monitoring of reaction products by connecting
the output of a FT-SDCM to external analytics such as ICP-MS
[17,18] or ICP–OES [19,20]. Essentially a FT-SDCM is a mobile electrochemical flow-through cell that can be easily moved to another
spot on the sample. Additionally it addresses only a small part of
the working electrode. This allows for example high throughput
screening of thin-film material libraries which is nearly impossible
using conventional flow cells. One drawback of any capillary based
approach is that additional expensive equipments e.g. capillary
puller, high precision polishing machine and an optical microscope
54
J.P. Kollender et al. / Journal of Electroanalytical Chemistry 740 (2015) 53–60
for manufacturing and assembly are needed. Additionally, the used
capillaries are rather fragile and great care is needed when handling them. Therefore, a robust and easy way to manufacture an
alternative to the capillary based-SDCM is of high interest and
would make SDCM more easily accessible for other researchers.
The most obvious approach would be to fully machine a SDCM
from bulk material using high-precision 5-axis machining centres
as done in some commercially available [21,22] and previously
reported solutions [23]. Unfortunately, using this approach, the
cost even of a simple SDCM can substantially rise.
During the last 10 years significant advances have been made in
3D-printing in terms of available materials, printing resolution and
price. Today, materials with good chemical stability for most electrochemical applications e.g. acrylonitrile butadiene styrene (ABS),
epoxy, polycarbonate (PC), polypropylene (PP), titanium and stainless steel are available [24]. Printing resolutions down to 100 lm
are available on commercial 3D-printers. A clear advantage of
3D-printing is that even very complex geometries including undercuts and embedded curved channels can be easily realised since
the object is built layer by layer without needing mechanical tools.
The price of the printed object is essentially only determined by
the amount of consumed building material independent of how
complex the object is and the time need to print the object if something like a depreciation of the printer is calculated. Today, even no
ownership of a 3D-printer is required since 3D-printing service
companies are available offering various materials and printing
techniques combined with instant online quoting [25,26]. This
eliminates the high investment costs for a 3D-printer and need
of specifically trained staff. Prices for 3D-printing have dropped
dramatically for selected materials which makes 3D-printing a relatively cheap method compared to CNC-milling. CAD-drawings
required for 3D-printing can be easily made using low-cost and
user-friendly CAD-programs [27,28]. This combination of userfriendly CAD-programs and availableness of 3D printing service
companies makes it very easy to obtain highly customized SDCMs
at low cost. An expensive in-house mechanical workshop is therefore no longer needed.
An attempt for 3D-printing of electrochemical flow cells was
already made in 2010 [29]. However this 3D printed cell has not
been coupled with online downstream analytics of reaction products and has no scanning capabilities. Therefore in this paper a
3D printed FT-SDCM with a complex geometry specifically
optimised for online downstream analytics of reaction products
is presented. Additionally the presented FT-SDCM can also be used
for automated high-throughput experimentation on thin film
material libraries. The cell was characterised using computational
fluid dynamics (CFD) simulations and tested under different experimental conditions.
2. Experimental and theoretical approaches
2.1. Theoretical simulations and previous FT-SDCM designs
Two previous designs of the FT-SDCM are presented in this
work and their particularities in terms of electrolyte flow are calculated and compared for a better understanding of possible design
drawbacks which can be improved. All simulations shown in this
publication were carried out in 3D space using the ANSYS FLUENT
software (Ansys Inc., U.S.A). The flow is assumed to be steady,
incompressible and isothermal. Thus, the steady incompressible
Navier–Stokes equations are solved [29]. Furthermore, the
influence of possible dissolved species is neglected and the calculation uses, therefore, the properties of water at room temperature.
For the discretisation of the convective term in the momentum
equation, a second-order upwind scheme is used. The derivatives
appearing in the diffusion term are computed by a least squares
method and the pressure–velocity coupling is achieved by the SIMPLEC algorithm. The face pressures are computed as the average of
the pressure values in the adjacent cells (linear interpolation). A
flow rate of 3 mL min 1 was used in the calculations which can
be considered laminar since the channel as well as the gap Reynold
numbers are between 10 and 100 depending on the droplet cell
geometry. At the flow guide separating the in- and outflow channels a no-slip boundary condition was applied while at the inflow
boundary condition the mass flow rate is set. At the outflow
boundary the pressure has to be prescribed, which is set to the surrounding pressure. Finally, a polyhedral mesh was used with a grid
spacing of approximately 5 % of the local channel diameter, which
leads to a much finer grid in the gap than in the in- and outflow
regions of the FT-SDCM.
Until now basically two different types of FT-SDCMs are used.
The first one is the theta capillary based approach (see Fig. 1a)
[4]. A glass capillary is split into two channels by a vertical wall
which is partially removed at the cell tip for allowing the electrolyte flow. In this case the electrolyte is streaming from one compartment of the cell into the other. Typically a silicone gasket at
the tip of the capillary is used to seal the wetted area. The microreference and counter electrodes are inserted from the top part
in one of the electrolyte channels, while the investigated area
becomes the working electrode. This design ensures wetting of
the sample even at low flow rates. The CFD simulation presented
in part (a) of Fig. 1 was calculated using a flow rate of 3 mL min 1
exemplifying this aspect. Starting from the point of highest velocity positioned at the centre of the wetted area, the calculations
show a radial decrease of the fluid velocity with more than 50%
along a distance equivalent to a channel diameter. The fluid velocity mapping shows a U-shaped profile for velocities higher than
75% (considered as a minimum useful limit) from the maximum
achievable velocity. This rather sharp vectorial profile indicates
that removing gas bubbles formed on or particles released from
the surface may be accentuated in the centre of the wetted region
and not as effective toward the edges. Apart from this, a major
drawback of the theta capillary approach is the fragility of the cell
and the long manufacturing time required. Additionally, the process of manually removing the separating wall may not be very
reproducible so the geometry of the flow channel near the working
electrode can greatly differ between various cells. This can be overcome if all geometrical values are exactly known and a calibration
series of dip etching in hydrofluoric acid is performed from which
the tips are reproducibly prepared. Nevertheless, it remains
laborious.
The second type of flow cell is typically manufactured from a
polymer block by drilling two channels into the block [23]. The
two channels are intersecting each other under a certain angle at
the tip of the cell creating an elliptical opening. The sealing at
the tip of this V-cell is typically applied manually. The CFD simulations of this type of cell are presented in part (b) of Fig. 1. In order
to allow a direct comparison with the theta-capillary cell, the same
flow rate and the same inner diameter of the flow channels were
used. Also, the distance between the investigated surface and the
lower point of the wall, separating both channels is identical with
the previous case. The maximum fluid velocity calculated within
the V-cell is approximately 3 times larger than the one found
for the theta-capillary, but the velocity profile is very different.
The distribution of the velocities higher than 75% from the maximum achievable velocity is in this case more located at the tip of
the cell. A certain elongation of this profile along the outlet channel
(right side) is accompanied by the presence of a vertex where a
dramatic velocity drop can be seen. From this simulation it can
be concluded that the probability that this type of cell produces
vortexes is higher as compared to the theta-capillary type. Each
J.P. Kollender et al. / Journal of Electroanalytical Chemistry 740 (2015) 53–60
55
Fig. 1. (a) CFD-simulation for capillary based FT-SDCM. (b) CFD-simulations for polymer block-based FT-SDCM. Arrows indicate flow direction and the velocity given in m s 1
is colour-coded. The flow rate in both cases is 3 mL min 1 resulting in different flow rates and colour codes for the two different geometries. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)
vortex can pin a gas bubble due to the zero velocity at its centre
and is obstructing the fluid flow. In principle this is a major drawback of this design, despite being more robust. If the gas bubble
becomes larger with time by collecting gas from the solution it
may exceed a critical size so that the bubble is torn apart. If such
vortex will form even closer to the addressed surface, then gas
bubbles sticking or being produced at the working electrode are
not reliably removed by the flowing electrolyte. This leads to an
unwanted situation when the electrolyte is just flowing through
the cell without coming in contact with the surface to be investigated. As a result, with this design relatively high flow rates are
required to ensure constant wetting of the surface. Such high flow
rates on the other hand can cause unwanted changes in the reaction conditions up to types of tribocorrosion such as flow induced
corrosion [30].
2.2. Simulation of improved flow type SDCM and its fabrication by 3Dprinting
To overcome the limitations of the two previous designs, an
improved version of flow-type SDCM was designed and fabricated
using 3D-printing. Similar to previous designs, it consists of two
flow channels and an opening at the tip where the electrolyte
comes in contact with the sample. A schematic drawing of the flow
channels is presented in Fig. 2a while an outer view of the
cross-section of the entire cell block is presented in the part b of
the same figure. The channels intersect at an angle of 65° at the
tip, creating a circular opening. The particularity of the newly proposed design addresses the need of improving the electrolyte flow
on the surface of the investigated WE. At the intersection of the
two channels an additional vertical wall was added to better guide
the flowing electrolyte towards the surface. This ensures permanent contact between electrolyte and sample surface and efficient
removal of gas bubbles or particles from the surface even at low
flow rates. This additional wall can be easily manufactured by
3D-printing. This geometry is very difficult to be realized when
manufacturing the flow cell from a plastic block using conventional
mechanical methods e.g. by drilling the two flow channels.
Before fabricating the new design prototype, a theoretical
approach was used for closely studying the electrolyte flow details.
In order to understand the hydrodynamical properties of the new
flow cell, an extended series of CFD simulations was performed.
The sealing thickness upon compression (450 lm) is defining the
distance between the cell and the sample surface. The boundary
conditions for the simulation were set at room temperature,
aqueous solutions and a constant flow rate of 3 mL min 1 which
is similar to the flow rate later used during the experimental testing. The graphical representation of the results is shown in Fig. 3.
The velocity vectors are indicated by arrows, while their modulus
values are plotted as a coloured mapped profile. It can be seen that
Fig. 2. (a) Geometry of flow channels at the tip of the cell. (b) Cross-section of CAD model of the complete flow channels including threads for connectors and sealing (shown
in black).
56
J.P. Kollender et al. / Journal of Electroanalytical Chemistry 740 (2015) 53–60
Fig. 3. CFD-simulation of the improved FT-SDCM manufactured by 3D-printing. The
velocity scale is given in m s 1. The flow rate is 3 mL min 1.
the velocity of the electrolyte dramatically increases when it
passes the flow guidance. This increase in the velocity is due to
the continuity equation for incompressible fluids where the
velocity of a flowing media increases when passing a constriction.
The profile of the highest velocities (>75% of the full scale) at the
tip has a U-shape in this case which results from the combined
influences of the flow channels and vertical wall. The corresponding profiles observed in previous designs (Fig. 1) are much more
localised. A planar velocity profile for the theta capillary and a spot
profile for the classical V-cell combined with the lateral velocity
gradients discussed before result in lower bubble removal
efficiency as compared to the new design.
Locally increasing the velocity at the tip leads to reliable
removal of bubbles and particles from the sample surface. Also,
no back mixing could be observed. In order to compare the efficiency of gas bubble removal between the V-cell and the newly
proposed design, the trajectories of bubbles formed at the sample
surface was calculated. A bubble diameter of 1 lm was used in the
simulations and the results are summarized in Fig. 4. The classical
V-cell presented in the part (a) of Fig. 4 shows again the presence
of a vortex which affects the overall trajectories. This result is consistent with the flow simulation previously presented in Fig. 1(b).
The increased probability of vortex formation leads to a decrease
in the gas bubble removal efficiency. Moreover, the accumulations
of bubbles in a given area (pinned by a vortex) will result in a much
larger bubble being formed due to a coalescence of smaller bubbles. In extreme cases, this larger gas accumulation may eventually
lead to a complete blocking of the flowing electrolyte which will
render the V-cell unusable for the intended purpose. The bubble
trajectories of the new flow cell design are presented in Fig. 4(b).
In this case the bubbles are efficiently removed from the surface
since no vortex formation is observed. This dramatic change in
both flow profile and bubble trajectories is due to the smart tuning
of the geometry proposed for the new cell tip. The residence time
of the bubbles is also presented in Fig. 4 as a colour coded mapping
in seconds. Due to the looped bubble trajectories circulating the
vortex in the case of the classical V-cell, their velocity profile and
the subsequent residence time is expected to be strongly affected.
This is not directly observable in Fig. 4 due to the difficulty of providing accurate data regarding specific bubble parameters (e.g.
diameter, growth, coalescence, number of bubbles, etc.). These
parameters are specific to a given electrolyte and cannot have a
general description, therefore are omitted in the present study.
However, the calculated trajectories can be safely used for concluding upon the overall behaviour of the gas bubble dynamics.
The customised flow cell was manufactured using stereolithography and an ultraviolet curable photopolymer as building
material. By means of true to scale downsizing of the presented
design openings and flow channel diameters down to 800 lm
could be obtained on the used stereolithography machine. Further
downsizing can be achieved by using specialised 3D-printing processes offering resolutions down to 150 nm [31]. Flow channels
and opening of the manufactured cell had a diameter of 2 mm
(see Fig. 2). The vertical wall/flow guide had a thickness of
500 lm. To prevent leakage and ensure high reproducibility of
addressed area a sealing ring made from nitrile–butadiene rubber
(NBR) was installed at the tip of the cell. The inner hole of the sealing ring had a diameter of 2 mm to exactly match the size of the
opening of the cell. NBR was chosen as sealing material due to its
high chemical stability. Softer sealing materials like silicone or
natural rubber derivatives have been successfully used with the
presented flow cell. These NBR–sealing rings are a convenient
Fig. 4. Simulated gas bubble trajectories for the classical V-shaped cell design (a) and the new cell design (b). Gas bubbles 1 lm in diameter are released from the surface and
their residence time is colour coded in seconds. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
J.P. Kollender et al. / Journal of Electroanalytical Chemistry 740 (2015) 53–60
57
Fig. 5. (a) CAD model of the FT-SDCM setup. (b) Photography of the FT-SDCM.
replacement of the manually applied silicone sealings previously
used [6,23]. In order to connect the flow-cell to the electrolyte
reservoir and to downstream analytics, self-made gold-plated
screws equipped with a barbed fitting were used. One of these
gold-plated screws was used as counter electrode. As reference
electrode a l-Hg/Hg2(CH3COO)2/NaCH3COO system was used
[32]. A detailed CAD drawing of the complete SDCM in contact with
a sample is shown in part (a) of Fig. 5. The micro reference-electrode was fixed on the front side of cell using a specially designed
threaded port. Constant contact between reference electrode and
flowing electrolyte was achieved by positioning the tip of the
reference electrode slightly inside the flow channel which is connected to the electrolyte outlet. A photograph of the SDCM body
is presented in part (b) of Fig. 5 and the functions of all designed
electrolyte ports/holders are indicated.
2.3. New FT-SDCM test and measurement details
CAD files for CFD-simulations and 3D printing were created
using SolidWorks CAD software (Dassault Systems, France). The
customised flow-cell SDCM was manufactured on a ProJet 6000
stereolithography machine (3D Systems, Rock Hill, U.S.A.) using
VisiJet SL material (3D Systems, Rock Hill, U.S.A.). Data processing
for 3D printing was done using SLA-System Software (3D Systems,
Rock Hill, U.S.A.). To achieve high reproducibility of wetted area
the electrolyte was confined inside the flow cell using a sealing
ring installed at top of cell. This sealing was made from nitrile
butadiene rubber (NBR) with a measured thickness of about
870 lm (Semperit, Austria).The sealing ring had an inner diameter
of 2 mm matching exactly the geometry of the opening of the cell
and was manufactured using a custom punching tool. It was
assembled at the tip of the flow cell using cyanoacrylate glue
(UHU GmbH, Germany). To connect the flow channels to external
electrolyte reservoir and to downstream analytics screws equipped
with a barbed fitting were used. The screws had a central inner
diameter of 2 mm and were machined from a Ni rod. To increase
electrochemical stability the Ni-screws were coated with Au using
electroplating (Metakem, Germany). For additional sealing PTFE
tape was wrapped around the threaded part of the screws.
Tygon-tubes with an inner diameter of 1.85 mm were attached
to the barbed fittings of the screws to make final connections to
electrolyte reservoir and downstream analytics.
The investigated sample itself was mounted on a self-developed
motorized x-y-linear-stage with a travel range of 300 mm in each
direction. This large travel range allows for example easy scanning
of industrial grade steel plates. A threaded rod which is screwed
into the top of the cell is used to press the cell against the sample
by a self-developed high resolution linear-stage which is acting as
z-axis. This ensures a uniform distribution of the compression
forces acting on the sealing. The applied force is continuously
monitored by a force sensor (ME-Messsysteme GmbH, Germany)
and readjusted if necessary by feed back to the z-axis. The applied
force was 400 mN leading to a measured compression of the NBR
sealing to about 450 lm. When using softer sealing materials like
silicone or natural rubber derivatives the applied forces were
below 300 mN. The electrolyte is continuously pumped through
the cell by a peristaltic pump. The complete setup is controlled
by an in-house developed LabView program.
All chemicals used in the experiments were of analytical grade
with a minimum purity of 99.5%. All electrochemical experiments
were performed using a 3-electrode setup and a potentiostat (Amel
2059, Amel, Italy). Deionized water was used for preparation of all
solutions. To precisely determine the area being wetted by the cell
it was placed on a 1 mm thick Ti plate (99.6% purity, Alfa Aesar,
Germany) and anodised at 20 V under constant flow for 300 s in
0.1 M Na2SO4. Area of the coloured spot obtained after anodisation
was measured using a light microscope equipped with auto-colour
recognition software (Nikon NIS-Elements D, Nikon, Japan). To
study the effectiveness of the sealing on the micrometer scale a
Ni thin film on ITO-covered glass (Kintec Co.) was prepared by
thermal evaporation. The Ni film was dissolved at 0.1 M HCl using
the 3D printed cell at a potential of 0.7V (SHE) until the initial current value had dropped to a value of less then 98%. Topography of
the area near the sealing was studied using a Dektak Stylus Profilometer (Bruker Corporation, U.S.A.) equipped with a 2 lm radius
stylus.
Downstream analytics in all dissolution experiments were done
using a Hitachi Z-8230 polarized Zeeman atomic absorption
spectrophotometer (AAS) [33]. For chemical dissolution experiments a piece of a 2 mm thick Zn plate of 99.99% purity (Alfa Aesar,
Germany) was used. It was ground with Si–C grinding paper
(1200–4000 grades). Afterwards, the surface was rinsed with ethanol (p.a. grade VWR), deionized water and finally blown dry with
nitrogen.
To study the performance of the SDCM for electrochemical
dissolution experiments a 1 mm thick copper plate of 99.95% purity (Alfa Aesar, Germany) was used. Prior to the experiment the
copper plate was dipped in 5% nitric acid for 30 s to remove the
natural oxide layer. Next, it was carefully rinsed with water and
ethanol and finally blown dry with nitrogen.
58
J.P. Kollender et al. / Journal of Electroanalytical Chemistry 740 (2015) 53–60
3. Results and discussion
For studying the experimental behaviour of the cell, several different experiments were performed. In order to determine the area
wetted by the electrolyte and its reproducibility, a Ti plate was
anodised at ten different spots in 0.1 M Na2SO4 solution under constant flow (3 mL min 1) at 20 V for 300 s. The oxidised area had a
distinct colour which was used to determine the area effectively
wetted by the cell. In Fig. 6 a close-up photograph of an oxidised
spot is presented. The area of ten different spots was measured
using a light microscope equipped with auto colour recognition
software and found to be 3.57 ± 0.11 mm2. Three more examples
of addressed areas are presented in the lower part of Fig. 6. In
the high resolution optical imaging, the observed frazzled
boundary regions of the spot are due to the micro-roughness of
the Ti-surface caused by the polycrystalline nature of the material.
The measured 3% uncertainty of the wetted surface may be partially due to the material roughness and its small value indicates
a highly reproducible addressed area when automatic operation
of the SDCM is desired. The wetted area uncertainty can be
decreased below 1% if similar measurements are done on flatter
substrates (e.g. thin films deposited on glass or Si) but a worst case
scenario would be represented by a bulk sample, as used in the
present study.
In order to study the effectiveness of the sealing on the micro
scale and check for any possible crevice effects, a different experiment was designed. A Ni thin film deposited on an ITO-covered
microscope slide was used as a test substrate. The Ni film was electrochemically dissolved using the FT-SDCM under constant flow in
0.1 M HCl at 0.7 V (SHE) and the surface topography was studied
near the NBR sealing after the dissolution experiment was done.
The shape of the dissolved spot on the Ni film was mapped and
in Fig. 7 a cross section is presented as measured using a contact
profilometer. The height profile shown in Fig. 7 indicates that the
sealing used during the FT-SDCM dissolution of Ni is very effective.
At the left side of the profile (starting from position of 0 lm) a Ni/
ITO thin film thickness of approximately 210 nm is observed in a
first thickness plateau approximately 50 lm long. This plateau
resided under the NBR sealing and therefore the Ni dissolution
did not occur. At the other extremity starting from position of
Fig. 6. Optical images of the anodised Ti-samples used for determination of wetted
area-close-up of a single (top part) – three selected spots for reproducibility
measurements (lower part).
Fig. 7. Cross section showing the boundary area of electrochemically dissolved Nithin film on glass measured by a profilometer.
100 lm the flat glass substrate is evidenced by zero Ni/ITO thin
film thickness (the electrolyte used is also dissolving the ITO).
The transition between both thin film and bare glass plateaus
occurs in a distance of approximately 50 lm which characterizes
the thickness profile of the dissolution pit. This sharp edge of the
dissolution pit demonstrates an efficient sealing. Also, a high efficiency of dissolved material removal from the surface is indicated
by this simple experiment, as expected from the theoretical investigations. An inefficient removal of dissolved material would have a
high probability for re-deposition inside or along the edges of the
pit, which is not observed here.
Two additional types of dissolution experiments (chemical dissolution of Zn and electrochemical dissolution of Cu) combined
with online downstream analytics were carried out to evaluate
the efficiency of material removal when using the FT-SDCM by
quantifying the transported species. These results can be directly
used to crosscheck the data obtained from the CFD simulations.
In the first experiment, the chemical dissolution of Zn was tested.
The FT-SDCM was used to address the surface of a mechanically
polished zinc plate. With a flow rate of 3 mL min 1, 0.1 M HCl
was pumped through the cell into an AAS which continuously
measured the amount of dissolved Zn. Fig. 8 shows the total
amount of chemically dissolved zinc plotted against the time
during which the zinc surface was in contact with the flowing electrolyte. Each dissolution experiment was done on a separate spot.
The graph shows a nearly linear relationship between the contact
time and the amount of dissolved zinc for the first 12 min. After
approximately 12 min a slight increase in the amount of Zn dissolved per time unit is observed. A visual inspection of the
obtained corrosion pits on the Zn substrate using an optical
microscope revealed a significant increase of the surface roughness
for longer etching times. This is a known phenomenon appearing
Fig. 8. Amount of chemically dissolved zinc as measured by AAS over time.
J.P. Kollender et al. / Journal of Electroanalytical Chemistry 740 (2015) 53–60
during chemical etching of polycrystalline materials. The increase
of the effective surface due to roughing leads to higher etching
rates. This indicates that the slightly non-linear behaviour of the
dissolution curve presented in Fig. 8 is not a result of a sealing failure but rather a direct description of the increase of the effective
dissolution area. In the case of a NBR sealing issue, the shape of
the dissolved mass vs. time curve would be expected to show a
more accentuated deviation from a linear mass-time dependence
e.g. as a result of a pit diameter increase. Additionally, no change
of spot shape could be observed even for longer etching times.
In the second experiment, electrochemical dissolution of copper
was performed. The tip of the FT-SDCM was placed on a copper foil
and 0.1 M HCl was pumped through the cell. In a 3-electrode setup
a potential of 0.75 V (SHE) was applied for different time periods
(5, 10, 15, 20, 25 and 30 s) using a potentiostat. The dissolved
amount of copper was continuously measured using AAS and the
corresponding dissolution curves are shown in Fig. 9. In all experiments the curves have a small rise-time below 20 s. This slightly
longer time (as compared with expected times from the simulations) is related to the flow of electrolyte through the entire SDCM
and AAS tubing system. This issue had been addressed in an earlier
work by determining the retention time sum function for a flow
cell. It was shown that this value is about 17 s if coupled to the
AAS system that was also used in this work. This value agrees well
with rising and falling time of the concentration transients demonstrating that the delay and retention is due to the volume in the
tubing of the AAS. At the end of one dissolution sequence, after
switching off the applied voltage, a fast and complete removal of
all dissolution products from the flow system can be observed
for all experiments as indicated by reaching the background signal
level. The fall time of all curves is nearly independent from the
dissolution time which means that all dissolved species are immediately removed from the surface and transported out of the cell. In
turn, this suggests that no dead zones exist inside the cell at the
edges of the wetted areas, in agreement with the CFD-simulations.
Upon numerical integration of the time dependant Cu concentrations, as continuously measured by AAS (see Fig. 9), the mass
of electrochemically dissolved Cu can be calculated taking into
account the flow rate. The integration limits for each concentration
peak presented in Fig. 9 were chosen in a way coming closest to the
background values. Plotting the amount of dissolved Cu vs. dissolution time reveals a linear relationship as presented in Fig. 10.
The mass of dissolved Cu ranged from approximately 0.3 lg (after
5 s) to more than 2.1 lg (after 30 s). A linear regression was performed using the least square method and a fit precision of 0.995
was calculated. Similar to the previously presented case of chemical dissolution, this suggests that no electrolyte leakage occurred
and all the dissolved Cu was successfully transported away. The
combination of these two beneficial effects makes the proposed
59
Fig. 10. Amount of electrochemically dissolved copper as measured by AAS over
time. Values were determined by integrating data from Fig. 9 and multiplication
with the flow rate of 3 mL min 1.
improved design of a FT-SDCM suitable for online downstream
analytics.
4. Conclusion
In this article previously described designs of a FT-SDCM, the
theta capillary based approach and the V-shaped cell, were
carefully analysed by CFD-simulations. Based on the obtained simulation results an improved version of FT-SDCM with a special
geometry was designed. The complex geometry has proven to be
very difficult to be manufactured by conventional methods like
multi-axis CNC machining. To overcome this problem 3D printing
was used for the first time in the fabrication of an FT-SDCM. The
effectiveness of the sealing was checked on the microscale using
a high-precision contact profilometer with sub-nanometer resolution. Reproducibility of the area addressed by the FT-SDCM was
checked by anodisation of Ti. The area of the coloured oxide spot
was determined by means of computer-controlled pattern recognition software. To experimentally verify the results obtained by CFD
simulation two different types of dissolution experiments combined with online downstream-analytics via AAS were performed.
Both, the chemical dissolution of Zn and the electrochemical etching of Cu revealed that all species released from the surface are
transported out of the cell and no leakage was observed even in
long-time experiments. Also, no backmixing or trapping inside
the cell could be observed leading to peaks with small rise and fall
times in downstream analytics.
Conflict of interest
There is no conflict of interest.
Acknowledgments
The financial support by the Austrian Federal Ministry of
Science, Research and Economy and the National Foundation for
Research, Technology and Development through the Christian
Doppler Laboratory for Combinatorial Oxide Chemistry (COMBOX)
is gratefully acknowledged.
References
[1]
[2]
[3]
[4]
Fig. 9. Copper dissolution curves as measured by AAS over time. Arrows are
indicating duration of applied potential pulses.
A.W. Hassel, M.M. Lohrengel, Electrochim. Acta 42 (1997) 3327–3333.
A.I. Mardare, A.D. Wieck, A.W. Hassel, Electrochim. Acta 52 (2007) 7865–7869.
A.W. Hassel, M. Seo, Electrochim. Acta 44 (1999) 3769–3777.
S.O. Klemm, J.-C. Schauer, B. Schuhmacher, A.W. Hassel, Electrochim. Acta 56
(2011) 4315–4321.
[5] A. Vogel, J.W. Schultze, Electrochim. Acta (1999) 3751–3759.
[6] J.P. Kollender, A.I. Mardare, A.W. Hassel, Chem. Phys. Chem. 14 (2013) 560–
567.
60
J.P. Kollender et al. / Journal of Electroanalytical Chemistry 740 (2015) 53–60
[7] J.A. Haber, C. Xiang, D. Guevarra, S. Jung, J. Jin, J.M. Gregoire, Chem. Electro.
Chem. 1 (2014) 524–528.
[8] A.I. Mardare, A. Ludwig, A. Savan, A.D. Wieck, A.W. Hassel, Electrochim. Acta 55
(2010) 7884–7891.
[9] M. Sakairi, F. Sato, Y. Gotou, K. Fushimi, T. Kikuchi, H. Takahashi, Electrochim.
Acta 54 (2008) 616–622.
[10] A.I. Mardare, M. Kaltenbrunner, N.S. Sariciftci, S. Bauer, A.W. Hassel, Phys.
Status Solidi A 209 (2012) 813–818.
[11] C.M. Siket, A.I. Mardare, M. Kaltenbrunner, S. Bauer, A.W. Hassel, Electrochim.
Acta 113 (2013) 755–761.
[12] M.M. Mennucci, M. Sanchez-Moreno, I.V. Aoki, M.-C. Bernard, H.G. Melo, S.
Joiret, V. Vivier, J. Solid State Electrochem. 16 (2012) 109–116.
[13] M. Sánchez, J. Gamby, H. Perrot, D. Rose, V. Vivier, Electrochem. Comm. 12
(2010) 1230–1232.
[14] G. Buytaert, J. de Premendra, L. Wit, B. Katgerman, H. Kernig, H. Brinkman, Surf.
Coat. Tech. 201 (2007) 4553–4560.
[15] M. Breimesser, S. Ritter, H.-P. Seifert, S. Virtanen, T. Suter, Corros. Sci. 55 (2012)
126–132.
[16] M. Lohrengel, S. Heiroth, K. Kluger, M. Pilaski, B. Walther, Electrochim. Acta 51
(2006) 1431–1436.
[17] S. Hochstrasser-Kurz, D. Reiss, T. Suter, C. Latkoczy, D. Guenther, S. Virtanen,
P.J. Uggowitzer, P. Schmutz, J. Electrochem. Soc. 155 (2008) C415–C426.
[18] N. Homazava, A. Ulrich, M. Trottmann, U. Krähenbühl, J. Anal. At. Spectrom. 22
(2007) 1122.
[19] S. Lebouil, A. Duboin, F. Monti, P. Tabeling, P. Volovitch, K. Ogle, Electrochim.
Acta 124 (2013) 176–182.
[20] M. Serdechnova, P. Volovitch, F. Brisset, K. Ogle, Electrochim. Acta 124 (2013)
9–16.
[21] Versascan SDC, Princeton Applied Research, Oak Ridge (TN), USA.
[22] SDS 470, Bio-Logic SAS, Claix, France.
[23] A.K. Schuppert, A.A. Topalov, I. Katsounaros, S.O. Klemm, K.J.J. Mayrhofer, J.
Electrochem. Soc. 159 (2012) F670–F675.
[24] A.J. Capel, S. Edmonson, S.D.R. Christie, R.D. Goodridge, R.J. Bibb, M. Thurstans,
Lab Chip 13 (2012) 4583–4590.
[25] Shapeways Inc., New York, USA.
[26] Sculpteo, Issy les Moulineaux, France.
[27] Tinkercad, Autodesk Inc., San Rafael (CA), USA.
[28] SketchUp, Trimble Navigation, Sunnyvale (CA), USA.
[29] M.E. Snowden, P.H. King, J.A. Covington, J.V. Macpherson, P.R. Unwin, Anal.
Chem. 82 (2010) 3124–3131.
[30] A.J. Smith, S. Martin, A.W. Hassel, Electrochim. Acta 51 (2006) 6521–6526.
[31] 3D Laser Lithography, Nanoscribe GmbH, Karlsruhe, Germany.
[32] K.A. Lill, A.W. Hassel, J. Solid State Electrochem. 10 (2006) 941–946.
[33] M. Voith, G. Luckeneder, A.W. Hassel, J. Solid State Electrochem. 16 (2012)
3473–3478.