Corrosion Science 83 (2014) 226–233

Contents lists available at ScienceDirect

Corrosion Science

journal homepage: www.elsevier .com/locate /corsc i

Hydrogen evolution rate during the corrosion of stainless steelin supercritical water

http://dx.doi.org/10.1016/j.corsci.2014.02.0190010-938X/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +1 705 748 1011x7063; fax: +1 705 748 1625.E-mail addresses: kchoudhry@trentu.ca (K.I. Choudhry), rcarvajalorti@trentu.ca

(R.A. Carvajal-Ortiz), dimitrioskallikragas@trentu.ca (D.T. Kallikragas), isvishchev@trentu.ca (I.M. Svishchev).

Kashif I. Choudhry, Ruth A. Carvajal-Ortiz, Dimitrios T. Kallikragas, Igor M. Svishchev ⇑Trent University, Department of Chemistry, 1600 West Bank Drive, Peterborough, Ontario K9J 7B8, Canada

a r t i c l e i n f o

Article history:Received 20 September 2013Accepted 8 February 2014Available online 18 February 2014

Keywords:A. Stainless steelB. Modelling studiesC. High temperature corrosion

a b s t r a c t

The interaction of water with metal surfaces at high temperatures leads to the significant release ofhydrogen gas. A systematic investigation of hydrogen evolution from fresh and oxidized stainless steel(SS316) surfaces is carried out in a tubular reactor, at supercritical water conditions. A linear relationshipis found between the reactor surface area and the rate of hydrogen gas released. Results show that theevolution of hydrogen gas is a zero-order reaction, with the activation energy of 105.9 kJ mol�1 for theoxidized surface.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Worldwide energy consumption is rapidly increasing and isaccelerated by economic development and increasing population.The expected increase in energy demand and the growing aware-ness of global warming has led to international efforts to developa new advanced generation of nuclear energy systems. These Gen-eration IV (GEN-IV) proposals include the Supercritical Water-Cooled Reactor (SCWR) as one of the six advanced nuclear reactorconcepts under the program. The SCWR is a reactor which usessupercritical water (SCW) as both moderator and coolant. It willgenerate electricity with an outlet temperature of 625 �C and pres-sure of 25 MPa, which are much greater than those of existingpressurized water reactors (PWRs) [1]. The SCWR is expected tobecome one of the principle sources for the large scale productionof hydrogen gas.

Since the SCWR operates at higher temperatures and pressuresthan existing PWRs and fossil fuel plants, the corrosion of struc-tural materials under these extreme conditions, has been identifiedas a critical problem [2]. The concentrations of dissolved O2 and H2

in the coolant, as well as pH, are all crucial parameters affecting thespeciation and solubility of the passivation layer formed by themetal oxides and hydroxides that accumulate on the inner surfacesof the SCW bearing components. The high temperatures and pres-sures make SCW very aggressive in terms of corrosion, particularlyin an oxidizing environment. The in-core generation of oxidizing

species such as hydrogen peroxide H2O2, OH radical, and oxygen,due to the radiolytic breakdown of water, can increase the corro-sion of reactor components, as well as have an effect on the trans-port and deposition of corrosion products [3]. The high dissolvedoxygen concentration is a cause of concern for both the generalcorrosion, and the stress corrosion cracking susceptibility of thestructural material in the reactor [4]. Corrosion in the system cancause increased maintenance time, cost, and a loss of power gener-ation [5–7]. In order to minimize the corrosion in the heat trans-port systems of the PWRs and Boiling Water Reactors (BWRs),hydrogen water chemistry (HWC) is being examined to includethe addition of oxygen scavenging species such as hydrogen gasand hydrazine to the coolant [1,8–10], as well as lithium hydroxide[11] for alkalinity control. Hydrazine (N2H4) is a strong reducingagent, currently used both in thermal and nuclear power plants be-cause of its ability to eliminate dissolved oxygen effectively, andprotect structural materials against corrosion. Power plants withWater-Cooled-Water-Moderated Energy Reactors (VVERs orWWERs) use hydrazine as a source of the hydrogen that is requiredto suppress the accumulation of the oxidizing species generated byradiolytic processes in the coolant water [9]. However, it is not yetclear whether these strategies will provide adequate chemistrycontrol in the SCWR environment.

Austenitic stainless steels, nickel-based alloys and zirconium al-loys are widely used in the design of nuclear power plants and areregarded as possible candidate materials for the water bearingcomponents of the proposed SCWR [12–15]. These materials willbe subject to corrosion, particularly in the harsh supercritical waterenvironment at which these next generation designs will operate[15,16]. It is well known that at high temperatures, significantamounts of hydrogen gas can be released during the oxidation of

K.I. Choudhry et al. / Corrosion Science 83 (2014) 226–233 227

a metal surface in the formation of a corrosion layer. In the case ofstainless steel, this corrosion product is typically a magnetite/hematite oxide layer on the surface of the metal. Of particular sig-nificance to the SCWR is that hydrogen is not only produced duringthe oxidation of the metal surface, but is also a primary product ofthe radiolytic breakdown of water in the in-core region of the reac-tor water loop.

There has been a long history of study in the role of hydrogenand the effect it has on the corrosion rate of a material [17–24].It is worthwhile mentioning that the presence of excess hydrogencan become a crucial factor in the failure of materials in nuclearpower systems. For instance, hydrogen can diffuse into a metaland precipitate as a hydride such as Zr2H3 in the case of zirconiumalloys [25]. Such absorption of hydrogen can lead to phenomenasuch as stress corrosion cracking, hydrogen embrittlement, and de-layed hydride cracking, which can weaken the structure and con-tribute to the catastrophic failure of the reactor components[1,14,26–28].

The reactions that take place in the corrosion layer are very sim-ilar to those that are currently being investigated for the develop-ment of catalytic water splitting cycles intended for large scalehydrogen production. There are several thermochemical watersplitting mechanisms under examination involving base metal oxi-des. For a detailed review of water splitting using various metaloxides, the reader is refereed to Souza [24].

From a viewpoint of a nuclear reactor chemistry control strat-egy, hydrogen generated in the heat transport system of the SCWRmay potentially have a vital role as an oxygen scavenger, effec-tively removing the corrosive radiolytic products from both thein-core and out-of-core regions. An understanding of how muchhydrogen will be produced in the SCWR pressure tube is importantin estimating the amounts of other oxygen scavenging compounds,such as hydrazine, which may be added to the coolant for corrosioncontrol.

In our previous work [29], the dissolved hydrogen concentra-tion in the effluent of a tubular stainless steel reactor was foundto be rather high at conditions relevant to the operation of theSCWR. In this study, we systematically investigate the effects thatthe reactor surface area, temperature, pressure, and the residencetime of deoxygenated water, have on the rate of hydrogen evolu-tion during corrosion, on both fresh and oxidized stainless steelsurfaces. Hydrogen evolution was studied in a flow-through reac-tor using SS316 tubing with volumetric at pump flow rates rang-ing from 0.1 to 1.5 mL min�1. Reactor temperatures and pressuresranged from 650 to 750 �C and 5 to 25 MPa, respectively. Oncethe hydrogen evolution rate had reached a steady state, oxygen-ated water was introduced into the flow system. An estimate wasobtained of how much hydrogen reacts with the dissolved oxy-gen. The apparent corrosion rate at SCWR flow conditions wasalso estimated, through the determination of metal loss to theeffluent. The metal concentrations in the reactor’s effluent wereanalysed using voltammetry methods, and the concentrations ofO2 and H2 were continuously monitored by a Dissolved Gas Ana-lyser (DGA).

Molecular dynamics simulations of supercritical water at theiron hydroxide interface were conducted at the temperature of640 �C for total water densities of 0.044 and 0.055 g cm�3. Thesedensities are similar to those found in the SCWR. Iron hydroxideis the primary corrosion product produced when ferric metals rustat SCW conditions. A corrosion crevice was modelled by confiningthe water between parallel iron hydroxide surfaces spaced 40 nmapart. Simulations support the experimental findings of greaterhydrogen evolution rates at lower densities. Water was found topenetrate into the surface at the lower density configuration,where once within the iron hydroxide structure, catalytic watersplitting may occur.

2. Experimental and simulation details

2.1. Apparatus

A stainless steel flow-through reactor was used to producehydrogen and to examine the hydrogen-water chemistry (HWC)at conditions relevant to the SCWR. A schematic diagram of theexperimental set-up is shown in Fig. 1. All parts of the flow-through apparatus were constructed using stainless steel (SS316)and connected using SS316 capillary tubing (OD = 1.59 mm,ID = 0.75 mm) and zero dead volume unions (Valco�). The feed linewas connected to three bottles, each containing air saturated,hydrogen saturated or deoxygenated water. This was achieved bysparging with ultrapure air, hydrogen and helium gas (Praxair highpurity gases, UHP 5.0), respectively. Low pressure check valves(anti-backflow, Valco�) were installed on the sparging lines justoutside the bottles to prevent water backflow from the bottles intothe lines. The volumetric flow rate of the water was continuouslymonitored by weighing a collected volume of water over a set timeperiod. Three separate HPLC pumps (Waters� 590) were used forcontrol of the volumetric flow rates. Pump 1 and 2 were used tocontrol the flow rate of air saturated and hydrogen saturated waterrespectively. Pump 3 was used to control the flow rate of the deox-ygenated water. All three flow channels were combined by using azero dead volume cross (Valco�) and were connected to a switch.With the switch in position 1, the water flows through the oven/reactor and when the switch is in position 2, the water flows di-rectly through to the DGA for calibration.

Experiments were carried out using SS316 tubing with differentinternal diameters resulting in varying surface-to-volume ratios,from 7.79 to 53.33 cm�1. Table 1 shows the elemental compositionof the tubing. The reactor tube was placed in a custom-made sandbath inside a programmable muffle furnace (Isotemp� 650 Series,Fisher Scientific). Furnace temperature was maintained above thewater’s critical temperature by the built-in proportional integralderivative (PID) controller equipped with a type K-thermocouple.The effluent exiting the reactor was cooled to ambient temperatureusing a custom built heat exchanger (length = 2.5 m, made ofSS316 tubing), coupled to a cold plate (TE Technology�, Inc.). Thetemperature of the heat exchanger was maintained by a tempera-ture controller (TC-24-25, TE Technology�), and the pressure in thesystem was controlled by an adjustable back pressure regulator(P-880, Upchurch). Both the temperature and pressure of the sys-tem were monitored and recorded using a data logger (PrTC-210,Omega�). After cooling, the depressurized effluent was then passedthrough the dissolved gas analyser (SRI Instruments�, 8610C)which was equipped with a helium ionization detector (HID) forthe measurement of the gas composition. The dissolved oxygenconcentration and pH of the effluent were also monitored, usinga NeoFox system (Ocean Optics�) equipped with a FOSPOR oxygenprobe and Orion 5 Star metre (Thermo Scientific) respectively.

2.2. Calibration curve for hydrogen analysis

An HPLC pump was used to maintain a steady flow rate of0.1 mL min�1 and the actual flow rate was verified by collecting avolume of water over a set time period. The pump flow rate wasadjusted until the desired amount of water was collected, corre-sponding to the target rate of 0.1 mL min�1. A calibration curvefor hydrogen gas was obtained by first feeding hydrogen saturatedwater with a known hydrogen concentration of 1.6 mg L�1 (ppm)into a dissolved gas analyser. The hydrogen gas was separated fromthe continuously flowing water using concentric tubes, with anouter gas collection tube made of glass, and an inner permeablesilicone extraction tube through which the gas was able to effuse,

Fig. 1. Schematic diagram of the experimental set-up for the determination of hydrogen evolution.

Table 1Chemical composition of the stainless steel 316 reactor tubing (in wt%).

Metals Fe Cr Ni Mn C P S Si Mo

Percentage 65.00 17.00 12.00 2.00 0.08 0.045 0.03 1.00 2.50

228 K.I. Choudhry et al. / Corrosion Science 83 (2014) 226–233

and subsequently enter the chromatography coil of the gas ana-lyser. An extraction time of 1.25 min was used to generate the firstpoint on the curve. Since the amount of dissolved hydrogen in thesample is the saturation value, this first point corresponds to ahydrogen concentration of 1.6 mg L�1. The extraction time wasthen doubled to 2.5 min. By doubling the volume of water flowingthrough the extraction tube compared to the extraction time of1.25 min, twice the amount of hydrogen, or 3.2 mg L�1, was ob-tained. This pattern of doubling the extraction time was repeatedfor 5, 10, and then 20 min extraction times, yielding concentrationsof 6.4, 12.8, and 25.6 mg L�1 respectively, giving a linear relation-ship between peak area and concentration. It was observed thatthe calibration curve would deviate from linearity at concentra-tions beyond 25.6 mg L�1 and hence this is the operating limit ofthis technique.

2.3. Experimental procedure

The hydrogen evolution rate was examined in a stainless steeltubular reactor with varying surface-to-volume ratios rangingfrom 7.79 to 53.33 cm�1. Temperatures and pressures ranged from650 to 750 �C and 5 to 25 MPa. The at-pump flow rates of deoxy-genated water were varied from 0.1 to 1.5 mL min�1, in order toachieve different water residence times in the reactor. Separatewater bottles containing air saturated water, hydrogen saturatedwater, and deoxygenated water were prepared, and were continu-ously sparged during the experiments with either air, hydrogen orhelium gas. Deoxygenated water with an oxygen concentration ofless than 10 lg L�1 (ppb) was first pumped through the system at avolumetric flow rate of 0.1 mL min�1 and pressurized to the de-sired pressure of 25 MPa by adjusting the back pressure regulator.The Isotemp muffle furnace was then turned on, and the tempera-ture was raised to 650 �C. The effluent exiting the reactor wascooled and depressurized prior to analysis, and gas compositionwas measured using a gas chromatograph (SRI Instruments�

8610C). The effluent was continuously collected and monitoredfor the concentrations of nickel, chromium and iron using adsorp-tive stripping voltammetry, via a hanging mercury drop electrodemethod, using a 797 VA Computrace system (Metrohm�). Themethods used for the trace detection of nickel, chromium and ironare reported in the VA Application Notes [30–32]. At the end ofeach experimental run, the system was first depressurized andthen cooled down to room temperature.

2.4. Simulation details

Classical molecular dynamics (MD) simulations were performedusing the M.DynaMix simulation software developed by Lyubartsevand Laaksonen [33]. The SPC/E model of water was used as it accu-rately describes thermodynamic properties particularly at hightemperatures. The equation of state obtained from the SPC/E modelcorrelates well with experimental results and through the corre-sponding sates principle, is an accurate method of investigatingwater systems at the elevated temperatures and pressures associ-ated with the supercritical phase [34,35]. Interaction parametersfor the Fe(OH)2 surfaces were represented by the relatively recentCLAYFF force field which has been successfully implemented in themodelling of the interaction energies of amorphous solids, oxides,layered hydroxides and interfacial systems [36,37].

Two parallel electrostatically neutral Fe(OH)2 slabs were con-structed, separated by a 40 nm gap. Simulations were conductedat the SPC/E model temperature of 640 �C, corresponding to a realwater temperature of 650 �C, for total water densities of 0.044 and0.055 g cm�3. Details on the formulation and structure of theFe(OH)2 surface, simulation cell size, and numbers of molecules,can be found in Svishchev et al. [38]. A random configurationwas initially used for the placement of the water molecules andthe system was allowed to equilibrate for the first 50 ps. The Ver-let algorithm was used to integrate the equations of motion using a1 fs time step. The simulations were performed using an NVTensemble and the temperature was maintained using the Nose–Hoover thermostat. Periodic boundary conditions were employedand the atomic pair interactions were calculated using the Lor-entz–Berthelot mixing rules. Long-range Coulombic interactionswere handled via the Ewald Summation method [39]. Intermediateaveraging was performed every 1000 time steps and the simula-tions were allowed to proceed for 1 ns.

Fig. 3. Total dissolved metal ion concentrations, in lg L�1, in the effluent versusexposure time. Temperature, pressure and flow rate were held constant at 650 �C,25 MPa and 0.1 mL min�1.

K.I. Choudhry et al. / Corrosion Science 83 (2014) 226–233 229

3. Experimental results and discussion

The influence of the reactor surface on the hydrogen evolutionrate was first studied using deoxygenated water with an at-pumpvolumetric flow rate of 0.1 mL min�1, at the SCWR conditions of650 �C and 25 MPa. The measured hydrogen gas concentration inthe depressurized effluent was normalized by the density of waterat 650 �C and 25 MPa to give actual dissolved hydrogen concentra-tions, in moles per litre of supercritical water, in the reactor tube.Both the reactor tube and the small capillary section within thefurnace were accounted for in the calculation of the exposed sur-face area and total volume of SCW. Fig. 2 presents the dissolvedhydrogen concentrations as a function of time for different surfaceareas. It can be seen that at the beginning of each experiment, withfresh metal surface, the hydrogen evolution in the reactor is thehighest, and decreases exponentially with time. About 1 weekwas required for the dissolved hydrogen concentrations at the exitof the reactor to reach a steady state. Hydrogen yields from the oxi-dized SS316 surface levelled off at 0.13 � 10�3, 0.21 � 10�3 and0.25 � 10�3 mol L�1, for the reactor surface areas of 42.88, 60.77and 65.97 cm2, respectively. The initial drop in hydrogen evolutioncan be presumably attributed to a decrease of bare metal availablefor the reaction with water, as the reactor surface develops anoxide layer over time, resulting in a subsequently lower hydrogenevolution rate.

The apparent corrosion rate was estimated according to themethod described by Svishchev et al. [29], by collecting samplesof the effluent water which were analysed for the concentrationsof the dissolved metal ions of Fe, Ni and Cr. These metals makeup the majority of the composition of stainless steel as shown inTable 1. Their concentrations in the effluent provide an indicationof the level of corrosion in the system. The daily metal concentra-tions in the effluent were calculated by averaging the concentra-tions found in 3 samples taken each day. In all samples, theconcentration of Cr was found to be below 0.7 lg L�1, which isthe detection limit of this method. This is because the formationof soluble chromate ion HCrO4�, was not expected with very lowconcentrations of oxygen in the water. Fig. 3 shows the daily aver-age concentration of Fe and Ni in the effluent for the experimentalrun performed at a constant reactor temperature of 650 �C andpressure of 25 MPa. The daily dissolved metal concentrations showthat it also takes about 1 week to reach a steady state in the metalrelease rate. This is likely the time required for the formation of theprotective oxide layer on the bare, fresh metal. It also matches thetime required to achieve a steady state of hydrogen evolution of0.13 � 10�3 mol L�1, shown in Fig. 2, for the reactor surface area

Fig. 2. Hydrogen evolution in the reactor versus exposure time. Temperature,pressure and flow rate were held constant at 650 �C, 25 MPa and 0.1 mL min�1.

of 42.88 cm2. Based on these observations, the first week of datawas ignored in the calculation of the apparent corrosion ratethrough loss of metals to the solution. The apparent corrosion ratewas calculated to be about 2.0 � 10�5 mm y�1. This value for theapparent corrosion rate is significantly lower than that reportedby Luo et al. [2] from autoclave data using weight loss methodsafter the descaling of corrosion coupons. This indicates that mostof the oxide produced in our experiments at conditions relevantto SCWR remained on the inner surface of the reactor tube.

The relationship between the steady state hydrogen evolution,measured by the dissolved hydrogen concentrations at the exit ofthe reactor, and the reactor surface area is shown in Fig. 4. Resultsshow a linear relationship between the hydrogen concentrations inthe effluent, and the surface area of the reactor tube. The trend ofincreasing hydrogen evolution with an increase in the reactor sur-face area is caused by an increase in the area available for the sur-face oxide catalysed reaction of water decomposition.

The effect of the flow rate, and hence residence time, of wateron the steady state hydrogen evolution rate was investigated withvolumetric flow rates ranging from 0.1 to 1.5 mL min�1 and reactortemperatures and pressures ranging from 650 to 750 �C, and 5 to25 MPa. A stainless steel reactor tube with an inner volume of0.176 cm�3 was used with different at-pump volumetric flow ratesto achieve different water residence times in the reactor. The resi-dence time s, of the supercritical water in the reactor was calcu-lated according to the flow rate and the density of the feed waterunder the reaction conditions using Eq. (1):

Fig. 4. Steady state hydrogen evolution versus the reactor surface area. Temper-ature, pressure and flow rate were held constant at 650 �C, 25 MPa and0.1 mL min�1.

Fig. 5. Effect of the flow rate on steady state hydrogen evolution in the reactor.Temperature and pressure ranged from 650 to 750 �C and 5 to 25 MPa.

Fig. 6. Effect of the residence time on steady state hydrogen evolution at a constanttemperature of 650 �C. Pressure was varied from 5 to 25 MPa.

Fig. 7. Effect of the residence time on steady state hydrogen evolution at a constant

230 K.I. Choudhry et al. / Corrosion Science 83 (2014) 226–233

sðT; p; f Þ ¼ Vf� qðT; pÞqðT�; p�Þ ð1Þ

where V is the volume of the reactor, f is the volumetric flow rate atthe pump, q(T, p) is the density of water at the operating tempera-ture and pressure, shown in Table 2, and q(T⁄, p⁄) is the density ofthe water at ambient conditions. The results for hydrogen evolutionlevels at different volumetric flow rates are shown in Fig. 5. At a lowvolumetric flow rate of 0.1 mL min�1, higher hydrogen evolutionwas observed. It can be seen that hydrogen evolution generally de-creases exponentially with an increase of volumetric flow rate. Atreactor conditions of 650 �C and 25 MPa, the hydrogen concentra-tion decreased to about one sixth of the initial value when the vol-umetric flow rate was increased from 0.1 to 0.5 mL min�1. At avolumetric flow rate of 1.5 mL min�1, the hydrogen concentrationwas reduced to one half of the value obtained at a flow rate of1.0 mL min�1. This decrease can be explained by a lower contacttime between water and the reactor surface. In these experimentsthe residence time and hence the contact time varied from 0.1 to6.87 s. The residence time of approximately 0.5 s, corresponds tothe residence time of the working fluid in the SCWR pressure tube.The linear relationship between increasing residence time and stea-dy state hydrogen evolution is explicitly shown in Figs. 6 and 7. Inour previous work, the hydrodynamic behaviour of a similar flow-through reactor was examined from ambient to supercritical waterconditions. Our residence time distribution (RTD) results indicatedthat this reactor system exhibits plug flow behaviour over widetemperature and flow rate ranges [40]. Our current flow reactor,which operates at similar conditions, showed a linear relationshipbetween the hydrogen evolution rate and the residence time. Thisindicates that hydrogen evolution is a zero-order reaction with re-spect to water as a reactant.

Fig. 4 shows a linear relationship between the hydrogen levelsin the reactor and the surface area of the reactor tubing. This indi-cates that the reaction is surface catalysed, and occurs at the sur-face of the stainless steel reactor tube. Thus the surface areaexposed to the coolant water is the limiting factor in hydrogen evo-lution at elevated temperatures and pressures.

The effective rate constant for hydrogen evolution was deter-mined using the experimental data for water residence times rang-ing from 0.1 to 6.87 s, and reactor temperatures and pressuresranging from 650 to 750 �C and from 5 to 25 MPa. The effective rateconstant keff, in mol cm�2 s�1, was calculated using Eq. (2):

keff ¼10�3CM

s SV

� �" #

ð2Þ

Here, CM is the molarity, or mol L�1, of the dissolved hydrogen inthe effluent at the exit of the reactor, S/V is the surface area-to-vol-ume ratio of the reactor tube in cm�1, and s is the residence time inseconds. Slopes of the graphs in Figs. 6 and 7 were divided by thereactor surface area-to-volume ratio to obtain effective rate con-stants for different temperature and pressure conditions. A valueof 1.58 � 10�10 mol cm�2 s�1 was obtained as an effective rate con-stant for hydrogen evolution at a constant reactor temperature of650 �C and pressure of 25 MPa.

Table 2Density of water in the test unit at the temperatures and pressure examined.

Temperature (�C) Pressure (MPa) Density (g cm�3)

650 5 0.012650 15 0.037650 25 0.065700 25 0.060750 25 0.056

pressure of 25 MPa. Temperature was varied from 650 to 750 �C.

The effect of pressure on steady state hydrogen evolution wasstudied in the range of 5–25 MPa at a constant reactor temperatureof 650 �C. It can be seen in Fig. 6 that as the pressure of the systemdecreased from 25 to 5 MPa, the hydrogen concentration at the exitof reactor increased somewhat, yielding effective rate constants of1.66 � 10�10 mol cm�2 s�1 at 15 MPa and 1.83 � 10�10 mol cm�2

s�1 at 5 MPa. The average effective rate constant for pressures

Fig. 9. Hydrogen evolution in the reactor versus exposure time for the first 72 h ofoperation. Temperature, pressure and flow rate were held constant at 650 �C,25 MPa and 0.1 mL min�1. The dotted line is an extrapolation to t = 0.

K.I. Choudhry et al. / Corrosion Science 83 (2014) 226–233 231

ranging from 5 to 25 MPa, at temperature of 650 �C, is estimated tobe 1.69 ± 0.13 � 10�10 mol cm�2 s�1. Therefore, the dissolvedhydrogen concentration at the exit or at any point along the stain-less steel tubular reactor can be estimated by multiplying theeffective rate constant with the surface area-to-volume ratio ofthe reactor tube and the residence time of the fluid in the reactor,or in the section of interest.

The influence of temperature on hydrogen evolution was inves-tigated by varying the reactor temperature from 650 to 750 �C at aconstant pressure of 25 MPa. Fig. 7 indicates that the experimentalresults show a linear relationship between the reactor temperatureand the steady state hydrogen concentration in the effluent at theexit of the reactor. Fig. 5 shows that at a flow rate of 0.1 mL min�1

and a pressure of 25 MPa, steady state hydrogen levels were5.43 � 10�5, 10.44 � 10�5 and 18.7 � 10�5 mol L�1 for the temper-atures 650, 700 and 750 �C, respectively. These experimental re-sults clearly demonstrate that the increase in the reactortemperature can significantly enhance the amount of hydrogenproduced. The increase in hydrogen evolution with increasing tem-perature was expected since the reaction rate increases with tem-perature, according to the Arrhenius equation. The effective rateconstants of 1.58 � 10�10, 3.25 � 10�10 and 6.08 � 10�10

mol cm�2 s�1 were obtained for steady state hydrogen evolutionat a constant pressure of 25 MPa and temperatures of 650, 700and 750 �C. We have produced an Arrhenius plot for the effectiverate constants, as shown in Fig. 8, to determine the activation en-ergy of 105.9 kJ mol�1 for the oxidized surface. The value for theactivation energy obtained in this study is comparable to the re-ported value of 110 kJ mol�1 for hydrogen production using mixediron oxides coated over ceramic substrates [41].

The effect of reactor temperature on the amount of hydrogeninitially evolved from the bare metal surface was also examined,by varying the temperature from 650 to 750 �C at a constant reac-tor pressure of 25 MPa. Fig. 9 shows a plot of the natural logarithmof the hydrogen concentration versus time for the first 72 h, yield-ing a straight line. This allowed us to estimate the initial hydrogenevolution rate from the bare, fresh metal surface by extrapolatingthe concentration to t = 0, as shown by the dotted lines in Fig. 9.The initial hydrogen concentrations evolved from the bare, freshmetal surface, as measured at the exit of the reactor were foundto be 0.5 � 10�3, 0.7 � 10�3 and 0.9 � 10�3 mol L�1, for the reactortemperatures of 650, 700 and 750 �C, respectively. The effectiverate constants of 3.01 � 10�10, 4.54 � 10�10 and 6.24 � 10�10

mol cm�2 s�1 were obtained for the temperatures of 650, 700 and750 �C, respectively. The activation energy for this initial evolutionperiod was obtained from the Arrhenius plot in Fig. 10, and found

Fig. 8. Arrhenius plot of the effective rate constants for the steady state hydrogenevolution. The calculated activation energy is 105.9 kJ mol�1.

to be 57.42 kJ mol�1. This initial activation energy is much lowerthan the activation energy for the steady state, which was foundto be 105.9 kJ mol�1.

In our previous paper we have shown the elemental composi-tion analysis of the corrosion layer using scanning electron micros-copy (SEM) coupled to energy dispersive X-ray diffraction (EDX)obtained from line scans across the metallographic cross-sectionof the exposed reactor tube surface at 500 �C and 25 MPa [29].The results show a double oxide layer of about 5 lm consistingof an inner Cr enriched layer and outer Fe enriched layer. The for-mation of the oxide layer, the metal dissolution and the hydrogenevolution are all closely related. The high concentration of hydro-gen and iron in the effluent during the first 24 h is attributed tothe high solubility of the iron hydroxide, Fe(OH)2, and hydrogengas, both formed from the reaction of the bare stainless steel sur-face with water, shown in Eq. (3). The growth of the oxide film istypically described by a diffusion controlled reaction mechanism,and the main reactions are shown in Eqs. (4)–(8).

At lower temperatures, during the first hours of the exposure ofthe metal to SCW, the outward chromium diffusion and inwardoxygen diffusion forms the thin inner layer assumed to be Cr2O3.At higher temperatures, this layer transforms into the more stablespinel phase of FeCr2O4. The affinity of iron for oxygen and the sta-bility of the iron oxide phases at high temperatures are likely tocause diffusion of iron outward to the water–metal interface,where it forms the outer iron oxide layer. The absence of a Ni rich

Fig. 10. Arrhenius plot of the effective rate constants for the initial hydrogenevolution. The calculated activation energy is 57.42 kJ mol�1.

Fig. 11. Atomic density profile of water at the iron hydroxide surface at 640 �C andtotal water density of 0.044 g cm�3. Hydrogen of the water is shown in blue andoxygen in red. Surface iron is shown in green, hydrogen in light blue and oxygen inorange. (For interpretation of the references to colour in this figure legend, thereader is referred to the web version of this article.)

Fig. 12. Molecular configuration of the water-iron hydroxide interface showing thepenetrating water molecule and its environment where water splitting may occur.Water oxygen is shown in red and hydrogen in violet. The surface irons are shownin blue and the surface oxygens in yellow. Surface hydrogens have been omitted forclarity. (For interpretation of the references to colour in this figure legend, thereader is referred to the web version of this article.)

232 K.I. Choudhry et al. / Corrosion Science 83 (2014) 226–233

layer is probably due to the formation of NiFe2O4. This mixed oxideis known to be an effective catalyst in the water-splitting reaction,being unstable at very high temperatures and pressures. The for-mation and growth of the passivation layer happens over the firstweek of the experimental run, after this the metal dissolution andhydrogen evolution both reach a steady state.

Fresh metal surface:

Feþ 2H2O! FeðOHÞ2 þH2 ð3Þ

Passivation oxide layer:

3FeðOHÞ2 ! Fe3O4ðmagnetiteÞ þ 2H2OþH2 ð4Þ

2Fe3O4 þH2O! 3Fe2O3 þH2 ð5Þ

6FeðOHÞ2 ! 3Fe2O3ðhematiteÞ þ 3H2Oþ 3H2 ð6Þ

2H2O! 2H2 þ O2ðcatalysts NiFe2O4; FeO=Fe3O4Þ ð7Þ

4Fe3O4 þ O2 ! 6Fe2O3 ð8Þ

It is known that the application of hydrogen water chemistry innuclear plants is expected to reduce oxidizing species and mitigatecorrosive environments. Once the hydrogen evolution rate hadreached a steady state of around 0.25 � 10�3 mol L�1, for the reac-tor surface area of 65.97 cm2 (see Fig. 2), air saturated water, withan oxygen concentration of around 8 mg L�1, was introduced intothe flow system. The extent of oxygen consumption was obtainedby measuring the hydrogen and oxygen concentrations in the efflu-ent. A very low concentration of dissolved oxygen of around40 lg L�1 was observed at the reactor exit, indicating that almostall of the oxygen was consumed, by the hydrogen produced fromthe surface reaction. This suggests that the surface catalysed ther-mochemical reactions may generate sufficient hydrogen to sup-press the accumulation of oxidizing species generated byradiolytic processes in the coolant water of the SCWR.

4. Simulation insights

The results of our computer simulations of supercritical water,in contact with a surface of iron hydroxide, published in a previousstudy, showed that the density of the surface water layer relativeto the bulk increases when the total water density in the systemis decreased from 0.055 to 0.044 g cm�3, at the SPC/E model tem-perature of 640 �C [38]. The higher fraction of water present atthe interface relative to the bulk, in a lower density system, nicelysupports current experimental findings that the hydrogen evolu-tion reactions occur on the surface of the oxide/hydroxide layer,as evident in the greater hydrogen evolution rates in the lowerdensity systems (see Fig. 6).

An interesting phenomenon observed in these simulations wasthe penetration of water, roughly 4 or 5 Å into the surface. Theatomic density profile of the SCW at the surface interface is shownin Fig. 11. The hydrogen of the water molecules is shown in blueand the oxygen in red. It can be seen in this figure that a small frac-tion of water works its way into the surface, evident by a smallshoulder on the water density profile, at about 200–205 Å fromthe centre of the simulation cell. One may speculate that hydrogenevolution can occur by hydrolysis of the water, catalysed by themetal hydroxide. It was shown by Nahtigal and Svishchev thatwater splitting can occur in NaCl nanoclusters formed at supercrit-ical water conditions via proton transfer from confined water tochloride ions, followed by hydrolysis product partitioning. It wasshown that the localised electrostatic fields in an NaCl nanoclustercan reach upwards of 1010 V m�1, which is more than sufficient todissociate the water molecule captured within a nanocluster [42].In this scheme, once water penetrates the iron oxide/hydroxide

layer the proton dissociates from the water molecule along theelectric field of the water-iron interaction. The hydroxide groupis bound by the potential well of the iron, creating the surfacehydroxide and the newly freed proton combines with other pro-tons and evolves from the system as hydrogen gas, with the elec-trons being supplied by the surrounding Fe2+ metal ions. Fig. 12shows the molecular configurations at the iron hydroxide-SCWinterface, in which the water molecule that has worked its wayinto the ionic surface can be seen.

5. Conclusion

In this study a flow-through reactor is used to determine thehydrogen evolution rate from fresh and oxidized stainless steelsurfaces, at supercritical water conditions. The hydrogen concen-tration in the flow at the exit of the reactor reaches a steady stateafter about 1 week of exposure of fresh steel surface to supercriti-cal water. A linear relationship is found between the steady statehydrogen concentration and the exposed surface area of the reac-tor tube. Effective rate constants for hydrogen evolution are deter-mined. The results show that the evolution of hydrogen gas is azero-order reaction, with the activation energy of 105.9 kJ mol�1

for the oxidized surface. Molecular dynamics simulations indicatethat at supercritical conditions, water molecules can penetrate

K.I. Choudhry et al. / Corrosion Science 83 (2014) 226–233 233

the iron hydroxide surface where catalytic water splitting may oc-cur due to the local electrostatic fields in the corrosion layer. Final-ly, the results suggest that the hydrogen evolved from the reactorsurface may be sufficient to reduce the accumulation of dissolvedoxygen, thus inhibiting corrosion during the operation of theSCWR.

Acknowledgement

The authors are grateful for the financial support of the Gener-ation IV Energy Technologies Program. Funding to the GenerationIV Program was provided by Natural Resources Canada throughthe Office of Energy Research and Development, Atomic Energyof Canada Limited, and the Natural Sciences and Engineering Re-search Council of Canada.

References

[1] D. Guzonas, F. Brosseau, P. Tremaine, J. Meesungnoen, J.-P. Jay-Gerin, Waterchemistry in a supercritical water-cooled pressure tube reactor, Nucl. Technol.179 (2012) 205–219.

[2] X. Luo, R. Tang, C. Long, Z. Miao, Q. Peng, C. Li, Corrosion behaviour of austeniticand ferritic steels in supercritical water, Nucl. Eng. Technol. 40 (2007) 147–154.

[3] I.M. Svishchev, D. Guzonas, Particle nucleation: implications for waterchemistry control in a GEN IV supercritical water-cooled nuclear reactor, J.Supercrit. Fluids 60 (2011) 121–126.

[4] G.S. Was, P. Ampornrat, G. Gupta, S. Teysseyre, E.A. West, T.R. Allen, K.Sridharan, L. Tan, Y. Chen, X. Ren, C. Pister, Corrosion and stress corrosioncracking in supercritcal water, J. Nucl. Mater. 371 (2007) 176–201.

[5] F. Nordmann, Aspects on chemistry in french nuclear power plants, in:Proceedings of the 14th International conference on the properties of waterand steam, Kyoto, Japan, 2004, pp. 521–530.

[6] F. Cattant, D. Crusset, D. Féron, Corrosion issues in nuclear Industry today,Mater. Today 11 (2008) 32–37.

[7] P.J. Millett, K. Fruzzetti, Status of application of amines in US PWRs, PPChem 7(2005) 599–603.

[8] K. Ishigure, Power cycle chemistry in nuclear cycles: technology developmentand importance of fundamental data, in: Proceedings of the 14th internationalconference on the properties of water and steam, Kyoto, Japan, 2004, pp. 512–520.

[9] O.P. Arkhipov, V.L. Bugaenko, S.A. Kabakchi, V.I. Pashevich, Thermal andradiation characteristics of hydrazine in the primary loop of nuclear powerplants with water-cooled, water-moderated reactors, At. Energy 82 (1997) 92–98.

[10] P.J. Millett, C.J. Wood, EPRI, Recent advances in water chemistry control at USPWRs, in: 58th International water conference, Pittsburg, USA, 1997.

[11] R.A. Carvajal-Ortiz, A. Plugatyr, I.M. Svishchev, On the pH control atsuprecritical water-cooled reactor operating conditions, Nucl. Eng. Des. 248(2012) 340–342.

[12] U. Ehrnstén, Corrosion and Stress Corrosion Cracking of Austenitic StainlessSteels, in: R.J.M. Konings, T.R. Allen, R.E. Stoller, S. Yamanaka (Eds.),Comprehensive Nuclear Materials, Elsevier, Amsterdam, 2012, pp. 93–104.

[13] S. Fyfitch, Corrosion and Stress Corrosion Cracking of Ni-Base Alloys, in: R.J.M.Konings, T.R. Allen, R.E. Stoller, S. Yamanaka (Eds.), Comprehensive NuclearMaterials, Elsevier, 2012, pp. 69–92.

[14] T.R. Allen, R.J.M. Konings, A.T. Motta, Corrosion of Zirconium Alloys, in: R.J.M.Konings, T.R. Allen, R.E. Stoller, S. Yamanaka (Eds.), Comprehensive NuclearMaterials, Elsevier, Amsterdam, 2012, pp. 49–68.

[15] C.R.F. Azevedo, Selection of fuel cladding material for nuclear fission reactors,Eng. Fail. Anal. 18 (2011) 1943–1962.

[16] T.R. Allen, Y. Chen, X. Ren, K. Sridharan, L. Tan, G.S. Was, E. West, D. Guzonas,Material Performance in Supercritical Water, in: R.J.M. Konings, T.R. Allen, R.E.Stoller, S. Yamanaka (Eds.), Comprehensive Nuclear Materials, Elsevier,Amsterdam, 2012, pp. 279–326.

[17] M. Blat, D. Noel, Detrimental Role of Hydrogen on the Corrosion Rate ofZirconium Alloys,’’ Zirconium in the Nuclear Industry, in: E.R. Bradley, G.P.Sabol (Eds.), Zirconium in the Nuclear Industry: Eleventh InternationalSymposium, ASTM STP 1295, 1996, pp. 319–337.

[18] M. Blat, L. Legras, D. Noel, H. Amanrich, Contribution to a better understandingof the detrimental role of hydrogen on the corrosion rate of zircaloy-4 claddingmeterials, in: G.P. Sabol, G.D. Moan (Eds.), Zirconium in the Nuclear Industry:Twelfth International Symposium, ASTM STP 1354, 2000, pp. 563–591.

[19] M. Harada, R. Wakamatsu, The effect of hydrogen on the transition behavior ofthe corrosion rate of zirconium alloys, in: B. Kammenzind, M. Limbäck (Eds.),Zirconium in the Nuclear Industry: 15th International Symposium, ASTM STP1505, 2009, pp. 384–402.

[20] S. Abolhassani, R. Restani, T. Rebac, F. Groeschel, W. Hoffelner, G. Bari, W. Goll,F. Aeschbach, TEM examinations of the metal-oxide interface of zirconiumbased alloys irradiated in a pressurized water reactor, in: P. Rudling, B.Kammenzind (Eds.), Zirconium in the Nuclear Industry: FourteenthInternational Symposium, ASTM STP 1467, 2005, pp. 467–493.

[21] M. Oskarsson, E. Ahlberg, U. Sodervall, U. Andersson, K. Pettersson, Pre-transition oxidation behaviour of pre-hydrided Zircaloy-2, J. Nucl. Mater. 289(2001) 315–328.

[22] Y.S. Kim, H.K. Woo, K.S. Im, S.I. Kwun, The cause for enhanced corrosion ofzirconium alloys by hydrides, in: G.D. Moan, P. Rudling (Eds.), Zirconium in theNuclear Industry: Thirteenth International Symposium, ASTM STP 1423, 2002,pp. 274-296.

[23] E. Chan, Application Note: Measuring hydrogen in water steam cycle forcorrosion trend study, SWAN Analytische Instrument AG, Switzerland, 2013.

[24] L. D’Souza, Thermochemical hydrogen production from water using reducibleoxide materials: a critical review, Mater. Renew. Sustain. Energy 2 (7) (2013)2–12.

[25] Y.-S. Kim, Y.-H. Jeong, S.-B. Son, A study on the effects of dissolved hydrogen onzirconium alloys corrosion, J. Nucl. Mater. 444 (2014) 349–355.

[26] H. Je, A. Kimura, Stress corrosion cracking susceptibility of oxide dispersionstrengthened ferritic steel in supercritical pressurized water dissolved withdifferent hydrogen and oxygen contents, Corros. Sci. 78 (2014) 193–199.

[27] M. Koyama, E. Akiyama, T. Sawaguchi, K. Ogawa, I.V. Kireeva, Y.I. Chumlyakov,K. Tsuzaki, Hydrogen-assisted quasi-cleavage fracture in a single crystallinetype 316 austenitic stainless steel, Corros. Sci. 75 (2013) 345–353.

[28] Y. Mine, T. Kimoto, Hydrogen uptake in austenitic stainless steels by exposureto gaseous hydrogen and its effect on tensile deformation, Corros. Sci. 53(2011) 2619–2629.

[29] I.M. Svishchev, R.A. Carvajal-Ortiz, K.I. Choudhry, D.A. Guzonas, Corrosionbehavior of stainless steel 316 in sub- and supercritical aqueousenvironments: effect of LiOH additions, Corros. Sci. 72 (2013) 20–25.

[30] Metrohm, VA Application Note No V-82: Different Chromium Species in SeaWater, in: M. AG (Ed.), VA Application Notes, Metrohm AG, 2003.

[31] Metrohm, VA Application Note No V-123: Iron (total) in ethylene grlycol with2 3 dihydroxynaphtalene, in: M. AG (Ed.), VA Application Notes, Metrohm AG,2003.

[32] Metrohm, VA Application Note No. V-87: Nickel and cobalt in drinking water,in: M. AG (Ed.), VA Application notes, Metrohm AG, 2003.

[33] A.P. Lyubartsev, A. Laaksonen, M.DynaMix – a scalable portable parallel MDsimulation package for arbitrary molecular mixtures, Comp. Phys. Commun.128 (2000) 565–589.

[34] A. Plugatyr, I.M. Svishchev, Accurate thermodynamic and dielectric equationsof state for high temperature simulated water, Fluid Phase Equilib. 227 (2009)141–151.

[35] A. Plugatyr, R. Carvajal-Ortiz, I.M. Svishchev, Ion-pair association constant forLiOH in supercritical water, J. Chem. Eng. Data 56 (2011) 3637–3642.

[36] R.T. Cygan, J.J. Liang, A.G. Kalinichev, Molecular models of hydroxide,oxyhydroxide and clay phases and the development of a general force field,J. Phys. Chem. B 108 (2004) 1255–1266.

[37] R.T. Cygan, V.N. Romanov, E.M. Myshakin, Molecular simulation of carbondioxide capture by montmorillonite using an accurate and flexible force field, J.Phys. Chem. 116 (2012) 13079–13091.

[38] I.M. Svishchev, D.T. Kallikragas, A.Y. Plugatyr, Molecular dynamics simulationsof supercritical water at the iron hydroxide surface, J. Supercrit. Fluids 78(2013) 7–11.

[39] M.P. Allen, D.J. Tildesley, Computer Simulation of Liquids, Clarendon Press,New York, 1989.

[40] A. Plugatyr, T.M. Hayward, I.M. Svishchev, Thermal Decomposition ofhydrazine in sub- and supercritical water at 25 MPa, J. Supercrit. Fluids 55(2011) 1014–1018.

[41] M. Neises, M. Roeb, M. Schmücker, C. Sattler, R. Pitz-Paal, Kineticinvestigations of the hydrogen production step of a thermochemical cycleusing mixed iron oxides coated on ceramic substrates, Int. J. Energy Res. 34(2010) 651–661.

[42] I.G. Nahtigal, I.M. Svishchev, Generation and Integration of NaOH into NaClclusters in supercritical water: a molecular dynamics study on hydrolysisproduct partitioning, J. Phys. Chem. B 113 (2009) 14681–14688.