~ Pergamon
`
`Talanta, Vol. 41, No. 2, pp. 211-215, 1994
`Copyright «) 1994 Elsevier Science Ltd
`Printed in Great Britain. AU riahts reserved
`0039-9140/94 $6.00 + 0.00
`
`REMOVAL OF DISSOLVED OXYGEN FROM WATER:
`A COMPARISON OF FOUR COMMON TECHNIQUES
`
`IAN B. BUTLER; • MARTIN A. A . ScuooNEN2 and D AVID T. RlcKARD1
`' Department of Geology, Univenity of Wales, College of Cardiff, Cardiff, CF! 3YE, U.K.
`2Department of Earth and Space Sciences, SUNY at Stony Brook, Stony Brook, NY I I 794-2100, U.S.A.
`
`(Received 30 March 1993. Revised 9 August 1993. Accepted 9 August 1993)
`
`Sum...-y- Four commo n techniques for the removal of dissolved oxygen fro m water have been examined:
`boiling at I atm, boiling under reduced pressure, purging with N2 and sonication under reduced pressure.
`After treatment, the residual oxygen in solution was analysed by the Wink.Jee method. Nitrogen purging
`for 20--40 min at flow rate of 25 mL/ s was found to be the most effective oxygen removal method. Boiling
`at 1 atm was found to be the least effective, None of the techniques evaluated here lead to complete
`removal of oxygen. 111e C'\ll~tration of re$id~I di$$01Ved oxy~ after purgi11g for 20-40 minutes with
`nitrogen is 0.2--0.4 ppm.
`
`The removal of dissolved oxygen from water
`is important when experimenting with redox
`sensitive reagents. For example, it is vital
`when conducting experiments with metal sul(cid:173)
`phide systems. 1 Dissolved oxygen can be
`removed from solution by both chemical
`and physical means. Chemical methods can be
`extremely effective. For example, addition of
`titanium (Im citrate removes dissolved oxygen
`and poises the EH(pe) of a neutral solution
`anywhere between -480 mV (pe - -8.11)
`(equilibrium potential reported by Zehnder and
`Wubnnann)2 and -300 mV (pe = -5.07)
`(Rickard and Oldroyd, unpublished work).
`Addition of Ti(III) citrate has been used by
`Butler and Rickard3 in the synthesis of fram(cid:173)
`boidal pyrite, but it may not be suitable in
`experiments where titanium ions may catalyse
`reactions or sorb onto surfaces. In experiments
`where addition of a redox poise is undesirable,
`a physical method must be employed to remove
`oxygen. Although often used, the effective(cid:173)
`ness of physical methods is rarely reported.
`The objective of this study is to evaluate the
`effectiveness of four physical methods that are
`in common usage. These arc: purging with
`nitrogen, argon or a similar inert gas,4 boiling
`at 1 atm/·6 sonication unde, "vacuum"7 and
`boiling under " vacuum". Other methods that
`have been used in a few studies, but that are
`not evaluated here are degassing by passing
`
`water through a gas permeable plastic tube
`in a vacuum chamber and boiling water while
`bubbling with nitrogen.9
`
`EXPERIMENT AL
`
`Methods for dissolved oxygen removal were
`as follows.
`
`Nitrogen purging
`One litre of deionized water was placed in a
`PTFE stoppered pyrex reaction kettle. A glass
`bubbler was fitted and the water was bubbled
`with prepurified grade nitrogen for 5, 20, 40 and
`60 min at room temperature. No stirring was
`used because the mixing caused by the passage
`of the nitrogen was considered sufficient to keep
`the water well mixed. Nitrogen was allowed to
`escape by means of a small vent hole. No air
`trap was fitted because the high N 2 flow rates
`made such a device o bsolete. The water purged
`for 20 min with pre-purified grade nitrogen
`was used lo make all analytical reagents for
`this study; these reagents were preserved under
`N 2 during the balance of the investigation.
`The oxygen content of the prepurified nitrogen,
`according to the distributor, is < 5 ppm. In one
`set of experiments ultra high purity nitrogen
`with an oxygen content of <0.5 ppm was used
`to purge water for 30 min. The flow rate of
`the nitrogen was measured by timed collection
`of the expelled gas in an upturned measuring
`cylinder full of water. The flow rate used in these
`experiments was approximately 25 ml of gas
`
`21 1
`
`Eton Ex. 1082
`1 of 5
`
`

`

`212
`
`IAN B. Bun.ER et al.
`
`(at l atm) per sec. The deionized water used in
`this study has a resistance of J 8 Mn.
`
`Boiling at I atm
`One litre of deionized water was placed in a
`conical flask and boiled for 30 min on a hot
`plate in open air. The boiling water was then
`poured into a plastic bottle and the remaining
`air in the bottle was squeezed out before tightly
`capping. The bottle was then cooled in running
`tap water.
`
`Boiling under reduced pressure
`Five hundred millilitres of deionized water
`were placed in a stoppered Buchner flask and
`attached to a hand vacuum pump. A trap
`was placed between the Buchner flask and the
`pump to prevent significant quantities of water
`getting into the pump. After boiling the water
`for 30 min, the flask was sealed and cooled in
`running water.
`
`Sonication under reduced pressure
`The procedure was the same as for boiling
`under vacuum except that the Buchner flask was
`placed in a sonic bath and sonicated for periods
`of both 30 min and l hr.
`Trace oxygen in water was analysed by a
`Winkler titration as described by Clesceri et al.10
`Addition of Mn(II) sulphate solution plus a
`strongly alkaline solution of Nal to the system
`results in the precipitation of manganese hy(cid:173)
`droxides of higher oxidation states. Iodide in
`solution is in tum oxidized by the manganese
`hydroxides to iodine. The amount of iodine
`produced is in proportion to the amount of
`oxygen in solution. Iodine was then titrated
`against a solution of sodium thiosulphate
`which had been previously standardized against
`0.0250N ± 0.0005N iodine solution. The overall
`reaction scheme for this method is summarized
`in Table I.
`All water samples for analysis were taken and
`treated in 250 ml conical flasks with ground
`glass stoppers. Analysis of the contents of
`each flask was carried out immediately after
`treatment.
`
`RI.sULTS
`The results of this study arc summarized in
`Table 2. The errors given with each oxygen
`concentration in Table 2 are the random
`errors due to uncertainties in the experimental
`
`Table I. Summary of the reaction scheme for the
`Winkler mdhod. After Hitchman 11
`Mn2 ' + 2OH- = Mn(OH),
`2Mn(OH>, + ½0. - HiO = 2MD(OH),
`2Mn(OH)3 + 6H+ + 31- = 2Mnz+ + lj + 6H2O
`lj=l,+ l -
`12 + s,~ = 21- + s.~-
`
`technique. Reagents were prepared using water
`purged with nitrogen; when water prepared by
`that method was analysed it was assumed that
`any oxygen in the reagents represented neither
`an addition to, or subtraction from, the system.
`Thus it is true that any error due to oxygen
`in the Mn(ll) sulphate and Nal solutions is
`appreciably smaller than the random error and
`so is negligible.
`There is a systematic error present due to
`oxidation of Mn(II) to higher oxidation states in
`the solid during storage. This error was found to
`
`Method
`N2 purging
`(pre-purified grade,
`<5 ppm 0 2)
`
`(Ultra high purity
`grade, <0.5 ppm O,)
`
`Boiling at I aim
`
`Table 2. O1.ygen c:owxntratiom of deionized water after
`treatment to remove dmolved 01.ygm'
`Concentration 0 2
`Time
`5 mins 0.47 ppm± 0.05 ppm
`5 mins 0.56 ppm ± 0.05 ppm
`5 mins 0.55 ppm ± 0.05 ppm
`20 mins 0.35 ppm ± 0.05 ppm
`20 mins 0.33 ppm ± 0.05 ppm
`20 mins 0.42 ppm ± 0.05 ppm
`20 mins 0.34 ppm ± 0.OS ppm
`20 mins 0.36 ppm ± 0.05 ppm
`40 mins 0.18 ppm±0.05 ppm
`40 mins 0.23 ppm ± 0.05 ppm
`60 mins 0. I 9 ppm ± 0.05 ppm
`60 mins 0.22 ppm ± 0.05 ppm
`30 mins 0.28 ppm ± 0.05 ppm
`30 mins 0.24 ppm± 0.0S ppm
`30 mins 0.30 ppm ± 0.05 ppm
`30 mins 0.2S ppm ± 0,05 ppm
`30 mins 1.20 ppm ± 0.05 ppm
`30 mins 1.21 ppm ± 0.05 ppm
`30 mins 0.87 ppm ± O.OS ppm
`30 mins 0.93 ppm ± 0.05 ppm
`30 mins 0.77 ppm± 0.05 ppm
`30 mins 0.75 ppm± 0.05 ppm
`30 mins 1.00 ppm ± 0.0S ppm
`30 miiu 0.8S ppm ± 0.0S ppm
`30 mins 0.34 ppm ± 0.05 ppm
`30 mins 0.3 I ppm ± 0.05 ppm
`30 mins 0.31 ppm± 0.05 ppm
`30 mins 0.28 ppm ± 0.05 ppm
`30 mins 0.28 ppm ± o.os ppm
`30 mins 0.26 ppm ± 0.05 ppm
`30 mins 6.92 ppm± 0.05 ppm
`60 mins 0.67 ppm ± 0.05 ppm
`30 mins 2.38 ppm ± 0.05 ppm
`30 min• 1.18 ppm ± 0.05 ppin
`60 mins 0,55 ppm± 0,05 ppm
`60 mins 0.49 ppm ± 0.05 ppm
`
`Boiling under
`reduced pressure
`
`Sonication under
`reduced pressure
`
`'Concentrations derived from analysis of iodine produced
`directly from oxidation of manganese hydroxides by
`oxygen. Errors are the random errors due to uncertainty
`in experimental technique.
`
`Eton Ex. 1082
`2 of 5
`
`

`

`Removal of dissolved oxygen from water
`
`213
`
`be +0.05 ppm 0 2 • Hence this represents a
`systematic over estimation of the 0 2 content
`by +0.05 ppm. The data in Table 2 have been
`corrected to take this error into a<:eount.
`
`DISCUSSION
`Purging with nitrogen appears to be both a
`quick and effective procedure to scrub deionized
`water of dissolved oxygen. The optimum
`duration of the scrubbing at the fl.ow rate used
`in these experiments is about 30-40 min (Fig. I).
`Purging for longer periods, some workers
`bubble solutions for up to I day or more, will
`not further reduce the oxygen concentration.
`The kinetics of the degassing process depend
`upon the gas/liquid interface area, i.e. the re(cid:173)
`moval of dissolved oxygen is faster when nitro(cid:173)
`gen is introduced as small bubbles. Therefore,
`a sintered glass bubbler to produce small N2
`bubbles is a critical element in a purging setup.
`The data also seem to indicate that the use
`of expensive ultra high purity grade nitrogen
`( <0.5 ppm 0 2 ) does not measurably aid
`the removal of dissolved oxygen from water
`(Fig. 1). In this investigation no difference was
`found between the concentration of residual
`oxygen expected using pre-purified nitrogen
`( < 5 ppm 0 2) and that observed using ultra
`high purity nitrogen as the purge gas.
`
`Error ~r, ore t 1om~ord error of tk~ m.:on.
`
`O.GO
`
`E 0.!iO
`Q.
`
`Q. so.~o
`-0 " "a 0.30
`
`~
`0
`_0.20
`g
`in 10
`"'
`
`'O
`
`1 o.oo 20.00 J0.00 •O 00 ~ .00 60.00 70 00
`Durolion of Purging, mins .
`
`Fig. I. Residual dissolved oxygen concentration after purg(cid:173)
`ing 1 1 of distilled/deionized waler with nitrogen al a flow
`rate of 25 ml/sec at I aun as a function of time. The initial
`oxygen concentration of the water was 8.5 ppm (measured
`with oxygen efectrode) and the pH was near neutral. Most
`depletion occurred in the first 5 min. Further purging for 40
`min to I hr l'Cj!ulted in a constant residual oxygen concen(cid:173)
`tration of about 0.25 ppm. II is clear from this graph that
`extended durations of nitrogen purging do nol noticeably
`further reduce the concentration of dis1<>lved oxygen in
`solution. Purging with ultra high purity nitrogen is no more
`effective · for the removal of dissolved oxygen than purging
`with prepurified grade nitrogen.
`
`TAL 41/2 D
`
`11.00
`
`8.5()
`
`E g;e.oo
`" 8'?.50 g
`
`"'7.00
`1
`ii: G.~O
`C/1 a
`G,0,0
`
`Fig. 2. Equilibrium dissolved 0 2 concentration vs tempera(cid:173)
`ture. The data are calculated using Henry's law constants
`taken from the SOL THERM databue. u In all calculations
`it is assumed that Po, equals 0.2L atm. At 2s~c. the
`equilibrium 102] is 8.52 ppm, thus N 2 purging for 40 min
`represents a greater than 40 times reduction in 102].
`
`Boiling water at I atm is probably the most
`common of the methods under investigation
`here. The equilibrium con<;entration of oxygen
`in water at I oo~c in contact with air is 6.17 ppm
`(calculated using the Henry's law constant for
`0 2 in water at 100°C12 and assuming a Po, equal
`to 0.21 atm). This is significantly higher than
`the values found in this study. Therefore the
`removal of oxygen by this method clearly is
`not an equilibrium process. During the boiling
`process bubbles of water vapour, depleted in
`oxygen, are produced and it is with these that
`gas exchange takes place. Dissolved oxygen is
`entrained in the bubbles and then liberated to
`the atmosphere at the liquid surface. The liquid
`surface during vigorous boiling is extremely
`turbulent and so it is probable that a back.
`reaction is taking place whereby atmospheric
`oxygen is able to redissolve in the water and is
`carried back into solution to be available once
`more for exchange at the liquid/bubble inter(cid:173)
`face. This process may account for both the
`variability of the analytical results and the rela(cid:173)
`tive ineffectiveness of boiling at 1 atm as a
`method for the deoxygenation of water.
`Boiling under reduced pressure is not only
`an effective method for removal of dissolved
`oxygen but it is also the most reprodudble
`method studied here. Since oxygen is removed
`once liberated from the solution, the P~ of
`the gas phase in contact with the water is lower
`than for boiling at I atm. Hence the back
`reaction (uptake of 0 2) will be of less import(cid:173)
`ance and removal of dissolved oxygen more
`effective. More complete deoxygenation could
`
`Eton Ex. 1082
`3 of 5
`
`

`

`214
`
`IAN 8. Bun.ER et al.
`
`Nitrogen,.
`
`' Nitrogen In.
`
`Reaction
`Kettle.,
`
`Fig. 3. Apparatus for N2 purging of I I of deionized water.
`With this set-up it is possible to achieve [02] equal to 0.2
`ppm with Nl purging for 40 min. The sintered glass bubbler
`is CMC11tial to obtain small bubbles, thus maximizing the
`surfa(;C area available for gas exchange. If a high N 2 flow
`rate is maintained then an 0 2 trap is not needed at the gu
`outlet.
`
`abiotic redo,i. reactions. Thus for hydrothermal
`experiments, boiling under reduced pressure or
`perhaps nitrogen purging followed by boiling at
`reduced pressure should be employed.
`It should be emphasized, however, that all the
`methods reviewed here leave some oxygen in
`solution and the complete elimination of oxygen
`from solution is not possible by any of the
`techniques evaluated in this study. In terms of
`the redox state of the water, the residual oxygen
`is significant. For example, 0.25 ppm residual
`0 2 at 25°C is equivalent to the equilibrium 0 2
`concentration in contact with an atmosphere
`with P0: equal to 10-2
` atm. It is clear from
`Fig. 4 a pe/pH diagram, that even the best
`physical methods for the removal of dissolved
`oxygen cannot produce even a mildly reducing
`solution. Reduced conditions can only be
`obtained by using a redox poise, such as
`Ti(III) citrate. However, for some experiments
`it may be a challenge to find a redox poise that
`will only react with the dissolved oxygen and
`not with any other redox sensitive species or
`surface.
`The inability to remove all dissolved oxygen
`with physical methods has important impli(cid:173)
`cations for the field of sulphide geochemistry.
`For example, the work by Sato•s indicates that
`most sulphide minerals would be oxidized by
`the deoxygenated water produced in this study.
`Hence, exposure of most sulphide minerals to
`deoxygenated water is likely to lead to some
`change in their surface properties.1 The residual
`dissolved oxygen becomes particularly import(cid:173)
`ant if the system under study has little or no
`
`-2
`
`probably be achieved by the use of a more
`efficient vacuum pump or aspirator. Water
`deoxygenated by this method needs relatively
`little cooling before it can be used and use of a
`stronger pump would reduce this factor still
`further.
`Sonication was used as an aid to bubble
`nucleation by Schoonen et a/.7 but the present
`study indicates that the process has little virtue
`other than to wann the water. Indeed the initial
`experiment resulted in an oxygen concentration
`of almost 7 ppm after 30 min; significant deoxy(cid:173)
`genation was only achieved after the wa.ter had
`reached 45°C. The boiling process produced in
`lhe sonic bath is less vigorous than that pro(cid:173)
`duced by a hotplate and this may account for
`the longer duration required for deoxygenation
`of the water. Clearly the sonication itself is
`oflittle value and similar results could probably
`be obtained using a water bath to maintain an
`evenly distributed temperature.
`It should be noted that the results of this
`study are in sharp contrast to statements made
`by HungateY Hungate has reported that boiling
`for l min results in almost complete deoxygena(cid:173)
`tion and that purging with gas for 30 min to 1 hr
`is less effective than boiling for this brief period.
`No raw data was presented by Hungate13 to
`back up these conclusions and no final 0 2
`concentrations were reported. Clearly, from
`the present study, significantly better removal of
`dissolved oxygen can be achieved by purging
`or boiling under vacuum than by boiling at
`1 atm.
`In summary, the authors recommend use
`of pre-purified grade nitrogen, a sintered glass
`bubbler, high gas flow rates and an air-tight
`container (with a small vent hole to let the purge
`gas oul) for effective removal of oxygen by
`this method (Fig. 3). This method may be
`further improved if 0 2 is removed from the
`N2 gas before passing it through the water. For
`example, oxygen may be removed by passing
`nitrogen over hot copper turnings. Nitrogen
`purging does also have the virtue that the
`solution is cool and ready for use immediately
`after oxygen removal, in contrast to other
`techniques reviewed here. A possible drawback
`of purging with nitrogen is that there will be
`nitrogen in solution after oxygen removal.
`At low temperatures (perhaps < 150°C) nitro(cid:173)
`gen is essentially inert and should not be a
`factor unless reactions involving nitrogen fixing
`bacteria occur. However, at higher temperatures
`( > l 50°C) dissolved nitrogen may be involved in
`
`Eton Ex. 1082
`4 of 5
`
`

`

`Removal of dissolved oxygen from water
`
`215
`
`20 ~ -~ - -~ -~ - -~ - - , - - - - - . 20
`
`15
`
`-5
`
`-~
`
`- 10 '--- -L--.L-----'----'----'--...=-, -10
`10
`9
`8
`5
`6
`7
`4
`pH
`Fig. 4. pe (E") vs pH diagram illustrating the redox state
`attained via nitrogen purging to remove dissolved oxygen.
`For comparison, the pe range obtained by adding a Ti(lll)
`citrate solution is also shown. The four physical methods
`evaluated in this study lead to oxygen depleted, but still
`oxidizing water. Data from Drever" were used to calculate
`this diagram. The diagram was calculated for 298 K and a
`total pressure of I atm. The pressure contours for 0 2 and
`H2 are given in aim. At 298 K , EH(mV) = 5.9•pe.
`
`redox buffer capacity. For example, in very
`dilute H2S solutions, the remaining 0 2 may
`react with a significant portion of the dissolved
`H2S and radically change the solution pe (EH)
`and the sulfur speciation.
`Finally it should be noted that although
`purging with N2 is an efficient method to remove
`dissolved oxygen from deionized water, it is a
`poor method to preserve soluti-ons contain(cid:173)
`ing redox-sensitive species. Commercial grade
`nitrogen contains significant quantities of
`oxygen ( < 5 ppm for prepurified grades and
`< 0.5 ppm for oxygen free and ultra high purity
`grades) and bubbling of N2 through a solution
`containing redox sensitive species or minerals
`represents a constant advection of 0 2 through
`the system. The 0 2 in the gas is essentially
`scrubbed by the solution it passes through and
`will cause oxidation if bubbling is carried out
`over extended periods. For example, Pourbaix
`and Pourbaix16 have shown, with reference
`to corrosion and aqueous sulphide systems,
`that occasional exposure to limited amounts of
`
`oxygen is not necessarily corrosive, but that
`continuous exposure of H2S solutions to traces
`of oxygen can produce a significant pH drop
`possibly leading to corrosion. A better method
`to preserve such solutions would be to maintain
`them under a nitrogen or methane atmosphere,
`so allowing a minimum of exchange with
`oxygen impurities in the gas.
`
`Acknowledtements- This
`research was conducted at
`SUNY-Stony Brook during a visit by the senior author and
`was supported by NSF-EAR9116846 to M.A. A. Schoonen.
`The senior author is supported by a postgraduale student(cid:173)
`ship awarded by NERC (GTU/90/0Sl27) and a NERC
`research grant (OR3/7476) to D. T. Rickard. The authors
`would like lo thank 0. Kacandes and H. L. Barnes for
`their thorough reviews and helpful suggestions to improve
`the manuscript.
`
`REFERENCFS
`
`I. D. Fornasiero, V. Eijt and J. Ralston. Coll. 1111(/ Surf,
`1992, 61, 63.
`2. A. J. B. Zehnder and K. Wuhrrn:mn, Science, 1976, 191,
`1165.
`3. I. B. Butler and D. Rickard, inorganic Se/f-organizalion
`in Pyrite Framboids (in prep).
`4. T. Arakaki and J. W. Morse, Geochim. et Cosmochim.
`Acta, 1991, 57, 9.
`S. M. A. A. Schoonen and H. L. Barnes, Geochim. et
`Cosmochim. Acta, 1991, 55, 1495.
`6. B. Takano, M. A. McKibben and H. L. Barnes, Anal.
`Chem., 1984, 56, 1.594.
`7. M. A. A. Schoonen, N. S. Fisher and M. Wente,
`Geochim. et Cosmochim. Acta, 1991, 56. 1801.
`8. D. Takano and K. Watanuki, Talanta .. 1988, 35, 847.
`9. J. W. Stucki, D. C. Golden and C. B. Roth, Clays
`Clay Min., 1984, 32, 191.
`IO. L. S. Clesceri, A. E. Greenburg and R. R. Trussell (eds),
`Standard Methods for the E:rnmi,wtion of Water and
`Waste Warer, 17th Ed., 1989.
`I I. M. L . Hitchman, Measurement of Dissolved Oxygen.
`Wiley, New York, 1978.
`12. M. H. Recd and N. F. Spycher, SOLTHERM: Data
`Base of Equilibrium Constants for Aqueous-Mineral-Gas
`EquUibria, 1990. Dept. of Geo. Sci. Uni of Oregon,
`Eugene, OR 97403.
`13. R . E. Hungate in Methods in Microbiology, J. R . Norris
`and D. W. Ribbons (eds). Academic Press, New York,
`1969.
`14. J. I. Drevcr, Geochemistry of Natural Waters. Prentice
`Hall, Englewood Cliffs, NJ.
`15- M. Sato, Geochim. et Cosmochim. Acta, 1992, 56, 3133.
`16. M. Pourbaix and A. Pourbaix, Geochiln. er Cosmochim.
`Acra, 1992, 56, 3157.
`
`
`
`View publication statsView publication stats
`
`Eton Ex. 1082
`5 of 5
`
`