JOURNAL OF APPLIED ELECTROCHEMISTRY 28 (1998) 10±17

Role of mass transfer on hydrogen evolution

in aqueous media
H. RIEGEL, J. MITROVIC, K. STEPHAN
Institut fuÈr Technische Thermodynamik und Thermische Verfahrenstechnik, UniversitaÈt Stuttgart,
70550 Stuttgart, Germany

Received 20 August 1996; revised 8 April 1997

The process of hydrogen evolution during alkaline electrolysis of aqueous solutions is governed by
mass transfer, growth of hydrogen bubbles and removal of hydrogen from the cathode. Two
mechanisms are decisive for hydrogen removal: (i) hydrogen dissolved in the solution is carried o€
from the cathode surface by di€usion and convection, and (ii) gas bubbles are transported by a two-
phase ¯ow. The paper describes experiments to determine the local concentration of dissolved
hydrogen and the void fraction of hydrogen bubbles in aqueous solutions. Measurements were
performed in a ¯ow channel by varying the height of the cathode (40±400 mm), the current density
(up to 6250 A mÿ2 ) and the mean velocity of the electrolyte (up to 0:95 m sÿ1 ). Two operating
regimes of the electrolyser are found. At high current densities a back ¯ow is observed leading to an
increase in the electrolyte resistance. Traces of dissolved oxygen are detected at high current densities.
At low current densities the two-phase ¯ow is con®ned to a thin layer along the cathode surface, the
concentration of dissolved hydrogen being small.

Keywords: hydrogen evolution, mass transfer, bubble growth, ¯ow channel, two-phase ¯ow, dissolved hydrogen

Nomenclature w mean velocity of electrolyte

c concentration Greek symbols

E electrical potential e void fraction
F Faraday constant # temperature in  C
I electrical current
K Henry coecient Subscripts
N number of moles AgCl silver/silver chloride reference electrode
N_ ¯ow rate El electrolyte
p pressure H‡ hydrogen ion
R electrical resistance H2 hydrogen molecule
Rm universal gas constant 0 without gas bubbles
T temperature in K sat saturated
U voltage

1. Introduction MuÈller et al. [1] investigated the concentration

pro®le with a platinum wire electrode as a function of
Modelling hydrogen production at a gas evolving the cathode distance. They assumed the solution to be
electrode requires a detailed knowledge of the mass saturated with hydrogen at a distance of about
transfer mechanisms. The mass transfer depends 500 lm and correlated the measured potentials of the
mainly on the concentration of the reaction products, supersaturated solution at smaller distances by the
which can be ascertained reliably only by experiment. Nernst equation. The potential was measured against
Such experiments are dicult to perform, and only the cathode, under the assumption that it can be
one experimental study, dealing with concentration described as a function of the measured IR-drop and
pro®les of hydrogen at a gas evolving cathode, seems the Nernst potential. We suggest a similar method to
to exist [1]; by contrast, a greater number of publi- determine the hydrogen concentration with a plati-
cations are devoted to mass transfer at hydrogen num wire electrode. However, we use a reference
electrode, measure the voltage between these elec-
evolving electrodes [2±4]. All these experiments were
trodes, and determine the concentration without any
performed with a tracer. The transfer of the tracer to
assumptions.
the electrode was measured, instead of the mass
The in¯uence of gas bubbles on the electrical re-
transfer of the reaction product from the electrode
sistance and methods of minimizing the resistance
into the liquid bulk.
10 0021-891X Ó 1998 Chapman & Hall
ROLE OF MASS TRANSFER ON HYDROGEN EVOLUTION 11

have been previously investigated [5±16]. The pio- does not exist, and a potential de®ned by Equation 2
neering paper was that of Tobias [5]. Hine et al. [6±9] cannot be assumed. Therefore, the metal has to act as
investigated the resistance of the electrode as a catalyst adsorbing hydrogen atoms. Because of the
function of current density, temperature, volume rate term ln cH2 in Equation 2 small concentrations of
and channel geometry. Two simple gas-bubble dis- dissolved hydrogen can be detected. However, small
tribution-functions were suggested: a linear decrease concentrations of impurities can also in¯uence or
of the void fraction with the cathode distance or a govern the voltage. In order to verify the validity of
quasihomogenous two-phase ¯ow region with a Equation 2 we checked the correlation between
bubble-free liquid outside this region. Sillen et al. voltage and concentration experimentally.
[10, 11], Janseen et al. [12±16] and Bongenaar-Schl-
enter et al. [17±19] studied the electrical resistance 3. Experiments
and the behaviour of gas bubbles during alkaline
electrolysis. They also investigated optically the dis- The experimental setup to verify Equation 2 is shown
tribution of bubbles perpendicular to the electrode in Fig. 1. It consists mainly of two electrodes, a
surface. However, these methods are not applicable platinized platinum wire (1) as redox and a silver/
to high current densities resulting in a high hydrogen silver chloride (2) as reference electrode. Both elec-
production rate. trodes were arranged in a closed plexiglass tank ®lled
Theoretical models for calculating the electrical with 8:2 dm3 of 1 M potassium hydroxide solution,
resistance of an electrolyser assume a constant initially saturated with air. The voltage between the
void fraction perpendicular to the electrode-surface two electrodes (3) was measured as a function of time
[21, 22]. Vogt [20] suggested a model which is based for a given constant hydrogen ¯ow rate. Hydrogen
on variable void fraction. was produced at the cathode (4), its area being
0:02 m2 . A magnetic stirrer (5) assured a homoge-
2. Theory nous distribution of the hydrogen in the potassium
hydroxide solution. A heater (6) kept the temperature
At the surface of a hydrogen electrode electrons in the constant at 25  C, and a constant current of 20 mA
platinum electrode, hydrogen ions in solution, and was imposed from the power supply (7), leading to a
dissolved molecular hydrogen are in equilibrium. The constant hydrogen ¯ow rate of N_ H2 ˆ 1:04  10ÿ7
concentration of the dissolved hydrogen is normally mol sÿ1 . The circuit was completed with an anode (8)
an independent variable [23]. When equilibrium is arranged in an open tank and a PTFE-tube (9) ®lled
maintained and when the partial pressure of hydro- with KOH solution. Diaphragms arranged between
gen in the gas phase is ®xed, the potential of the the tanks prevented mixing of dissolved gases.
hydrogen electrode is given through the well known A catalytic layer of black platinum on the plati-
expression (e.g. [23]). num wire was prepared with a solution of hexa-
chloro-platinum acid.
Rm T c ‡ Rm T pH
Ehydrogen ± electrode ˆ ln H ÿ ln 2 1† The volumetric void fraction of hydrogen and its
F c0;H‡ 2F p0;H2 distribution in forced convection ¯ow was deter-
The standard conditions are de®ned by cH‡ ˆ 1M and mined by measuring the electrical resistance of the
pH2 ˆ 1 bar. The partial pressure pH2 in Equation 1 two-phase mixture ¯owing up a vertical channel of
can be replaced by the concentration of the dissolved rectangular cross section. The apparatus used in these
gas using the relation for phase equilibrium. Henry's experiments is shown in Fig. 2. It consists of two
law pH2 ˆ KcH2 † is suitable for low partial pressures closed circuits, one for hydrogen and the other for
pH2 < 2 bar and # ˆ 25  C†. The proportionality oxygen. The gases were evolved on ten electrode pairs
constant, K, has a value of ˆ 1:723 bar m3 molÿ1 in mounted in the upper section of the electrolyser.
1M potassium hydroxide [24]. Thus, assuming equi- Cathodes and anodes were made of nickel, their size
librium between dissolved molecular hydrogen and was 4 cm  2 cm. The channel was divided vertically
the hydrogen ions, the voltage between a platinum by a diaphragm into two subchannels, and the elec-
electrode and a silver/silver chloride reference elec- trolyte pumped through the electrolyser at velocities
trode is a function of the amount of dissolved gas. up to 1:0 m sÿ1 . At the end of the channel, the two
From Equation 1 for a one molar hydroxide solution separated two-phase ¯ows streamed into reservoirs
cH‡ ˆ 10ÿ14 M†, as used in the experiments, where the liquids were purged from gases. The elec-
trolyte then passed through a heat exchanger, where
Rm T the temperature was regulated in the range from
UPt;AgCl ˆ EPt ÿ EAgCl ˆ ÿ1:029 ÿ ln cH2 2†
2F 20 to 60  C. The whole electrolyte volume amounted
where cH2 is given in mol mÿ3 and UPt;AgCl in V. The to about 100 dm3 .
equation is only valid for constant concentration of At the upper end of the channel, just above the last
H‡ . At large values of current density deviations can electrode pair, nine platinum electrodes (Pt-E1) were
occur if the concentration is strongly a€ected by mass mounted perpendicular to the cathode surface, Fig. 3.
transfer to the electrode [25]. At the surface of an The electrical current through the last electrode pair
inert metal electrode placed in a solution, containing consisted of a direct current superimposed on an
hydrogen molecules and hydrogen ions, equilibrium alternating current of small amplitude. A bipolar
12 H. RIEGEL, J. MITROVIC AND K. STEPHAN

Fig. 1. Experimental setup.

operational ampli®er (BOP) served as current source. REl 1 ‡ e=2

The frequency of the alternate current corresponded ˆ 3†
R0;El 1ÿe
to the frequency of an external alternating voltage,
supplied by the oscillator of a lock-in ampli®er, used to where REl represents the electrical resistance with, and
measure the a.c.-voltage between neighboured plati- R0;El without, gas bubbles.
num electrodes at the frequency of the oscillator. The When determining the concentration of dissolved
frequency of 1.1 kHz was high enough to avoid po- hydrogen the d.c. voltage between platinum electrode
larizing e€ects on the electrodes. With this technique, and reference electrode must be considered to consist
the electrical resistance of the ¯uid was measured as a of two additive parts. One is caused by the electrical
function of distance. The location of the platinum current through the electrolyte. This part, REl I, is the
electrodes (diameter 100 lm) was determined from the product of the total current produced by the BOP
electrolyte resistance without gas bubbles. and the resistance between the two electrodes (plati-
As is well known, the resistance of the electrolyte num and reference) which is also measured. The
depends on the void fraction. Several relations cor- second part is the di€erence between the e.m.f.'s of
relate the resistance and the void fraction e. Kreysa the two electrodes, given by Equation 2. Determina-
and Kuhn [26] examined the di€erent relations and tion of dissolved hydrogen therefore requires a mea-
recommended those derived by Bruggemann [27] and surements of the d.c. voltage between the platinum
Maxwell [28]. Since these equations di€er only slightly electrode and the reference electrode. From the
from each other, we used the Maxwell equation voltage is then obtained cH2 (see Equation 2).

Fig. 2. Apparatus for two-phase experiments. Fig. 3. Upper part of the channel.
ROLE OF MASS TRANSFER ON HYDROGEN EVOLUTION 13

Fig. 4. Voltage between platinum electrode and silver/silver chloride reference electrode in 1 M KOH solution as function of the amount of
hydrogen.

4. Results according to which, at NH2  3  10ÿ3 mol hydrogen

content, the concentration cH2 in the solution is zero.
The voltage between the platinum electrode and the As can be seen from the Figure, experiment and
silver/silver chloride reference electrode is plotted in theory agree well. Equation 2 can thus be used to
Fig. 4 as a function of the amount of hydrogen. Such determine the concentration of dissolved hydrogen at
a curve is typical for a redox titration. It can roughly a voltage below ÿ0:7 V.
be divided into three sections. The ®rst part is char- As the experiments showed, hydrogen reacts ®rst
acterized by a ¯at slope and a point of in¯ection. with an oxidizing agent. This is dissolved oxygen. In
Then follows a short transitional region with a steep the solution of distilled water and potassium hy-
decrease passing through another in¯ection point. droxide saturated with air the solubility amounts to
Finally, the voltage decreases slowly. The slow de- 0:926 mol mÿ3 at a partial pressure of oxygen of 1 bar
crease of the voltage in the ®rst section is caused by [24]. The amount of oxygen is 1:5  10ÿ3 mol at a
an oxidizing agent, which reacts with the reducing partial pressure of 0.2 bar in 8:2 dm3 of solution.
agent hydrogen. The reduction is completed at Hydrogen reduces oxygen on the platinum surface
amounts of hydrogen 3  10ÿ3 mol. After this, a according to O2 ‡ 2H2 ! H2 O. Therefore, twice as
further supply of hydrogen leads to an increase in the much hydrogen is consumed as oxygen is initially
hydrogen concentration causing a slight decrease in present in the solution. Since no hydrogen is present
the voltage with the amount of hydrogen. in the tank at the beginning of the titration, the volt-
Figure 5 shows the measured voltage as a function age cannot be determined from Equation 2. The
of the hydrogen concentration together with values voltage near the in¯ection point corresponds to the
calculated by Equation 2. The hydrogen concentra- standard potential of the titrated redox couple and,
tion follows from as Fig. 6 shows, amounts to ÿ84 mV.
The standard potentials of some platinum/plati-
NH2 ÿ 3:0  10ÿ3 mol num oxide couples [29] are also presented in Fig. 4.
cH 2 ˆ 4†
VKOH None of them ®ts the experiments. Obviously, a

Fig. 5. Voltage between platinum electrode and silver/silver chloride reference electrode as a function of the hydrogen concentration. The
®gure represents a comparison of measured and calculated data.
14 H. RIEGEL, J. MITROVIC AND K. STEPHAN

Fig. 6. Voltage between platinum electrode and silver/silver chloride reference electrode as a function of the supplied amount of hydrogen.
The voltage is governed by oxygen initially dissolved in the solution.

mixture of oxygen compounds exists on the surface, Figure 9 shows results at smaller mean electrolyte
so that a mixed potential arises, as already stated by velocity w ˆ 0:16 m sÿ1 †, and for a current density
Hoare [29]. i ˆ 3250 A mÿ2 . Here the gas bubbles are spread
Figure 7 shows the void fraction, e, for several over the channel. In the upper region the void frac-
current densities as a function of the cathode dis- tion as a function of the cathode distance has a
tance. The measurements were made with a 1M po- minimum in the centre between cathode and dia-
tassium hydroxide solution at a mean velocity of phragm. This is caused by back¯ow of the electrolyte
0:69 m sÿ1 and a temperature of 50  C. The ®rst three close to the diaphragm, resulting in higher void
electrode pairs below the platinum electrodes were fractions and electrical resistances. In a commercial
operated at current densities noted in the Figure. The electrolyser back¯ow must be avoided. It seemed
void fraction decreases with increasing cathode dis- worthwhile therefore to study the conditions for the
tance and vanishes at the diaphragm surface onset of back¯ow as a function of current density,
x ˆ 8 mm†. Increasing current density results in a number of electrode pairs and mean velocity.
higher void fraction at the cathode surface. Figure 10 shows the results. The current I is obtained
Figure 8(a) shows the in¯uence of the number of from the product of the current density, the number
electrode pairs on the void fraction as a function of of electrodes and the area of a single electrode.
distance. The void fraction increases with the number At working conditions right of the line back¯ow is
of electrodes. This e€ect is more pronounced at small avoided. A current I (A) at given mean velocity
cathode distances. Figure 8(b) shows a contour plot w m sÿ1 † smaller than
of these results. Along the boundary between single- I ˆ ÿ3:79 ‡ 128 w 5†
phase and two-phase ¯ow, the void fraction is 1%.
The white area represents single-phase ¯ow. The should be chosen. This equation is a result of linear
thickness of the two-phase boundary increases with regression of the measured values at mean velocities
the number of electrode pairs. above 0:03 m sÿ1 .
The concentration of dissolved hydrogen was de-
termined as follows. The voltage UPt;AgCl;sat between a
platinum electrode and reference electrode in potas-
sium hydroxide saturated with hydrogen at 50  C was
measured, followed by the voltage UPt;AgCl at the
unknown concentration. This voltage di€erence
results in
Rm T cH 2
UPt;AgCl ÿ UPt;AgCl;sat ˆ ÿ ln 6†
2F cH2 ;sat
In the case of a saturated solution the voltage dif-
ference disappears. Increasing voltage di€erences
correspond to a decreasing hydrogen concentration.
Figure 11 shows some results as a function of cathode
distance. The distance has no signi®cant in¯uence on
the voltages. The dissolved gases are, in contrast to
Fig. 7. Void fraction as a function of distance. Temperature 50  C,
mean velocity 0:69 m sÿ1 . Current density: (s) 6250, (h) 3250, the void fraction, homogeneously distributed over the
(n) 1500 and (,) 500 A mÿ2 . investigated cross section. The current density, how-
ROLE OF MASS TRANSFER ON HYDROGEN EVOLUTION 15

Fig. 8. Void fraction as a function of distance and number of electrodes. Current density, i ˆ 1500 A mÿ2 ; mean velocity, w ˆ 0:69 m sÿ1 ;
temperature, # ˆ 50  C.

ever, is of great in¯uence on the concentration. At electrode pairs. The voltage di€erence (hydrogen
100 A mÿ2 and 500 A mÿ2 oxygen can be detected. concentration) becomes smaller with increasing
Only at the two highest current densities of number of electrode pairs. It is noteworthy that the
3250 A mÿ2 and 6250 A mÿ2 can dissolved hydrogen voltage di€erence does not disappear. The smallest
be measured. value is 30 mV, corresponding to a maximum con-
Figure 12 shows the average values of the voltage centration of c=csat ˆ 12%, see Equation 6.
over a cross section as a function of number of Such a low value is not surprising, because the
platinum electrodes were then completely surrounded
by hydrogen bubbles, so that saturated solution can
be assumed. In this case each platinum electrode acts
as a hydrogen electrode. Its potential is shifted posi-
tively in the presence of dissolved oxygen. The plat-
inum electrode can be used as a detector for oxygen
impurities. Lorch [30] investigated the in¯uence of
oxygen on the e.m.f. of a hydrogen electrode.
Assuming electrolyte saturated with hydrogen an
amount of oxygen corresponding to 0:058 cH2 ; sat is
determined from the results of Lorch.

5. Summary

The concentration of hydrogen dissolved in an aque-

ous KOH solution can be determined using a platinum
electrode and a reference electrode. The voltage be-
tween the two electrodes depends on the concentra-
tion according to Equation 2, valid in the absence of
oxygen. This condition is justi®ed in the presence of a
catalyst for the reaction 2H2 ‡ O2 ! 2H2 O, and
when voltages are below ÿ0:7 V. When a solution is
saturated with hydrogen the e.m.f. of platinum elec-
trode can be used to determine the concentration of
traces of oxygen. The gas voidage in a ¯ow channel
Fig. 9. Void fraction as a function of distance and number of
electrodes. Current density, i ˆ 3250 A mÿ2 ; mean velocity, was determined from the electrical resistance between
w ˆ 0:61 m sÿ1 ; temperature, # ˆ 50  C. neighboured platinum electrodes. At small current
16 H. RIEGEL, J. MITROVIC AND K. STEPHAN

Fig. 10. Current, I, and velocity, w at onset of back¯ow. Current density: (s) 500, (h) 1500, (n) 3250 and (,) 6250 A mÿ2 .

Fig. 11. Voltage UPt;AgCl ÿ UPt;AgCl;sat as a function of cathode distance. The ®rst three electrodes below the platinum wires were operated
at 50  C and w ˆ 0:69 m sÿ1 . Current density: (s) 100, (h) 500, (n) 1500 and (,) 3250 and (e) 6250 A mÿ2 .

Fig. 12. Mean voltage, averaged across the channel width as a function of the number of operated electrodes at noted current densities at
mean velocity w ˆ 0:69 m sÿ1 and temperature # ˆ 50  C. Current density: (s) 100, (h) 500, (n) 1500, (,) 3250 and (e) 6250 A mÿ2 .
ROLE OF MASS TRANSFER ON HYDROGEN EVOLUTION 17

densities a two phase ¯ow was observed with in- [13] L. J. J. Janssen and E. Barendrecht, ibid. 28 (1983) 341±6.
creasing thickness of the boundary layer along the [14] L. J. J. Janssen, C. W. M. P. Sillen, E. Barendrecht and S. J.
D. van Stralen, ibid. 29 (1984) 633±42.
¯ow path. High current densities produce a back¯ow [15] L. J. J. Janssen and G. J. Visser, J. Appl. Electrochem. 21
close to the diaphragm, the void fraction and the (1991) 386±94.
electrical resistance then increase considerably. A re- [16] Idem, ibid. 21 (1991) 753±9.
[17] B. E. Bongenaar-Schlenter, E. Barendrecht, L. J. J. Janssen
lationship for the onset of back¯ow is given, with and S. J. D. van Stralen, DECHEMA-Monographien,
current as a function of electrolyte velocity. Vol. 9 Verlag Chemie, Weinheim (1985), pp. 445±61.
[18] B. E. Bongenaar-Schlenter, L. J. J. Janssen, S. J. D. van
Stralen and E. Barendrecht, J. Appl. Electrochem. 15
References (1985) 537±48.
[19] B. E. Bongenaar-Schlenter, L. J. M. Konings, C. J. Smeyers,
J. H. G. Verbunt, E. Barendrecht, L. J. J. Janssen, W.
[1] L. MuÈller, M. Krenz and K. RuÈbner, Electrochim. Acta 34 M. Sluyter and S. J. D. van Stralen, Eur. Com. EUR
(1989) 305±8. 8651 (Hydrogen Energy Carrier) (1983) pp. 206±18.
[2] N. Ibl, Chem. Ing. Technik 43 (1971) 202±15. [20] H. Vogt, Electrochim. Acta 26 (1981) 1311±17.
[3] F. N. Ngoya and J. Thonstad, Electrochim. Acta 30 (1985) [21] Idem, J. Appl. Electrochem. 17 (1987) 419±26.
1659±64. [22] J. M. Bisang, ibid. 21 (1991) 760±6.
[4] H. F. M. Gijsbers and L. J. J. Janssen, J. Appl. Electro- [23] D. J. G. Ives and G. J. Janz, `Reference Electrodes Theory
chem., 19 (1989) 637±48. and Practice', Academic Press, New York (1961).
[5] C. W. Tobias, J. Electrochem. Soc. 106 (1959) 833±8. [24] H. H. Landolt and R. BoÈrnstein, LoÈsungsgleichgewichte (6.
[6] F. Hine, M. Yasuda, Y. Ogata and K. Hara, ibid. 131 (1984) Au¯.) Bd. II 2. Teil, Springer-Verlag, Berlin (1962).
83±9. [25] H. Vogt, J. Appl. Electrochem. 23 (1993) 1323±25.
[7] F. Hine and K. Murakami, ibid. 128 (1981) 64±8. [26] G. Kreysa and M. Kuhn, J. Appl. Electrochem. 15 (1985)
[8] Idem, ibid. 127 (1980) 292±7. 517±26.
[9] F. Hine, M. Yasuda, R. Nakamura and T. Noda, ibid 122 [27] D. A. G. Bruggemann, Analen der Physik 24 (1935) 636±79.
(1975) 1185±90. [28] J. C. Maxwell, `A Treatise on Electricity and Magnetism'.
[10] C. W. M. P. Sillen and S. J. D. van Stralen, Altern. Energy Vol. 1 (2nd edn), Clarendon Press, Oxford (1881), p. 435.
Sources 4 (1980) 357±69. [29] J. P. Hoare, `The Electrochemistry of Oxygen', Interscience,
[11] Idem, ibid. 4 (1980) 119±32. New York (1968), p. 13±46.
[12] L. J. J. Janssen, J. J. M. Geeraets, E. Barendrecht and S. D. [30] A. E. Lorch, Trans. Electrochem. Soc. 60 (1936) 401±8.
J. van Stralen, Electrochim. Acta 27 (1982) 1207±18.