DIVISION S-l NOTES

STRIPPER FOR BUBBLE-FREE TENSIOMETRY

EDWARD E. MILLER* AND AMIR SALEHZADEH

AbstractUnwanted formation of air bubbles in the liquid water inside soil

water tensiometers has always been a major nuisance afflicting thissimple, fast, and accurate device. The stripper, a plastic wall (usuallyjust a tubing wall) between the tensiometer water and a wet vacuum(containing some droplets of liquid water), strips any dissolved air outof this water. Bubbles cannot grow, they will shrink and disappear,unless the water pressure inside the tensiometer is less than the vaporpressure of water (at the ambient temperature). Choices of desirableplastic materials and other considerations are based on 7 yr of devel-opment and employment of this method.

FOR MEASUREMENT of soil water pressure from 0 to75 kPa (3/4 bar), tensiometers that use modern

sensors are precise and quick to respond. Their Achillesheel is the eventual development of air bubbles withintheir water-filled cavities. Any small bubble nucleatedwithin these cavities grows with time, pushing waterout through the tensiometer membrane into the soil.This action perturbs the tensiometer reading. Crudelyspeaking, when bubbles are present the sensor readingrepresents a compromise between the soil water pres-sure and the equilibrium pressure of dissolved air inthe tensiometer water. If tensiometers are filled withfreshly deaired water, they must be refilled at frequentintervals.

Stripper DevelopmentWhen demonstrating a classical field type of tensiometer

(Hg sucked up into a vertical glass tube) in the laboratory, oneof the authors noticed that each week the long plastic tubeconnecting the glass manometer tube to the porous cup in thesoil body would be found loaded with small air bubbles, evenwhen the initial filling had been with deaired water. Investi-gation proved that this air could not have come from the porouscup; it represented molecular diffusion from the surroundingatmosphere through the plastic walls of the connecting tube.Inquiry disclosed that some experimenters already understoodthis problem and had either selected plastics with relativelylow rates of diffusion, or used metal and glass tubing through-out. Even with metal tubing, diffusion of air through the water-saturated tensiometer membranes would eventually lead to for-mation of bubbles that would destroy the reliability of themeasurements.

Thinking "If you can't lick 'em, join 'em," we tried re-versing the inward diffusion of air molecules from atmosphereto tensiometer water by replacing atmosphere with vacuum. Aconvection pump forced the tensiometer water to flow past theinside of the membrane that separated the tensiometer water

E.E. Miller, 1150 University Ave., Sterling Hall, Univ. of Wis-consin, Madison, WI 53706-1390; and Suite 180, Engineering 1-A, Univ. of Michigan, Ann Arbor, MI 48109-2125. Received 20Feb. 1992. *Corresponding author.Published in Soil Sci. Soc. Am. J. 57:1470-1473 (1993).

from the soil, then through a short length of plastic tubingsurrounded by a vacuum, then back through the pump. (Theconvection pump was simply a short U-tube, warmed on therising branch and correspondingly cooled on the descendingbranch.) Early experiments with this rather cumbersome ver-sion of what we now term a stripper (a device to removedissolved air) showed that it effectively removed almost alldissolved air from the tensiometer water.

If a stripper holds down the dissolved air in a tensiometerto some equilibrium (absolute) air pressure, pa, air bubbleswill always shrink whenever the absolute liquid water pressurein the tensiometer is greater than (pa + pvp) where pvp is thevapor pressure of water, say 2.5 kPa (1/40 bar) at room tem-perature. (As a numerical example, if dissolved air pressure isthus reduced to 5 kPa [2/40 bar], no bubble can possibly formabove 7.5 kPa [3/40 bar] absolute, or a suction of 92.5 kPa[37/40 bar].) We concluded that, with stripping, it should bepossible to extend the reliable pressure range of the water intensiometers down nearly to the vapor pressure of water, i.e.,to virtually 100 kPa (1 bar) of suction.

There were practical problems. Temperature effects on theexternal convection pumps produced erratic variations of ten-siometer readings, the water enclosed by the pump and strippertube acting like the liquid in a thermometer bulb. Salehzadeh(1990), in his thesis application, required five tensiometers,spaced a fraction of a centimeter apart in a 2-cm-high soilcolumn. This crowding made the placement of unwieldy cir-culating pumps virtually impossible. We decided to eliminatecirculating pumps altogether by switching the vacuum to theinside of a plastic stripping tube, which was then inserted intothe tensiometer water, in contact with the tensiometer mem-brane (We had feared at first that bubbles might form imme-diately inside the membrane if the water was not stirred pastit, but since these unstirred strippers have worked perfectlysince 1985, we have concluded that the finite thickness of thetensiometer membrane has been the saving feature, the dis-solved air concentration at the inside of the membrane beingreduced as it diffuses its way through the membrane. No bub-bles could form within the membrane thickness because themembrane pores are too small.) Scale modeling of diffusioninto plastic tubes that had the same ratio of outside to insidediameters showed that the rate of stripping per unit lengthwould be independent of absolute size (except for the effectof diffusion through the concentric layers of water outside thestripper tube). Conveniently, therefore, such internal-vacuumstripping tubes can be made very small.

Molecules of air diffuse through a given plastic into thevacuum, so why not molecules of water? Whatever liquid wateris evaporated from the tensiometer into the vacuum must bereplenished by soil water entering the tensiometer through themembrane. This inward flow rate of water through the mem-brane generates a pressure drop across the membrane that maybe of sufficient size to significantly perturb the tensiometerreading. Although diffusion of air is driven by the differencein effective air pressure across the plastic wall, diffusion ofliquid water is different, being driven by the difference — notof the liquid water pressure — but of the effective water vaporpressure. Therefore, when some drops of liquid water are in-troduced inside the stripper tube, instead of a true vacuum, thenormal vapor pressure of water will exist inside the tubing. Itis easy to show that vapor pressure depends only slightly onliquid pressure. By deliberately maintaining some droplets ofliquid water inside the vacuum — what we term a wet vacuum— we reduce the rate of diffusion of water by several ordersof magnitude, making this rate entirely negligible in practice.

The use of a wet vacuum, plus placement of the vacuum on

1470

NOTES 1471

used in the stripper (from listing in Huang et al., 1974).

Name ofplasticPolycarbonatePolyethylene

Density 0.915Density 0.918Density 0.928

PVCSilicone rubberTeflon

DiffusibilityfGas used

ZZ

OZ

ZZ

Z

Gas0.30

1.000.70

24.2150.

1.4

Water vapor1400

113

116-1231119

8.4

Tensiometer tube

t Flux density measure: [1010 (cm3 STP) (cm2 area)-' s~'] divided bypressure gradient measure, [cm Hg (cm thickness)-1], where (cm3

STP) treats vapor pressure as a perfect gas, converting its volume tostandard temperature and pressure.

the inside of the stripping tube, have been the real keys to oursuccess with stripping tensiometry.

MaterialsMolecular diffusion of gases and liquids has interested many

investigators. Hwang et al. (1974) published an extensive tab-ulation of such measurements. The five most relevant mate-rials, extracted from their lengthy listing, are shown in Table1.

Only a few of the plastics listed by Hwang et al. (1974) arecommercially available as tubing of sizes suitable for our pur-poses. Further, not all plastics are sufficiently rigid for use intensiometers requiring a fast response time. Teflon "spa-ghetti" tubing was an early contender — somewhat better thanpolyethylene. Unfortunately, it is difficult to form a reliablytight, dead-end closure on Teflon tubing.

Salehzadeh (1990) used tensiometers made from 4.8-mm(3/16-in.) o.d. tubing having a disk of screen-supported Nu-clepore membrane (Nuclepore Corp., Pleasonton, CA) epoxyglued to the end to serve as the soil-contact membrane (Fig.

rXI —II"02

m3.18 mm (1 /8") ID S.S. Tubing

RTV Coating

I"!; toP$S«S«SSS«SS««5S«J«SSS?SSS«S

1.59 mm (1/16") OD Porous S.S.

mm

Wet" Vacuum /

s Water droplets

Fig. 1. Business end of the version of stripper tensiometer usedin Salehzadeh's thesis apparatus.

Fig. 2. The sort of jig that was developed for gluing theNuclepore membranes onto the 4.8-mm (3/16-in.) o.d.tensiometer.

1). To fit this design, we finally selected for the strippingplastic a very soft silicone rubber compound, RTV630 (Gen-eral Electric Co., Silicone Products Div., RTV Products Dep.,Waterford, NY). Their RTV103 and RTV104 are also usable,though even softer. This RTV material is a two-componentproduct that is initially moderately pourable. It sets up into arubberlike material in = 1 h at 70 °C. Because of its mechanicalsoftness, it was coated in its liquid stage into the coarse outerpores of 1.6-mm (1/16-in.) o.d. porous stainless steel tubing,the combination constituting the stripper. (A short broken-endsample of the porous tubing was donated through the courtesyof S.S.I. Technologies, Janesville, WI. Unbroken, these areused in the nuclear industry. At the time, unbroken pieces wereunbelievably expensive. We cut our single piece into several25.4-mm [1-in.] lengths, one for each tensiometer, which servedadmirably.) Before being coated with RTV, the porous andsolid tubings were mechanically connected with epoxy,strengthened by a short bit of hypodermic needle inserted intoboth.

A magnified diagrammatic view of the business end of thefinished tensiometer as used by Salehzadeh is shown in Fig.1, while Fig. 2 shows the sort of jig that was developed forgluing the Nuclepore membrane onto the tensiometer, the ex-cess epoxy being trimmed off later with a lathe.

1472 SOIL SCI. SOC. AM. J., VOL. 57, NOVEMBER-DECEMBER 1993

—— Membrane 'Stripper'assembly , tubing

To pressure transducer

To wet vacuum

- Tensiometertube

Glass T

Solid 1/16SS tubing

Glass T and 3/16swage lock epoxyed

Fig. 3. Finished assembly plan for Salehzadeh's version ofstripper tensiometer.

Figure 3 shows the assembled tensiometer setup. Note thatseparate leads are extracted for the pressure transducer and thewet vacuum. A glass T provided rigidity without any addeddiffusion of gas or water to or from the atmosphere.

Figure 4 shows, diagrammatically, the method used for test-ing the RTV-coated porous stainless steel stripper for diffusionof either air or liquid water, the enclosure being filled, inadvance, with either dry air or liquid water. For the measure-ment of air diffusion, the height of the Hg pool is set to provideapproximately ambient air pressure. Using positive air pressureto drive the diffusion, we measured the rate of diffusion of airthrough the RTV-coated tubing as shown in Fig. 5. The resultsare expressed as standard-temperature-and-pressure (STP) vol-ume for a driving pressure difference of 100 kPa (1 bar).

These data cannot be converted to the Q values of Table 1because of the unknown effective thickness of the coating andthe unknown porosity and tortuosity of the porous tubing. Togive a rough operational comparison, for 2.54 cm (1 in.) of

Stripper (RTV coated porous S.S.)

To vacuum (dry/wet) Glass Enclosure

-t-X

Capillary

Mercury

Fig. 4. Diagrammatic view of scheme for measuring the rateof diffusion of either air or liquid water.

0.1-

0.08 —

0.06 —ca

to 0.04 —Otds3 0.02-o

\

RTV-coated porous SS25.4 mm (l") long, 1.59 mm (1/16") 0 DTemperature 21 °C

0I

12I

16 20TIME, (min)

Fig. 5. Diffusion of air through RTV-coated porous stainlesssteel tubing 2.54 cm (1 in.) long, 1.6 mm (1/16 in.) o.d.

1.6-mm (1/16-in.) o.d. Teflon tubing with a wall thickness of0.4 mm (0.015 in.), the calculated corresponding slope is es-timated as 0.24 cm3 of STP air h ~ ' m~' . In contrast, the slopein Fig. 5 for the RTV-coated tubing is 9.7 cm3 of STP air h-1

m-1, i.e., this same-size RTV stripper works 40 times faster.Figure 6 gives experimental confirmation of the effective

avoidance of diffusion of liquid water when a wet vacuum isemployed. Warning: if the drops of liquid water in the vacuumare not locally positioned inside the stripper tube itself, thelocation of these drops must be maintained at the same tem-perature as the stripper tube, or slightly higher. If the dropsare at a position of lower temperature, the vapor pressure feltwithin the stripper tube will be correspondingly lower and thewater loss by diffusion may become noticeable. (It may notbe necessary to introduce water drops if no vapor is beingremoved by the vacuum system. The first water to diffusethrough the stripper walls will constitute a layer of liquid waterto block further diffusion of water.)

We had originally used ordinary 6.4-mm (1/4-in.) o.d. poly-ethylene tubing for various external sensor-to-tensiometer-membrane connections. This external length of plastic tubingdecreased the rigidity of the system, lengthening its responsetime. When we took out the last remaining 51-mm (2-in.)length of the polyethylene tubing, the response time was short-ened fourfold.

0.01-

0.008 —

0.006 —

CAta

a

0.004 —

0.002 —

-0.002-

RTV-coated porous SS25.4 mm (1") long, 1.59 mm (1/16") ODTemperature 21 C

r-|-r ,

200 5 10 15 20 25TIME, (h)

Fig. 6. Diffusion of liquid water into wet (compared with dry)vacuum.

NOTES 1473

%Ua,O.

H<

-160 -

-20020000 40000 60000 80000 100000 120000 140000

TIME (s)

Fig. 7. Smooth experimental performance, in soil, of strippertensiometer down to —180 kPa ( — 1.8 bar) using thepressurized soil column.

For a system in which a very fast response time is not im-portant, there is still an important reason for eliminating ex-ternal plastic tubing. In our pressurized system (employed toattain suctions of much more than 100 kPa [1 bar], we neverallowed the liquid water pressure to go below the pressure ofthe atmosphere (for reasons of convenient control), which meantthat no air bubbles could have formed from room air diffusingin through the tubing. Water could have diffused out to theroom air, however, because relative humidity was usually farless than 100%. With this system, bubbles could form frominward diffusion of air through any connecting plastic tubing,whenever the soil water reaches negative gauge pressures. The-oretically, even without any external plastic tubing, room aircould eventually diffuse inward through the plastic body usedin some types of sensors. This problem, should it every arise,could be circumvented by enclosing the sensor body in a dif-fusion-proof envelope or coating.

Our pressurized system extends the range of tensiometermeasurement well beyond 100 kPa (1 bar) of suction. Suchdeliberate elevation of soil air pressure generates a correspond-ingly higher driving force for diffusion of soil air through themembrane into the tensiometer. The higher air-entry value thusrequired by elevated soil air pressure for our membranes didnot change their porosity and hence it changed neither thediffusive conduction of air through the water-saturated poresinside the membranes nor the parallel diffusion of air throughthe plastic regions of the membrane.

DiscussionMost of the work described here was directed specif-

ically to the difficult problems of tensiometry for Saleh-zadeh's thesis apparatus using pressurized soil air. Whenused with a thin contact bridge between soil and mem-brane made of wetable diatomaceous earth, these ten-siometers performed satisfactorily down to moderate soildryness. When we found that a thin layer of fine, het-erodisperse SiO2 beads (copiously supplied to us by R.D.Miller and available from the authors on request) provedconsiderably more effective for the remaining range ofdryness, this bridge material became our preferred choice.Figure 7 shows the smooth experimental performance(vs. time) of a stripper tensiometer in actual soil. Note

that it was still working beautifully down to the pre-planned turn-around suction of about 180 kPa (1.8 bars).

There are a host of less demanding applications forwhich stripper tensiometry promises to be practical andeconomical. In the press of using the pressurized soilsystems, none of the following suggestions for the useof stripper tensiometers in ordinary ambient air systemswas actually tested, but we see no reason why they shouldnot work analogously. In the field, a practical strippingvacuum could be supplied with a simple vacuum storagetank, pumped down at rather infrequent intervals asneeded. For field-scale systems, ordinary plastic tubingcould be used as the strippers, perhaps inserted full lengthinside somewhat larger plastic connecting tubing to re-move atmospheric air as fast as it diffuses in, thus per-mitting the convenient use of plastic tubing for makingconnections. Porous stainless steel tubes that are largerand far cheaper than our exotic 1.6-mm (1/16-in.) o.d.material are available that could be coated with RTV.Thus, even old-fashioned Hg-rise systems could be madeaccurate and long lasting, though degraded by their in-herently sluggish response times. Routine laboratory ex-periments could be "stripperized" to give reliablemeasurements down to nearly 100 kPa (1 bar) of suctionwith no worry about bubble appearance.

Note that published measurements of the diffusionproperties of plastic tubing should be viewed only asguidelines because of frequent marketing changes incommercial tubing. Two batches with the same com-mercial description could differ significantly in air dif-fusibility, so, when optimum performance is needed,experimenters should make their own measurements (seeAppendix) on the actual stripping material they have onhand. This is not difficult, as shown in Fig. 4, providedthat attention is paid to the thermometer-bulb effect andthe usual small correction for curved-meniscus suction.

Appendix: Measurement of Pressureof Dissolved Air

If a rigidly enclosed body of liquid water contains, in (unst-able) equilibrium, a single air bubble of large size — say 1-mm radius — the liquid water pressure then constitutes a roughway of describing the concentration of dissolved air. (For fullydeaired water, this absolute pressure would essentially be thevapor pressure of water, while for air-saturated water [keptlong in the air] it would be one atmosphere.) As ultimate re-finements, one might wish to subtract the vapor pressure ofwater and add the pressure increase inside the bubble causedby surface tension (2o/r) — about 0.14 kPa (1.5 cm of water)for a 1-mm bubble. In our experiments, we controlled thepressure, not the volume, monitoring the rate of expansion orshrinkage of a small introduced bubble with a reticled mag-nifier, and adjusting the water pressure as needed to attain(unstable) equilibrium. With this method, we not only satisfiedourselves that our earliest stripper was indeed removing dis-solved air, but we also measured its rate of stripping as afunction of the pressure of the remaining dissolved air.