AVAILABILITY OF INSTRUMENTS
Water-analysis instruments of the type described here are available from sekeral manufacturers under Du Pont license (4). For specific information on the several models currentkiavai1ab1e, inquiry should he made directly to Beckman Instruments, Inc., Fullerton, Calif.; Consolidated Elertrodynamics
Gorp.! Leeds 8~ Northrup, Philadelphia, Pa.; or Manu-
facturers Engineering and Equipment cOrP., Hatboro, Pa.
U. S. Patent 2,830,945 (April 15, 1958). ( 5 ) Taylor, E. S., Refrig. Eng. 64, 41 (4) Keidel, F. A,,
(1956).
LITERATURE CITED
(1) Brown, E. H., Whitt, C. D., Ind. Eng. Chem. 44,615 (1952). (2) Fontam, B. J., J . Am. C h a . SOC. 733 3348 (lg51). (3) Keidel, F. A. Proceedings of Ninth Annual Sympokum on Process Fluid Analyzers, Newark, N. J., April 1957.
RECEIVEDfor review April 8, 1958. Accepted September 15, 1959. Pittsburgh Conference on Anslytical Chemistr and Applied Spectroscopy,March 1956. k n t h Annual Symposium for Process Fluid Analyzers, New Jersey Section, Instrument Society of America, April 1957.
Continuous Coulometric Determination of Parts per Million of Moisture in Organic Liquids LELAND G. COLE, MICHAEL CZUHA, and RICHARD W. MOSLEY Consolidated Electrodynamics Corp., Pasadena, Calif.
DONALD T. SAWYER Department of Chemistry, University of Colirornio, Riverside, Calif. b Moisture in organic liquid$ is reliably determined with a continuous coulometric titrotion system. Moisture is removed from the sample stream by stripping and is passed through o coulometric cell consisting of a n onhydrous phosphorus pentoxide matrix between platinum electrodes. Accurate analyses can beobtained a t moisture levels down to 1 p.p.m.
A.
nunirmuc xii:,Iytii~:il twhriiqurs 1 t : t w l x m sugpc.strd for t h r deimnin:itinii of moisturr i n liquid3 (.?), thf, nbsentv of a gewctl :is w d as wliaM! n i e t l d h a s prornptrd contiiiuous reiwrrh. 1'1.1!rnnjority of the d s hatch mnnionly udrd ~ ~ n . t l i i ~ are tecltiiiqua. and require iainyk rollrrtion arid trnnsfw prior tc, lahorntory nnalysis. The muit generslly used prortdures vinploy thv ICirl Fisvhvr re3gi.nt ; this 11:~s hrcoinr tin. nrreptwl mrrliod for the ~lt~ti~rminati~,n of w i t i ~ in liquids during the p:~st 20 ycrrs. T1w gmeral prolh ms a i d limitations of tliis m e t l d h n w been ndcquatvly reviwid and diirussed hy \litrlirll and Smith in t h k (.xcrIIeIit trrntise on nqusini~try(%?)ant] I,! others ,3, 6-7,9). A l t h u g h the Kxrl Fisdwr method 1i:ta \vi& ;irwpt@nw,tivo iiiajor liniitations h a w iiidiratrrl tht, n e 4 for n ninre gmernl u o d reli.il,le metlird. I h t , nuinrrous iiintwiils 31.1 as interfrrrnrvs; parfivulorly trouhlesomr nrr strorig oxidizit.K or rriliicing agmts. Srcotid, the tldi,rtion of the Karl I'isrlier rnd ptiirit rcquirw cnreful twhniqiir xnd exprrii.ner if tlii' mrthod is to providi. rrlinl~h.nrialypes. I.TIIUITXI
2048
ANALWICAL CHEMISTRY
This paper describes an instrument based on an approach first suggested by Keidel (d), and compares the results obtained on this instrument with those obtained by conventional Karl Fischer titrations. Briefly, the instrument comprises a high-efficiency stripping column in which the water from a continuously flowing sample stream is quantitatively removed by a flowing gas stream and a coulometric cell which quantitatively measures the water in the effluent gas stream. EXPERIMENTAL
The Karl Fischer titrations were carried out with a system similar to the titration assembly described by Wernimont and Hopkinson (IO) using a deadstop end point. The Karl Fischer reagent was obtained from HartmanLeddon Co., Philadelphia, Pa., and was standardized against sodium tar0.05Y0 HnO. Moisture trate, 15.66 contents of samples were determined hy direct titration with Karl Fischer reagent; methanol was used as a solvent and was titrated to an end point prior to addition of the sample. The coulometric cell is the standard sensing element from a Consolidated Electrodynamics Corp. moisture monitor, an instrument based on the work of Keidel (8). This cell is shown in Figure 1 (prior to being potted in plastic) and provides a 75-em. path length to ensure complete electrolysis of the moisture in the gas stream. It consists of two platinum-wire electrodes 16 feet long which have been wound in a double helix, evenly spaced a t 0.003 inch, and encased in a glass tube. An interelectrade coating of essentially anhydrous phosphorus pentoxide (with a measured
*
Figure 1.
Electrolytic cell
resistivity of approximately 10'0 ohmem.) is obtained by filling the cell with phosphoric acid, allowing to drain, applying voltage to the syatem, and electrolyzing to dryness. The actual geometrical details of the platinumphosphorus pentoxide electrode system are more clearly shown in Figure 2. RESULTS AND DISCUSSION
The use of a coulometric system for the analysis of moisture in gases was first proposed by Keidel (8). The system is based on the principle of water absorption by a phosphorus pentoxide matrix between trvo closely spaced electrodes (Figure 2). Becmse of the high resistivity of anhydrous phosphorus pentoxide, a potential npplied
between the two electrodes results in current, flows which are less than a few tenths of 1 pa, Water absorbed in the phosphorus pentoxide is converted t o various phosphoric acid ionic species and the interelectrode matrix beconies conducting: PzOs
+ €I20
+
2HPOa
(1)
nPtAllNUM
LLECTRODES
ctass'
Figure 2.
Electrolytic sensing element
An applied potential effects accelerated ionic diffusion and subsequent electrol3 4 s of the absorbed water, converting it to oxygen and hydrogen a t the anode and cathode, respectively. 2 HPOs + Hz
+ '/z
0 2
+ PzOj
(2)
Thus, a provision is made for the continuous electrolytic regeneration of phosphorus pentoxide. Completely anhydrous phosphorus pentoxide is, therefore, always present in the cell to ensure complete removal of the water in the sample gas. By devising a system which provides a constant flow of gas over the aforementioned electrode assembly-e.g., 100 ml. per minute--n fixed number of molcs of gas is introduced per unit time. As the water is absorbed from the ga.s, a current flows between the electrodes which is proportional to the rate of water absorption in the cell; 0.5 mole of water is electrolyzed per faraday of electricity. I n other words, the milligrams of water per mole of gas, or per liter of gas, are directly related t o the electrolysis current that passes between the electrodes. Because liquids cannot be passed directly through the moisture cell without washing the phosphorus pentoxide matrix out of the cell in a relatively short time, moisture in liquid samples can best be determined by stripping the water from the liquid sample with a dry inert gas through use of a suitable stripping column. The stripping tower uses a falling film of liquid and a countercurrent stream of dry nitrogen; a diagram of this unit is shown in Figure 3. The sample is introduced by a metering pump a t the top of the tower and stripped by the nit'rogen stream which enters at the base of the tower. The gas, guided by a concentric cylindrical baffle (occupying 40y0 or more of the column void), passes up the threaded inner wall of the toner in intimate contact with the sample. The nitrogen from the stripping tower is then passed through the electrolytic cell. The stripping tower is also provided with a means for heating to aid in stripping viscous liquids-e.g., quantitative stripping of poly(propy1ene glycol) necessitated a column temperature of 100" C . The sample metering system consists of a rotary cavity-type metering pump driven by a synchronous motor permitting the. introduction of liquid samples a t a fixed and highly reproducible rate (0.215 ml. per minute). The liguid within the cavities is driven into the stripping column by dry nitrogen
A schematic flow di:tgrltni of the complete instrument with a bypLlss loop for the excess liquid is shown in Figure 4. To prevent carrj.-ovcr of particulate liquid, simple solvent stripper baffle is provided a t the top of the stripper column. The stripping ton-w, elcctro-
-WMP VALVE
STRIPPING
u 1i CAS
Figure 3.
trogen is dry, its rate of flow is not critical and may be adjusted to strip the liquid completely of its moisture; a rate of 200 to 400 ml. per minute was adequate for those liquid samples employed in the present study. This agrees with the theoretical and experimental work of Krynitsky and Garrett (1, C), who have shown that a five-throre'icalplate column would remove moisture from jet fuels down to less than 1 p.p.m. if a gas-fuel flow ratio of a t least 6 were used. The flow ratio used in the electrolytic moisture analyzer is a t least 1000 volumes of nitrogen pcr volume of liquid; the stripping tower of the present design possesses an efficiency exceeding five theoretical plates. These two factors dictate that the stripping tower should he lOOyo efficient in its removal of moisture from liquids to a t least 1 p.p.m.
Stripping column
LIQUID I N FILTER
Figure 4. schematic
Sample flow
DRYIN6 CELL
n
6
A
S
5
e
I
1
LIRUI(I_OuT
1 COLUMN-ORAln VALVE
E A S E
gas (cf. Figures 3 and 4). This metering pump was developed specifically for this purpose in the authors' laboratories. Because the sample system intr0ducc.s material at a fixed volume per unit time and the current represents grams per unit time, the indicated current for a sample is directly proportional to the moisture content:
.
2 =
grams HzO faraday X liter + equivalent wt. liters -(in amperes) ( 3 ) sec.
For the sample flow rate uscd in this work (0.215 ml. per minute), i = 38.5 ua. per p.p.m. by volume ( 4 ) Currents of this magnitude are conveniently measured by a 0- to 100-ga. meter (xith shunts) in series with the cell and a direct current source. The nitrogen gas, prior to passagc through the stripper, is passed through a separate electrolytic cell to dry it to less than 0.1 p.p.m. Because the ni-
lytic cells, metering pump, and sample system are shown in Figure 5 . Because the sample pump is, in effect, a constant displacement pump, the rate of sample flow into the instrument is independent of the viscosity of
Table
I.
Moisture Analyses
Moisture content, .p.p.m. (nip. liter) Coulometric Moisture Liquid Analyzer Jet fuel JP-4 72 f 1 (av. of 5 rung) 33 JP-4 JP-4 (H20 120 added) JP-5
97
Heavy 40 lubricating oil Insulating oil 20 Xylene 90 Stoddard solvent 210 Benzene 278 Poly(propy1ene 300
HtO per
Karl Fischer 74 i 3
38 f 3 118 f 5 102 f 5 40 i 3 23 f 3 9 l f 4 210 f 7 272 f 8 316 f 10
glycol)
VOL. 31,
NO. 12, DECEMBER 1959
2049
To check the reliabilitv and accwacv of the instrument, a s&ies of organk liquids has been analyzed with the instrument and the results have been compared with Karl Fischer titrations. The samples for the Karl Fischer titration were obtained with a hypodermic syringe directly ahead of the metering pump in the sample loop. The sample was then transferred from the syringe into a closed titration cell to exclude atmospheric moisture. The liquids used were selected to give a wide range of viscosity a well as a fairly broad moisture content range. Table I summarizes the results of the analyses. The values given for the coulometric moisture analyzer are the steady readings obtained after equilibrium and are stable to within a few tenths of 1 D.o.m. for an hour or more: the values & ; the Karl Fischer determinations represent averages of several separate determinations made on samples withdrawn as mentioned above during the course of a run. The precision shown for the Karl Fischer determinations represents average deviations. The agreement between the two methods is remarkably good and the comparisons are as good or better than the accuracy to be expected from the Karl Fischer titration. The efficiency of the stripper is demonstrated by the data for poly(propylene glycol), which would be expected to he one of the most difficult from which to strip moisture. However, the agreement between Karl Fischer titration and the coulometric analyzer indicates that the stripping column is efficient even for this material. Although samples have only been analyzed up to several hundred parts per million, the upper limit would seem to he governed by the design of the stripping column and metering pump. For very high moisture contents larger and more efficient stripping
20%
ANALYTICAL CHEMISTRY
I
-
.. .,..
Figure 5. analyzer
Coulometric moisture
I
towen could he employed; a metering pump with smaller volume could also he used. Although water contents below 20 p.p.m. can he reproducibly determined by the coulometric method, limitations of the Karl Fischer method prohibited quantitative comparison. Measurements on exhaustively desiccated licluids show a background current corresponding to a fract,i& of 1 p.p.m. Quantitative determinations of water contents to 1 p.p.m. should, therefore, be possible. The system described should be applicable to any liquid which does not give off vapors that would react with phosphorus pentoxide. Because of the relatively few liquids which can be vaporized, and subsequently react with phosphorus pentoxide, the instrument is capable of almost universal applicsr hility. For those systems which might interfere i t should he possible to insert a scrubber before the gas entered the moisture cell to eliminate the interfering substance. The coulometric technique for analyzing liquids for moisture provides a method for continuous on-stream analysis which is not easily realized with previous methods. The use of a phosphorus pentoxide matrix for water absorption provides an extremely versatile system which is applicable to almost all liquids. The instrument provides a means for analyzing samples over a wide r p q e of moisture contents. Finally, because the coulometric an-
equipment. LITERATURE CITED
(1) Garrett, W. D., Krynitsky, J. A., “Determination of Water in Jet Fuels
and Hydrocarbons,” Naval Research Laboratory, Washington, D. C., NRL Rept. 4997 (September 1957). (2) Keidel, F. A,, ANAL. CAEM.31, 2043 (1959). (3) Kelley, M. T., Stelzner, R. W., Laing, W. R, Fisher, D. J., Ibid., 31, 221 (1959).
(4),Krynitsky, J. A., “Theoretical Conaderstions Governing the Dehydration of Fuels by Gas Blowing,” Naval RB search Laboratory, Washington, D. C., NRL Memo. Rept. 684 (February
1957). (5) LeTourneau, R. L., ANAL. CHEM. 29 684 (1957). (6) Lveland, J. W., Wehater, T. B.,
Hablitzel, C. P., Reed, G. W., Zbid.,
30, 1316 (1958). (7) Meyer, A. S., Jr., Boyd, C. M., Ibid., 31,215 (1959). ( 8 ) Mitchell, J., Jr., Smith, D. M.,
“Aquametry,” Interscience, New York,
1O”Q
RECEIVEDfor review June 15, 1959. Acee ted September 14, 1959. Pittsburg{ Conference on Analytical Chemistry and Applied Spectroscopy, March 1959.
Spectrochemical Determination of Trace Impurities in Plutonium Nitrate Solutions -Correction I n the article on “Spectrochemical Determination of Trace Impurities in Plutonium Nitrate Solutions.” [ANAL. CHEM.31, 1643 (1959)], on page 1644 under Preparation of Samples the exposure in seconds should read “34” instead of “4.”
EDWARD VEJTODA
