Communications

Information about detailed particle size distribution is not References

provided for technical measurement reasons, or is nearly
unusable because of great measurement inaccuracy. Present-
ly demands for statements about particle number-applied
measurement results in case of ultra-fine particles (x < 0.1
lm) are not considered. It is necessary to provide support by
further examinations, fostering the choice of suitable mea-
surement methods to characterize the dustiness of powders.
Although there is a multitude of methods to characterize
the dustiness of powders also quantitatively [6], in most
cases no correlation of the data obtained succeeds either
with practical powder stressing and its resulting dust release,
or with the expected stress and strain for humans and natur-
al environment. Currently ongoing developments in the divi-
sion are subject to supplement existing apparatus with a
scanning mobility particle sizer (SMPS) to investigate the in-
fluence of ultra-fine particles on the risk of human health
and environmental safety.
Received: April 5, 2004 [K 3210]
[1] H. E. Wichmann et al., Gesundheitliche Wirkungen von Feinstaub,
ecomed-Verlag, Landsberg 2002.
[2] DIN EN 481, Festlegung der Teilchengrenverteilung zur Messung
luftgetragener Partikel, Beuth Verlag, Berlin 1993.
[3] DIN 55992, Bestimmung einer Mazahl fr die Staubentwicklung von
Pigmenten und Fllstoffen, Beuth Verlag, Berlin, Teil 1: Rotationsver-
fahren 1996, Teil 2: Fallmethode 1999.
[4] E DIN 33897, Arbeitsplatzatmosphre Bestimmung des Staubungsver-
haltens von Schttgtern, Teil 1: Grundstze, Teil 2: Kontinuierlicher Fall
im Gegenstrom, Teil 3: Verstaubung in ruhender Luft (Vorlage), Beuth
Verlag, Berlin 2002.
[5] Dustiness Estimation Methods for Dry Materials, Technical Guide No. 4,
British Occupational Hygiene Society, Derby 1985.
[6] F. Hamelmann, E. Schmidt, Chem. Ing. Tech. 2002, 74 (12), 1666.
This paper was also published in German in Chem. Ing. Tech. 2003, 75 (6), 696.
______________________

Symbols used Explosion Limits of Hydrogen/Oxygen

Mixtures at Initial Pressures up to 200 bar
b width
di inner diameter By Volkmar Schrder*, Bernd Emonts, Holger Janen, and
Ext extinction Heinz-Peter Schulze
h height
I actual light intensity In the Research Center Jlich, Germany, a prototype of a
I0 initial light intensity high pressure water electrolyzer was constructed. For safe
l length running the explosion limits of hydrogen/oxygen mixtures
m mass were determined experimentally at 20 C and 80 C and at
SA dust criterion (for respirable dust) initial pressures from 1 bar up to 200 bar. The measurements
SE dust criterion (for inhalable dust) were carried out according to a modified bomb method fol-
SF dust criterion (according to the single drop tester) lowing the draft of the new European standard for the deter-
SR dust criterion (according to the rotating drum) mination of the explosion limits (prEN 1839). The anomaly
t time in the pressure dependence of hydrogen explosion limits
x particle size (decrease of the upper explosion limit with increase in pres-
x50,3 mass median value of a particle size distribution sure), known for hydrogen/air mixtures, could be confirmed
x50,0 number median value of a particle size distribution for hydrogen/oxygen mixtures. With the exceeding of the
lower explosion limit (LEL) the explosion pressures and
pressure rising rates increase only weakly. This property of
Explanations hydrogen is in opposite to the behavior of other flammable
gases, e.g., hydrocarbons. On the other hand, close to the
respirable fraction (A-Dust) is the mass fraction of inhaled upper explosion limit, high explosion pressures could be ob-
particles penetrating into the served at a slight exceeding of the limit. Furthermore it was
non-ciliate airways shown that the impurities in the product gases of the electro-
inhalable fraction (E-Dust) is the mass fraction of all lyzer did not exceed 50 % of the LEL in any case.
suspended material inhaled
through mouth and nose
PM 10 Particulate Matter; particle
fraction with an aerodynamic
diameter less than 10 lm [*] Dr. rer. nat. V. Schrder (author to whom correspondence should be
addressed, volkmar.schroeder@bam.de), H.-P. Schulze, Federal Institute
thoracic fraction (T-dust) is the mass of inhaled of Materials Research and Testing (BAM), Dep. Chemical Safety
particles penetrating beyond Engineering, Unter den Eichen 87, D-12205 Berlin, Germany; Dr.-Ing.
the larynx B. Emonts, Dr.-Ing. H. Janen, Research Center Jlich, Institute for
Materials and Processes in Energy Systems (IWV), D-52425 Jlich,
Germany.

Chem. Eng. Technol. 2004, 27, No. 8 DOI: 10.1002/ceat.200403174  2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 847
 Communications
1 Introduction
Hydrogen as alternative source of energy is gaining more
and more significance. During combustion only environmen-
tal-neutral water steam arises under energy release. In oppo-
site to the combustion of fossil fuels carbon dioxide will not
be emitted. Hydrogen can be easy produced by means of
electrical energy by electrolytic water decomposition. There-
fore the combination of water electrolysis and fuel cell is a
promising possibility to store electrical energy. Due to the
low energy density of hydrogen high pressures are necessary
at storage and transport. In order to avoid energy losses at
the mechanical compression of hydrogen high pressure elec-
trolysis is carried out. At present there are prototypes of
electrolyzers where product gases are even produced at pres-
sures up to 120 bar. A further increase of the pressure to 200
bar is planned [1]. An essential problem is the gas diffusion
in the diaphragm of the electrolysis cells. This results in a
contamination of the product gases hydrogen and oxygen.
The contamination can create malfunction and in an ex-
treme case explosions in the apparatuses and gas cylinders.
In order to avoid such dangerous conditions the knowl-
edge of the explosion limits is necessary. The database
CHEMSAFE [2], published by DECHEMA, BAM and
PTB, contains a large number of evaluated safety related
properties, among others also explosion limits, which, how-
ever, are measured mostly according to standard procedures
under atmospheric conditions. Aim of this study was the ex-
act experimental determination of the pressure dependence
of the explosion limits at 20 C and 80 C. Furthermore ex-
plosion pressures and also pressure rises should be measured
for hydrogen fractions close to the explosion limits inside
the explosion range. Such data are important for construc-
tive explosion protection measures.
2 Experimental
2.1 Apparatus and Ignition Source
The ignition tests were carried out following the bomb
method according to prEN 1839. The used apparatus is sche-
matically presented in Fig. 1.
As ignition vessels two cylindrical stainless steel vessels
(pressure resistance: 600 bar) with a volume of 2.8 dm3 and
6.0 dm3 were used. For the initial pressures from 1 bar up to
20 bar the 6.0-dm3-vessel, for initial pressures from 20 bar up
to 200 bar the 2.8-dm3-vessel was used. Both vessels are of
similar construction. The high over diameter ratio was ap-
proximately 1. The ignition source was located in the middle
of the ignition vessels. At the top of the vessels a piezoresis-
tive pressure transducer was installed.
The mixture vessel was equipped with a stirrer for the pro-
duction of homogeneous gas mixtures. The mixing and igni-
tion vessels were installed in a protection room with a pres-
sure venting wall, which was suitable for explosion tests. The
848  2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
waste gas
O2 analysis
PI TIR PIR
waste gas
PI
hydrogen
oxygen
TIC
mixing autoclave
M
ignition autoclave
ERY
Figure 1. Experimental setup for the determination of explosion limits at ele-
vated initial pressures.
high pressure valves at the vessels were remote controlled
from the controlling room. Filling and explosion pressures
were measured by means of piezoresistive pressure trans-
ducers (KELLER Inc., type PAA-10, precision: 0.5 %),
which had been calibrated before with high precision man-
ometers of WALLACE & TIERNAN, class 0.066. The pres-
sure-time courses after the ignition were recorded by means
of an A/D- transformer card (Keithley DAS1402, 12 bit, lim-
it frequency: 100 kHz) and saved in the computer system.
The ignition source was a melting wire (exploding wire)
according to prEN 1839. A small wire was soldered with two
electrodes in a distance of 5 mm. By means of a special igni-
tion device the positive AC half wave of the secondary wind-
ing of an insulating transformer (1.5 kVA) has been switched
to the electrodes for the duration of 3 ms. The wire is melt-
ing and an electric arc ignites. The delivered ignition energy
thereby is about 10 J to 20 J.
2.2 Mixture Preparation and Test Procedure
The gas mixtures were prepared according to the partial
pressures of each component. It was possible to fill the com-
ponents at lower pre-pressures and then compress them for
the ignition tests using a compressor (SCHMIDT, Kranz &
Co. Inc.; type: Maximator). The lean component was filled
each time at first. To check the composition each mixture
was analyzed with a paramagnetic oxygen analyzer (SER-
VOMEX Inc.; type 572, resolution 0.1 mol-% O2) and in ad-
dition was examined in some cases by gas chromatography.
Afterwards the test mixtures were ignited inside the igni-
tion vessel. The pressure rise caused by the reaction was
measured to characterize the explosion ability of the test
mixture. Criterion for an ignition was a pressure rising factor
http://www.cet-journal.de Chem. Eng. Technol. 2004, 27, No. 8
 Communications

(explosion pressure (pex)/initial pressure (pa)) of F 1.05. Table 1. Explosion limits of H2/O2-gas mixtures (incl. real gas correction) [4, 5].
The portion of hydrogen in the test mixture was varied step initial pressure initial temperature LEL UEL
by step until the lower (LEL) respectively the upper (UEL)
explosion limit was reached. The step distance was [bar] [C] [mol-% H2] [mol-% H2]
0.2 mol-% hydrogen. At the explosion limit four repetition 1.0 20 4.0 95.2
tests were made where no ignition is allowed. The explosion 5.0 20 4.6 94.6
limit is defined as the hydrogen concentration where no igni- 10.0 20 5.0 94.2
tion takes place any more. 20.0 20 5.4 94.2
50.0 20 5.5 94.6
100.0 20 5.7 94.9
2.3 Measurement Uncertainty 150.0 20 5.7 95.1
200.0 20 5.9 95.1
In the draft of the European standard the maximum per-
missible measurement accuracy are indicated for the prepa-
1.0 80 3.8 95.4
ration of test mixtures. It refers to the amount of the test
5.0 80 4.4 95.0
substance (normally the fuel gas) in the test mixture. For
10.0 80 4.8 94.6
fractions x 2 mol-% a maximum error of 10 % (relative)
20.0 80 5.2 94.6
is required. At amounts of x > 2 mol-% the error of the mix-
50.0 80 5.3 95.0
ture preparation is limited to 0.2 mol-% (absolutely) of the
100.0 80 5.7 95.3
test substance. As the lower and upper explosion limits of
hydrogen are more than 2 mol-% for these examinations a 150.0 80 5.3 95.5

maximum permissible measurement uncertainty of 200.0 80 5.7 95.5

0.2 mol-% for all test series is required.

The test mixtures were produced according to the partial
pressure method at different filling pressures. Due to the 200

non-ideal behavior of the components at higher filling pres-

LEL 20 C UEL 20 C
LEL 80 C UEL 80 C
sures it was partly necessary to make real gas corrections
according to DIN ISO 6146. The real gas coefficients were 150

calculated with Soave's [3] method. The corrected mixture

initial pressure in bar

compositions were checked by the GC and oxygen analysis.

100
The maximum permissible errors were not passed in any
case.
50

3 Results and Discussion

0
3.1 Explosion Limits of Hydrogen/Oxygen Mixtures 3 4 5 6
fraction H2 in mol-%
94 95 96 97

Figure 2. Pressure dependence of the explosion limits of H2/O2-gas mixtures.

The experimental tests were executed at BAM by a stu-
dent's work [4] and a research project [5]. Tab. 1 and Fig. 2
summarize the results of the tests. The explosion limits were
rounded to 0.1 mol-% H2. 150

With rising initial pressure the lower explosion limits in-

crease and the upper explosion limits decrease, which means
the explosion range is reduced with increasing pressure. This
behavior of the hydrogen is different to the behavior of
initial pressure in bar

100

other flammable gases, e.g., hydrocarbons.

In 1995 Conrad and Kaulbars [6] executed measurements
up to 150 bar with air as oxidizing component (see Fig. 3).
50
The lower explosion limits of the hydrogen air mixtures and
hydrogen oxygen mixtures are nearly identical. The cause is
a surplus of oxygen at the lower explosion limit in both
cases. The nitrogen of the air has a similar heat capacity like 0
3 4 5 6 70 72 74 76 78 80
oxygen. That means for combustion processes in the fuel fraction H2 in mol-%
poor range that the oxygen/nitrogen mixture air has simi- Figure 3. Pressure dependence of explosion limits of H2/air-mixtures at room
lar combustion behaviors like pure oxygen. temperature [6].

Chem. Eng. Technol. 2004, 27, No. 8 http://www.cet-journal.de  2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 849
 Communications

At the upper explosion limit however oxygen is the poor Table 3. Explosion indices for H2/O2-gas mixtures at 20 C and an initial
pressure of 5 bar (6-dm3-vessel) [4].
component. Due to the nitrogen quantity in the air the ex-
plosion limits in pure oxygen are clearly higher. The course Test No. amount of H2 F (dp/dt)ex KG
of the upper explosion limit has additionally an anomaly, [mol-%] (pex/pa) [bar s1] [bar m s1]

which shall not be further discussed. In the beginning the UEG = 4.6 mol-% H2
upper explosion limit decreases up to 20 bar and then at H000052a 10.0 3.53 51.2 9.3
further increasing the initial pressure the explosion limit H900052a 9.0 2.69 22.5 4.1
rises again. This anomaly of the hydrogen appears with air H800052a 8.0 2.51 18.2 3.3
as well as with oxygen. H700052a 7.0 1.76 9.5 1.7
A comparison of the amount of oxygen at the upper ex- H600052a 6.0 1.40 4.3 0.8
plosion limits of H2/air-mixtures and H2/oxygen-mixtures re- H550052a 5.5 1.34 2.7 0.5
sults in similar values (see Tab. 2). The amount of oxygen in H500052a 5.0 1.11 0.9 0.2
air was considered with 21.0 mol-%. Nevertheless at initial H480052a 4.8 1.05 0.6 0.1
pressures from 100 bar significant differences appear. It
H460052a 4.6 1.01 0.4 0.1
seems that nitrogen influences the combustion reaction only
at high pressures.
OEG = 94.6 mol-% H2
O650052a 93.5 4.92 567.6 103.1
Table 2. Comparison of the upper explosion limits of H2 in mixture with air [6] O600052a 94.0 4.41 165.6 30.1
and oxygen [4, 5].
O580052a 94.2 3.85 45.1 8.2
initial pressure UEL in air UEL in O2 xO2(air)1) xO2(O2)1) O560052c 94.4 3.34 18.6 3.4
[bar] [mol-% H2] [mol-% H2] [mol-%] [mol-%]
O540052a 94.6 1.01 0.3 0.1
1 76.5 95.2 4.9 4.8

5 73.1 94.6 5.6 5.4

Table 4. Explosion indices for H2/O2-gas mixtures at 20 C and an initial
10 72.1 94.2 5.9 5.8 pressure of 50 bar (2.8-dm3-vessel) [4].
20 71.1 94.2 6.1 5.8
Test No. amount of H2 F (dp/dt)ex KG
50 73.8 94.6 5.5 5.4 [mol-%] (pex/pa) [bar s1] [bar m s1]

100 73.4 94.9 5.6 5.1 UEG = 5.5 mol-% H2

H000502a 9.9 3.16 133.2 18.8
150 72.9 95.1 5.7 4.9
H900502a 8.9 2.84 205.5 29.0
1)
xO2 = molecular fraction of oxygen at the upper explosion limit H800502a 7.9 1.93 75.1 10.6
H700502a 6.9 1.33 30.8 4.3
H620502a 6.1 1.08 8.5 1.2

3.2 Explosion Pressures and Pressure Rises close to the H600502a 5.9 1.05 5.9 0.8

Explosion Limits H580502c 5.7 1.06 5.8 0.8

H560502a 5.5 1.01 4.2 0.6
Besides the explosion limits (concentration limits of H2)
the question concerning the consequences of an explosion OEG = 94.6 mol-% H2
inside the electrolyzer at slightly exceeding limits was of in- O600502a 94.0 4.43 690.8 97.4
terest. For some pressures mixtures within the explosion O580502a 94.2 4.18 530.5 74.8
range were prepared and the explosion pressures (pex) as O560502a 94.4 4.16 467.2 65.9
well as the highest pressure rise (dp/dt)ex after the ignition O540502a 94.6 1.01 3.5 0.5
were measured. (dp/dt)ex depends strongly on the vessel vol-
ume. For a vessel volume of 1 m3 (dp/dt)ex is equal to the
KG-value according to the cubic law (see Eq. (1)). This value gen has a very high diffusion coefficient. Tab. 3 and Tab. 4
is a largely volume independent explosion indix, which de- clearly show that with the exceeding of the lower explosion
scribes the violence of an explosion. limit at some mol-% H2 the pressure rise factors and KG-val-
ues will increase only slightly. Explosion reactions are slow
KG = (dp/dt)ex V1/3 (1) deflagrations. This behavior of hydrogen is in opposite to the
behavior of most other fuel gases. Because of this fact it is
In Tab. 3 and 4 explosion indices for initial pressures of possible to apply pressure relief equipment (rupture disks,
5 bar and 50 bar are listed. safety valves) at this mixture compositions.
At the technical operation of water electrolyzers in partic- In the area of the upper explosion limit fast reactions and
ular hydrogen contamination in oxygen (in the range of the high explosion pressures were observed with minimal ex-
lower explosion limit) has to be considered, because hydro- ceeding the limit.

850  2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim http://www.cet-journal.de Chem. Eng. Technol. 2004, 27, No. 8
 Communications

In Fig. 4 two typical pressure/time-courses are shown for 2.5

reactions at the lower and upper explosion limit. The explo- high pressure water electrolyzer
prototype FZJ
sion limits were exceeded at about 1 mol-%. 2 stack temperature (average): 60 C

contamination in mol-%
0.1 A cm-2
30 0.2 A cm-2
1.5 0.4 A cm-2
Test No. O650052a; 93,5 Mol-% H2 H2 contamination in oxygen
25 pex Test No. H550052a; 5,5 Mol-% H2
1

20
pressure in bar

0.5
(dp/dt)ex
15 O2 contamination in hydrogen
UEL exeeded at about 1 mol-%

0
10 0 10 20 30 40 50 60 70 80 90 100 110 120
pex operating pressure in bar
(dp/dt)ex
5 Figure 5. Product gas contaminations in dependence of operating pressure
LEL exeeded at about 1 mol-%
and current density.

0
0 0.5 1 1.5 2 2.5
time in s
Figure 4. Pressure-time courses after the ignition of H2/O2-gas mixtures at an
initial pressure of 5 bar and 20 C
In Tab. 3 and Tab. 4 only single test results are shown.
They shall serve the presentation of the different reaction
courses at the upper and lower explosion limit and are not
suitable for the construction of safety devices. Due to the
fact that ignition tests do scatter very strongly, the explosion
indices, which shall serve for constructive protection mea-
sures, have to be fixed by means of a greater number of tests.
3.3 Conclusions for Running the Electrolyzer
Two effects during the production of hydrogen and oxygen
in an electrolyzer can be responsible for an appearance of a
contamination of the product gases with each other compo-
nent. On the one side there will be diffusion of hydrogen
through the diaphragm, which separates both gas rooms. On
the other side permanent low quantities of the liquid electro-
lyte (potash alkaline solution) were transferred in the rooms
of the electrodes. The gases H2 and O2, solved in the electro-
lyte, can degas and cause contaminations of the product gases.
This is possible because both electrolytes will be brought
together for the concentration balance behind the gas se-
parators. Measurements of the gas contamination for the
electrolysis at high pressure up to 120 bar show the signifi-
cant influence of the diffusion of the hydrogen through the
separating diaphragm [7]. Fig. 5 shows the values at an aver-
age stack temperature of 60 C for the contamination of O2
with H2 as well as H2 with O2 in dependence of operating
pressures and current densities.
For the examined operating area the contamination of the
hydrogen with oxygen does not show a significant depen-
dence of pressure respectively current density. Furthermore
the value with less than 0.25 mol-% O2 is very low. On the
other hand the contamination of the oxygen by hydrogen in-
creases with increasing pressure and falling current density.
3 3.5 4
At operating conditions with the nominal current density of
0,4 A/cm2 the gas contamination achieved 0,5 mol-% H2 in
O2 at 30 bar and 1.5 mol-% at 120 bar. Partial load of
0.1 A/cm2 results to a further increase of contamination up
to a maximum concentration of 2.0 mol-% H2 in O2.
These results give an evaluation concerning the safety dis-
tance of the measured values to the explosion limits. To the
upper explosion limit there will be a sufficient safety dis-
tance of about 3 mol-% in every working situation (< 0,25
mol-% O2). The lower explosion limit (> 4 mol-% H2) will
be achieved only for about 50 % with a maximum value of
about 2 mol-% H2 contamination. In case of a continually
H2 concentration measurement in the product gas it is possi-
ble to observe the safety distance to the lower explosion lim-
it as well as the fixing of an alarm threshold of 50 % lower
explosion limit for an automatically shut down of the elec-
trolysis process. Furthermore the determination of the ex-
plosion pressures as well as the pressure rises close to the
lower explosion limit has shown that constructive protection
measures like a pressure resistant construction in combina-
tion with safety valves offer sufficient security when slightly
exceeding the lower explosion limit.
Received: December 15, 2003 [K 3174]
References
[1] H. Janen, B. Emonts, 12. Internationales Sonnenforum, Freiburg,
Germany, July 2000.
[2] CHEMSAFE Database of Evaluated Safety Characteristics, DE-
CHEMA, BAM and PTB, Frankfurt/M., Germany.
[3] B. E. Poling, J. M. Prausnitz, J. P. O'Connel, The Properties of Gases and
Liquids, 5th ed., McGraw-Hill, New York 2001.
[4] R. Daubitz, Bestimmung der Explosionsgrenzen von Wasserstoff/
Sauerstoff-Gemischen bei erhhten Anfangsdrcken, Studienarbeit,
Technische Universitt Berlin 2001.
[5] V. Schrder, Sicherheitstechnische Untersuchungen fr die Hochdruck-
Wasserelektrolyse zur Speicherung regenerativer Energie, Abschluss-
bericht zum Vorhaben 2226, BAM, Berlin 2002.
[6] D. Conrad, R. Kaulbars, Chem. Ing. Tech. 1995, 67, 185.
[7] H. Janen, B. Emonts, H.-G. Groehn, H. Mai, R. Reichel, D. Stolten,
HYPOTHESIS IV, Stralsund, September 2001.
This paper was also published in German in Chem. Ing. Tech. 2003, 75 (7), 914.

Chem. Eng. Technol. 2004, 27, No. 8 http://www.cet-journal.de  2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 851