Coupling of a scanning flow cell with online electrochemical mass spectrometry forscreening of reaction selectivityJan-Philipp Grote, Aleksandar R. Zeradjanin, Serhiy Cherevko, and Karl J. J. Mayrhofer Citation: Review of Scientific Instruments 85, 104101 (2014); doi: 10.1063/1.4896755 View online: http://dx.doi.org/10.1063/1.4896755 View Table of Contents: http://scitation.aip.org/content/aip/journal/rsi/85/10?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Superhydrophilic graphite surfaces and water-dispersible graphite colloids by electrochemical exfoliation J. Chem. Phys. 139, 064703 (2013); 10.1063/1.4817680 Combinatorial microelectrochemistry: Development and evaluation of an electrochemical robotic system Rev. Sci. Instrum. 76, 062204 (2005); 10.1063/1.1906106 Electrochemical cell system for voltammetry of high purity solvents Rev. Sci. Instrum. 71, 516 (2000); 10.1063/1.1150233 A concerted assessment of potential-dependent vibrational frequencies for nitric oxide and carbon monoxideadlayers on low-index platinum-group surfaces in electrochemical compared with ultrahigh vacuumenvironments: Structural and electrostatic implications J. Chem. Phys. 111, 368 (1999); 10.1063/1.479279 Direct observation of infrared band intensity transfer between coadsorbates having widely separated oscillatorfrequencies: Intermixed NO/CO adlayers on ordered iridium electrodes J. Chem. Phys. 109, 4135 (1998); 10.1063/1.477018

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

129.21.35.191 On: Fri, 19 Dec 2014 15:50:36

REVIEW OF SCIENTIFIC INSTRUMENTS 85, 104101 (2014)

Coupling of a scanning flow cell with online electrochemical massspectrometry for screening of reaction selectivity

Jan-Philipp Grote,a) Aleksandar R. Zeradjanin, Serhiy Cherevko, and Karl J. J. Mayrhofera)

Department of Interface Chemistry and Surface Engineering, Max-Planck-Institut für Eisenforschung GmbH,Max-Planck-Strasse 1, 40237 Düsseldorf, Germany

(Received 27 June 2014; accepted 17 September 2014; published online 7 October 2014)

In this work the online coupling of a miniaturized electrochemical scanning flow cell (SFC) to amass spectrometer is introduced. The system is designed for the determination of reaction productsin dependence of the applied potential and/or current regime as well as fast and automated changeof the sample. The reaction products evaporate through a hydrophobic PTFE membrane into asmall vacuum probe, which is positioned only 50–100 μm away from the electrode surface. Theprobe is implemented into the SFC and directly connected to the mass spectrometer. This uniqueconfiguration enables fast parameter screening for complex electrochemical reactions, includinginvestigation of operation conditions, composition of electrolyte, and material composition. Thetechnical developments of the system are validated by initial measurements of hydrogen evolutionduring water electrolysis and electrochemical reduction of CO2 to various products, showcasing thehigh potential for systematic combinatorial screening by this approach. © 2014 AIP Publishing LLC.[http://dx.doi.org/10.1063/1.4896755]

I. INTRODUCTION

The increasing importance of renewable energies boostedinterest for sustainable energy conversion and storageconcepts like water electrolysis or electrochemical CO2reduction.1 The development and implementation of thesetechniques relies to large extent on fundamental research ofelectrochemical reactions, which are still insufficiently un-derstood to date.2 To gain adequate insights and develop theknowledge about the phenomena at the electrochemical inter-face it is of essential importance to design and employ appro-priate analytical tools.

One of the key aspects in the study and in optimizationof electrochemical reactions is the understanding of reactionpathways, which can be achieved via analysis of reactionproducts in parallel to electrochemical measurements. Thetime-resolved correlation of electrode potentials and faradaiccurrents to fragment intensities in mass spectrometers is cer-tainly one of the most illustrative approaches. Bruckenstein3

and his coworkers developed the so called membrane intro-duction mass spectrometer (MIMS), which builds the funda-ment for similar analysis techniques used today.4, 5 In partic-ular differential electrochemical mass spectrometry (DEMS)has been widely utilized in research for the investigation ofgaseous/low volatile products dissolved in the electrolyte.6–9

Although DEMS was successfully assembled by Wolterand Heitbaum10 already 30 years ago, no commercial systemis available up to date. Most of the used systems share thebasic principle of reaction product evaporation through a hy-drophobic PTFE membrane into the mass spectrometer. In aclassical DEMS setup this membrane is coated with catalystmaterial by, for instance, sputter deposition and can there-

a)Authors to whom correspondence should be addressed. Electronic ad-dresses: grote@mpie.de and mayrhofer@mpie.de. Tel.: +49 211 6792 160.Fax: +49 211 6792 218.

fore be directly used as working electrode. Those setups showvery fast response time and high sensitivity, but are not suit-able for solid electrodes whereby single crystals cannot beinvestigated.11 This issue is, e.g., addressed by the online elec-trochemical mass spectrometry (OLEMS), where the PTFEmembrane is placed very close (10 μm) to an independentworking electrode and therefore allows the use of variouselectrode types.4

One of the present challenges is the more or less com-plex electrode exchange that, in combination with the widenumber of alterable parameters in electrochemical measure-ments (electrode material, operation conditions, electrolyte,and so on), lead to time consuming work in the lab. Related tothis issue, a high throughput combinatorial approach for fastparameter and product analysis can be a strong support forcurrent efforts in catalyst development and can also provideexperimental data theoreticians need to verify their models.

After successful development of an electrochemicalscanning flow cell and its coupling to an inductive coupledmass spectrometer for catalyst dissolution analysis,12 the SFCis now coupled to a mass spectrometer with electron impactionization. This system combines the fast parameter screen-ing possibilities of the SFC with the direct product analysisof an OLEMS. First results investigating hydrogen evolutionand CO2 reduction on copper electrodes are presented in thiswork, in order to confirm the feasibility of the approach.

II. EXPERIMENTAL

The electrochemical scanning flow cell is mounted in afixed position over a xyz-translational stage (Physik Instru-mente, Germany) that holds the working electrode and en-ables fast and precise positioning of the sample. The changeof position on the working electrode is achieved via move-ment in xy-direction during non-contact time, while electrical

0034-6748/2014/85(10)/104101/5/$30.00 © 2014 AIP Publishing LLC85, 104101-1

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

129.21.35.191 On: Fri, 19 Dec 2014 15:50:36

104101-2 Grote et al. Rev. Sci. Instrum. 85, 104101 (2014)

FIG. 1. (a) Detailed scheme of a classical SFC13, 15–17 with possible coupling to ICP-MS in comparison with (b) the modified SFC for SFC-OLEMS coupling.Reaction products can evaporate through a hydrophobic PTFE membrane, which is mounted ≈50 to 100 μm away from the electrode surface, into the vacuumsystem of the mass spectrometer. (CE: counter electrode, RE: reference electrode).

contact is established when the working electrode and theSFC are approached towards each other. The applied force be-tween working electrode and SFC is controlled by a force sen-sor and is typically adjusted to 1500 mN in contact mode. Theclassical design of the SFC has been modified in this workto enable coupling to the mass spectrometer for detection ofvolatile species.13 A comparison between the classic and theadjusted cell configuration is shown in Figure 1. A more de-tailed scheme is available in the supplementary material.14

In comparison to earlier versions, the channel diameteris extended to 3 mm. This provides sufficient space to in-troduce the PTFE tip from the top of the cell through anextra vertical channel. A 50 μm thick PTFE Gore-Tex mem-brane with a pore size of 20 nm, through which products canevaporate into the vacuum system of the mass spectrometer,is mounted onto the very end of the tip. The small pore sizeand the thickness of the membrane lead to the reduction ofthe water background pressure inside the vacuum and pro-vide a good response time. The position of the tip with re-spect to the working electrode is controlled via the thicknessof the silicone sealing around the cell opening and the appliedcontact force. It is optimized to achieve (1) good responsetime by short diffusion paths and (2) high sensitivity by highproduct concentration. With a ring thickness of 125 μm, amembrane thickness of 50 μm, and the typically applied pres-sure, the approximate distance from the tip to the electrode is50–100 μm.

A graphite counter electrode, made out of a 20 mm longgraphite pipe with an inner diameter of 3 mm, is mounted atthe outlet of the SFC to avoid influences of the counter reac-tion on the mass spectrometer signal. The high surface arealeads to lower current densities and therefore to lower cur-rent fluctuations due to bubble formation and/or detachment.A Ag/AgCl/3 M KCl (Metrohm, Switzerland) reference elec-trode is mounted at the inlet channel of the SFC, all poten-tials are provided versus Standard Hydrogen Electrode (SHE).For the evaluation of the system a 0.5 mm copper foil with a

purity of 99.99% is used as a cathode material, which has beencleaned with sandpaper and electropolished in 5% H3PO4(pH 2) at 1.5 V for 1 min. The electrolyte consists of a0.1 mol KHCO3 (Honeywell Specialty Chemicals Seelze.Germany) aqueous solution prepared from ultrapure water(PureLab Plus system, Elga, 18 M�, TOC < 3 ppb) and isstarted to be purged with Argon or CO2 30 min before themeasurements. No systematic IR drop correction is performedat this stage, although required for more exact evaluationswith this large SFC. The slightly oval contact area of the elec-trolyte is estimated to be 12.5 mm2 by anodic polarizationof Cu and subsequent measurement of the differently coloredsurface by an optical microscope.

A potentiostat (Reference 600, Gamry Instruments,USA) is used to perform electrochemical measurements. Sat-uration of electrolyte with different gases is realized by a gaspurging system, while the electrolyte flow over the electrodesurface is imposed by a peristaltic pump. The utilized Vitontubes (Saint-Gobain Performance plastics, France) with lowgas permeation rates avoid the loss of dissolved gases and by-pass diffusion of atmospheric gases into the electrolyte. More-over, additional side purging channels in the SFC provide op-tional Argon flow around the cell opening, with what ambientair diffusion through the silicone sealing is suppressed.

All hardware of the system is controlled by an in housedeveloped LabVIEW software, which enables full automationof electrochemical measurements.18

Figure 2 shows a scheme of the coupling between theSFC and the mass spectrometer. A 1/16 in. steel pipe serves asconnection between the SFC and the mass spectrometer vac-uum system and gives best sealing against atmospheric gases.Moreover, an electrical heating wire tempers the inlet systemto roughly 65 ◦C and therefore prevents condensation and en-hances desorption of reaction products from the inner pipewall. The usage of a pre-pump (Trivac D10E, Pfeiffer Vac-uum, Germany) enables inlet-chamber pressure adjustmentvia a metering valve. The second valve functions as a pin hole

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

129.21.35.191 On: Fri, 19 Dec 2014 15:50:36

104101-3 Grote et al. Rev. Sci. Instrum. 85, 104101 (2014)

FIG. 2. Scheme of the coupling between SFC and mass spectrometer. Pres-sure can be controlled by pre-pump and metering valves (V).

with a variable hole size and grants control over gas flow be-tween inlet- and main-chamber. A membrane inlet mass spec-trometer (MIMS) (Max 300 LG, Extrel, USA) is modifiedthrough removal of the original membrane and connection tothe above described inlet system. Evaporated molecules areionized by electron impact ionization at an operation pressureof ∼5 μTorr, while after mass selection either a faraday detec-tor or an electron multiplier can be used to detect molecules.Mass to charge ratios ranging from 2 to 225 amu can be mea-sured down to ppm level with the faraday plate or down toppb by the electron multiplier.

III. RESULTS AND DISCUSSION

The evaluation of the OLEMS coupled to a SFC is firstperformed by investigating the hydrogen evolution on cop-per electrodes. Both steady state and transient measurementsshow the relation between the electrochemical and mass spec-trometry data.

Five potential steps between 0.1 V and different lowerpotential limits, increasing from −0.8 V to −2 V, are appliedwith a duration of 30 s each. The base potential in between thesteps is set to 0.1 V for 120 s (Figure 3(a)). The correspondingcurrent and the intensity of m/z = 2 fragment for hydrogen areshown in the upper part of the figure. Due to the rather com-plex technical solution for the time-resolved study of prod-uct distribution in a screening approach, the mass spectrom-eter intensity is not calibrated for absolute concentrations butarbitrary values are given. Nevertheless, a semi-quantitative,comparative analysis of the selectivity between different mea-surements with the same cell is still feasible. The mass spec-trometer response is synchronized with the potentiostat signalwithout correction for any delay time. The response time forhydrogen from its origin at the electrode surface to the de-tector in the mass spectrometer is in the range of 1–3 s. Incomparison to that, the recovery time of the mass signal tothe base line is ∼1.5 min for −2 V, and somewhat shorterfor more positive potentials, although the current signal dropsinstantaneously as expected. The long recovery time can be

FIG. 3. Steady state and transient measurements, as well as the resulting calibration curve. The mass spectrometer signal versus time recorded for (a) potentialsteps, (b) current steps, (c) cyclic voltammograms (CV) with different upper potential limits. The applied or measured current signal is overlaid in red, whilethe corresponding potential profile is shown below the graphs (a), (b), and (c). The integrated charge is compared to the measured hydrogen amount in (d). Datafrom 3 different measurement locations for each method are summarized to confirm the reproducibility.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

129.21.35.191 On: Fri, 19 Dec 2014 15:50:36

104101-4 Grote et al. Rev. Sci. Instrum. 85, 104101 (2014)

due to two reasons, namely (1) an insufficient mass trans-port in the proximity of the tip causing an enrichment of theelectrolyte with the reaction product and eventually even bub-ble formation and/or (2) too slow evacuation of the productfrom the vacuum system. At high potentials and high bub-ble evolution rates the system shows stable signals with suffi-ciently low noise. Figure 3(a) indicates that the detection limitlies between the first and the second potential step. This re-flects a current density of ∼0.8 mA cm−2 and a correspondinghydrogen production rate of 4 × 10−9 mol cm−2 s−1.

Figure 3(b) shows the galvanostatic measurements withsteps between 1 mA cm−2 and 15 mA cm−2, with a spec-trometric signal closely resembling the one recorded in thepotentiostatic regime. Differences in current densities for ap-proximately the same potentials in (a) and (b) may arisefrom unavoidable time-induced decay in performance due tobubble formation and CO poisoning. The pre-history of theelectrode, in this particular case depending on the amountof potential/current steps and the rest time at the base-potential/current, obviously plays an important role.

Figure 3(c) shows cyclic voltammograms with differentupper potentials ranging from −1.25 to −2 V with a scan rateof 10 mV s−1. Again the mass spectrometer signal is follow-ing the signal of the measured current without significant de-lay and with sufficiently short recovery time to resolve the CVeven at this scan rate.

The reproducibility of the measurements is confirmedby repeating the described methods on three different pointseach. The resulting current i of each reduction feature is inte-grated (Eq. (1)), and the obtained charge Q is plotted againstthe integral A under the corresponding mass spectrometer sig-nal I (Eq. (2)). The boundaries for integration are chosen asindicated in Figure 3(a):

Q =∫ t1

t0

i dt, (1)

A =∫ t2

t0

I dt. (2)

The deviation from the expected linear dependence ofA to Q in Figure 3(d) is due to lower collection efficiencyat higher reaction rates. An improved product transport intothe bulk electrolyte passing the PTFE tip could be causedby enhanced bubble evolution rates, however, the true rea-son remains unknown. Nevertheless, all data points can beproperly fitted by an exponential curve, which proof the re-spectable reproducibility and reliability of the system forsemi-quantitative analysis in combination with the screeningapproach.

The clear strength of mass spectrometry is the poten-tial for parallel multi-component analysis. An intensively dis-cussed electrochemical reaction at the moment is the reduc-tion of CO2, which, depending on the choice of electrode ma-terial, can proceed via various reaction pathways and leadto products like H2, CO, HCOOH, C2H4, CH4, CH3OH,and many more.19, 20 Figure 4 shows two cyclic voltammo-grams, one in Ar purged (pH 9.5) and one in CO2 purged(pH 7.3) 0.1 M KHCO3 electrolyte. Analysis of differencesin current-voltage dependence seems to be obvious, but cau-

FIG. 4. Cyclic voltammograms of a Cu electrode in Ar and CO2 purgedKHCO3 electrolyte. The CVs are recorded with a scan rate of 50 mV s−1

from −0.4 V to −1.6 V vs. SHE. Copper oxide reduction peaks were removedby previous potentiostatic polarization at −1.0 V for 30 s and at −0.2 V for10 s.

tion should be exercised, when conclusions about CO2 reduc-tion are drawn just from electrochemical data. It is especiallycritical to reason higher current densities in carbon dioxidesaturated solution to CO2 reduction, because on the one handthe decreased pH could lead to an increased hydrogen evolu-tion current, while on the other hand possible CO adsorptionduring CO2 reduction might induce the opposite.19 In our casethe latter effect seems to play a role, as the current density forCO2 saturated electrolyte is lower than the one in Ar purgedsolution. A valid statement about processes during CO2 re-duction cannot be accomplished just from this data, as CO2reduction can only be confirmed when proper product analy-sis is done. This again emphasizes the need of an instrumentfor precise online product analysis including the possibility toinvestigate transient electrochemical measurements.

In order to illustrate the selectivity of the reaction, H2,CH4, and C2H4 are recorded as relevant products at a massto charge ratio of 2, 15, and 26, respectively. Despite the lowmasses used for detection, only low influences of other prod-ucts are expected on these fragments.

In Figure 5 a potential sweep from 0 to −2 V vs. SHE andthe intensities of the different fragments are shown. The signalis corrected for the baseline intensity (Ibaseline) and divided bythe maximum intensity (Imax), like shown in Eq. (3):

Inormalized = I − Ibaseline

Imax

. (3)

The onset potentials for the different reactions are a first semi-quantitative descriptor of the electrocatalytic performance forthe generation of certain products. The hydrogen evolutiononset potential is ∼−1.12 V, ethylene production starts at−1.53 V, and methane at −1.64 V. Obviously, hydrogen evo-lution starts at a lower overpotential than the reduction of CO2under these conditions. For an exact comparison of the re-duction rates, however, a more quantitative assessment usingother analytical techniques would be necessary. Interestingly,the onset potentials are 200 to 300 mV more negative than re-ported ones in literature.6, 19 This shift might be a consequence

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

129.21.35.191 On: Fri, 19 Dec 2014 15:50:36

104101-5 Grote et al. Rev. Sci. Instrum. 85, 104101 (2014)

FIG. 5. Linear potential sweep in CO2 saturated 0.1 M KHCO3 on a copperelectrode (Scan rate: 10 mV s−1). Evolution of hydrogen (m/z = 2), ethylene(m/z = 26) and methane (m/z = 15) is detected by mass spectrometer for thepurpose of estimation of selectivity and onset potentials.

of different effects. Firstly, the electrochemical CO2 reduc-tion has been conducted without previous pre-electrolysis ofthe electrolyte, so the influence of impurities is not negligible.Secondly, no particular pretreatment of the copper sample hasbeen used in this study with the focus on system evaluation,although it is well known that the purity is important.21 Inaddition to that, the system has different response times fordifferent products, due to differences in diffusion rates in theelectrolyte, transfer rates through the membrane, and adsorp-tion behavior in the inlet system. For methane and ethylenethe response time is 8 s in the current system, which will re-sult in a shift of 80 mV for the shown sweep measurement,while for hydrogen this shift is only about 20 mV. Further-more, the influence of IR drop can lead to a negative shiftof the onset potentials of products that appear at high currentdensities. Nevertheless, the trends in onset potentials respondreasonably well to the literature trends, even without correc-tion for these differences.

IV. SUMMARY

A combination of an electrochemical scanning flow cellwith online mass spectrometry together with a gas purgingsystem has been developed. This setup enables fast screeningof electrochemical reactions with analysis of volatile reactionproducts in dependence of material and electrolyte parame-ters. The operation in various steady state and transient modesfor hydrogen evolution and CO2 reduction are shown to

confirm reproducibility. Addressing the issues of current sim-ilar online product analysis systems, the measurement timefor different parameters is reduced to a level where screen-ing investigations become possible. Due to these advantagesit is expected that this approach will lead to acceleration of(1) catalyst material research and of (2) determination of op-timal conditions for operation of practical devices.

ACKNOWLEDGMENTS

We thank Dr. Klaas Jan Schouten and Professor MarcKoper from the University of Leiden, Netherlands andDr. Abd El Aziz Abd El Latif and Professor Helmut Bal-truschat from the University of Bonn, Germany, for thediscussions during the design phase and the BMBF for thefinancial support in the framework of the project ECCO2(Kz: 033RC1101A).

1N. Lewis, MRS Bull. 32, 808 (2007).2I. Katsounaros, S. Cherevko, A. R. Zeradjanin, and K. J. J. Mayrhofer,Angew. Chem., Int. Ed. Engl. 53, 102 (2014).

3S. Bruckenstein and R. R. Gadde, J. Am. Chem. Soc. 93, 793 (1971).4A. H. Wonders, T. H. M. Housmans, V. Rosca, and M. T. M. Koper, J. Appl.Electrochem. 36, 1215 (2006).

5A.-E.-A. A. Abd-El-Latif, J. Xu, N. Bogolowski, P. Königshoven, and H.Baltruschat, Electrocatalysis 3, 39 (2011).

6K. J. P. Schouten, Y. Kwon, C. J. M. van der Ham, Z. Qin, and M. T. M.Koper, Chem. Sci. 2, 1902 (2011).

7E. Mostafa, A.-E.-A. A. Abd-El-Latif, R. Ilsley, G. Attard, and H. Baltr-uschat, Phys. Chem. Chem. Phys. 14, 16115 (2012).

8S. J. Ashton and M. Arenz, J. Power Sources 217, 392 (2012).9Z. Jusys, J. Kaiser, and R. J. Behm, Electrochimica Acta 47, 3693(2002).

10O. Wolter and J. Heitbaum, Ber. Bunsenges. Phys. Chem. 88, 2 (1984).11H. Baltruschat, J. Am. Soc. Mass Spectrom. 15, 1693 (2004).12S. O. Klemm, A. A. Topalov, C. A. Laska, and K. J. J. Mayrhofer, Elec-

trochem. Commun. 13, 1533 (2011).13A. K. Schuppert, A. A. Topalov, I. Katsounaros, S. O. Klemm, and K. J. J.

Mayrhofer, J. Electrochem. Soc. 159, F670 (2012).14See supplementary material at http://dx.doi.org/10.1063/1.4896755 for

technical drawing of the SFC-OLEMS design, (n.d.).15S. Cherevko, A. A. Topalov, A. R. Zeradjanin, I. Katsounaros, and K. J. J.

Mayrhofer, RSC Adv. 3, 16516 (2013).16A. A. Topalov, I. Katsounaros, M. Auinger, S. Cherevko, J. C. Meier, S. O.

Klemm, and K. J. J. Mayrhofer, Angew. Chem., Int. Ed. Engl. 51, 12613(2012).

17A. R. Zeradjanin, A. A. Topalov, Q. Van Overmeere, S. Cherevko, X. Chen,E. Ventosa, W. Schuhmann, and K. J. J. Mayrhofer, RSC Adv. 4, 9579(2014).

18A. A. Topalov, I. Katsounaros, J. C. Meier, S. O. Klemm, and K. J. J.Mayrhofer, Rev. Sci. Instrum. 82, 114103 (2011).

19Y. Hori, Mod. Asp. Electrochem. 42, 89 (2008).20K. P. Kuhl, E. R. Cave, D. N. Abram, and T. F. Jaramillo, Energy Environ.

Sci. 5, 7050 (2012).21R. Kas, R. Kortlever, A. Milbrat, M. T. M. Koper, G. Mul, and J. Baltru-

saitis, Phys. Chem. Chem. Phys. 16, 12194 (2014).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

129.21.35.191 On: Fri, 19 Dec 2014 15:50:36