Acknowledgement

This invention was made with government support under Grant No. 94-33610-0898 awarded by the U.S. Department of Agriculture, under the Small Business Innovation Research Program of the United States of America. Technical Field The present invention relates to soil moisture sensing devices and more particularly to an osmotic device for measuring soil moisture available to plants, whereby horticultural and irrigation practices may be improved. Background Art

With limited water resources in much of the world, farmers must increasingly adopt conservative irrigation practices. One method of optimizing irrigation practices is by measuring available soil moisture for plant growth, and irrigating in response to soil moisture depletion levels at which plant growth becomes stressed. Plants draw water from most soils essentially by osmosis into rootlets. A portion of the moisture in soil is bound to the soil by various chemical and physical attractions between water and the soil particles. The availability of this water is described as the matric potential of the soil moisture. Total soil water potential is a sum of the matric potential, the pressure potential of loose (fluid) water in the soil, and the osmotic potential of soil water (i.e., the contribution of salts and other osmotic species dissolved in soil water). Ideally, any sensor operating on soil moisture content for irrigation scheduling purposes will sense the soil matric potential, modified by the osmotic potential whenever at significant levels, rather than total soil moisture content. In this way, the sensor will truly monitor the soil moisture actually available to a living plant. The removal of bound water from the soil particles involves expenditure of energy. The free energy of a molecule such as water is determined by forces acting on the molecule. The free energy of water in soil is usually referred to as "water potential," and is expressed in units of pressure such as bars (one bar is equal to 0.1 MPa). Osmotic pressure of a water solution is describable by the same terms of measurement (i.e.,

pressure). Soil matric potential may vary from 0 bar (0 MPa) when fully wet to about -25 bar (-2.5 MPa) near total dryness. Permanent wilting of many plants is experienced at matric potentials approaching or exceeding -15 bar (- 1.5 MPa). For most soils, moisture release curves show that the majority of the available water is held between -0.1 and -5 bar. Two primary research tools for directly measuring soil matric potential are tensiometers and thermocouple psychrometers. Tensiometers are limited to a range of 0 to about -0.8 bar (0 to -.08 MPa). Thermocouple psychrometers are useful over a range of -0.5 to -2.5 MPa. Tensiometers are incapable and thermocouple psychrometers are essentially inapplicable in the range of -0.1 MPa to -0.5 MPa. This is a range of great significance for both horticultural studies and for scheduling of irrigation events. Furthermore, tensiometers often see a degradation in performance due to fouling of a porous ceramic cup component which is immersed in soil. An osmotic device for directly measuring soil matric potential was disclosed in U.S. Pat. 3,455,147 (15 July 1969). U.S. 3,455, 147 disclosed the use of an osmotically active polyethylene glycol solution in water confined within a chamber, one wall of which was a cellulose dialysis membrane whose outward side abutted a finely porous plate in contact with a soil sample. Osmotic absorption of soil moisture through the dialysis membrane into the polymer solution produced a hydraulic pressure in the chamber. Reverse passage of moisture out of the chamber in response to soil dryness also occurred. A means for quantitatively sensing and indicating pressure changes inside the chamber was provided. A commercial embodiment of this device was at one time available under the name

"Aquapot," but was discontinued for reasons believed to center primarily on unreliability in field use. Biofouling of the porous plate and biological degradation of the dialysis membrane may be anticipated to occur during soil immersion of the device. An essentially identical device was disclosed in figure 9 of WO 92/03916, wherein a pliable (reverse osmosis) membrane permeable to water and retentive to sugar and salt solutions was supported on a porous plate mounted to a chamber.

A variety of other techniques for measurement of soil moisture content

emerged since the device of U.S. 3,455, 147. Electrical conductance sensors have been disclosed in U.S. Patents 3,803,570, 3,882,383, 4, 1 22,389, 4,540,936, 4,796,654, 4,929,885, 5,060,859, 5,260,666, 5,424,649, 5,430,384, 5,445, 1 78, and, 5,479, 104, all directed at soil resistivity as a measurable indicator of soil moisture content. All encounter the problems of variable soil resistivity as a function of soil heterogeneity, density, compaction, salt and fertilizer content, stray electrical disturbances, and changes in electrical lead surface resistivities. Modified electrical conductance sensors that measure resistivities of preselected porous media (such as a gypsum block) in equilibrium contact with natural soil are disclosed in U.S. Patents 4,216,789, 4,513,608, 44,531 ,087, 4,561 ,293, and 5, 179,347. These media ameliorate some of the problems of soil variability, compaction and resistivity, but introduce a response lag time. Heat dissipation into soil from sensors have been disclosed in U.S. Patents 4, 197,866 and 4,310,758 as a way of monitoring soil moisture. For similar reasons, sensors measuring air permeation rates through soil or through porous bodies in contact with soil were disclosed in U.S. Patents 44,055,200 and 4,040,436. Expansion and contraction of polyacrylamide gel beads in response to soil moisture levels was disclosed in U. S. Pat. 5,329,081 . Even space satellite measurement of soil moisture content by spectral means has been described in U.S. Pat. 4,310,758, this disclosure also including a discussion of other methods and their shortcomings.

This continuing activity in the field of soil moisture sensors is testament to the long felt need in agriculture for a reliable method of signalling and controlling the schedule of irrigation as a function of soil dryness. This continuing activity is also testament to the fact that known techniques for directly measuring soil matric potential, specifically tensiometers, thermocouple psychrometers and the Aquapot device, have failed to meet the specific needs of farmers for reliability or practicality in a field. In addition, none of these alternative techniques actually measure soil matric potential, in contrast to the aforementioned tensiometer, thermocouple psychrometer and "Aquapot" osmotic device.

It is an object of this invention, therefore, to provide a reliable device

capable of direct measurement of soil matric potential.

It is a further object of this invention to provide a reliable device for, and a means thereby of, measuring soil matric potential in a field as an indicator for timing irrigation events, and to provide a means to measurement of soil matric potential through multiple irrigation cycles.

It is a further object of this invention to provide a reliable device that is resistant to biofouling during long term immersion in soil.

It is a further object of this invention to provide reliable device for and a means to measurement of soil matric potential unattended in a field. It is a further object of this invention to provide a device and means for measurement of soil moisture, wherein actual soil matric potentials may be quantitatively measured and the measurement data recorded or otherwise stored.

These and other objects of the invention will become evident to one of ordinary skill in the art through examination of the following specification, figures, and the claims appended. Disclosure of Invention

It has now been found that an improved device can be made and used which directly measures soil matric potential by an osmotic phenomenon. Specifically, an improved soil moisture sensor has now been developed that employs a rigid, inorganic, permselective membrane forming a portion of the boundary walls defining a chamber filled with an osmotically active fluid (hereinafter referred to as an "osmoticum"), whereby water can transfer between soil and the osmoticum via interchange (i.e., transport) through rigid membrane. Fluid pressure within the chamber varies as a function of the soil moisture level, and more specifically, as a function of soil matric potential, due to this water interchange, and the fluid pressure is quantitatively measured.

A particularly improved soil matric potential sensor has also been developed which comprises a rigid, cylindrical, inorganic, microporous permselective membrane, an enclosed chamber primarily in the form of an annulus or annular space adjacent the inner or lumen surface of the cylindrical membrane, an osmoticum contained in the enclosed annulus, and a means for

measuring the fluid pressure of the osmoticum. In addition, an improved rigid, inorganic, microporous membrane has been found, developed and used in the osmotic device, wherein the soil-contacting surface of the rigid permselective membrane has been first treated with a low temperature gas plasma so as to deposit an ultrathin, organic, moderately hydrophilic, polymeric plasma polymerizate thereon, which renders the membrane more resistant to biofouling.

A preferred embodiment of the invention employs a rigid cylindrical membrane suitably mounted into a generally tubular-shaped device. The rigid cylindrical membrane is preferably a microporous silica glass composition, but may otherwise be a of ceramic or glass/ceramic composition. One or more end members are affixed to this cylindrical membrane so as to define an enclosed chamber for containing an osmoticum. A rod-like insert is preferably located within the lumen space of the cylindrical membrane so as to define an annulus and to thereby increase the surface-to-volume ratio of membrane to osmoticum. The rod-like insert may consist of a protuberance from one or both end members. The osmoticum is typically an aqueous solution of a water-soluble polymer such as polyethylene glycol (PEG) or polyvinyl pyrrolidinone (PVP). The pressure sensing means is typically a pressure transducer, but may be a bourdon tube gauge, a spring gauge, or other such measurement device. The pressure sensing means may be connected to a data recording device or even to a servo-mechanism for actuating an irrigation device, or alternatively may transmit a signal to a remote receiving device for the signal. The rod-like insert may be omitted in the sensor design, but results in a lower surface-to-volume ratio of membrane to osmoticum. In the case of gauge-mounted sensors, the reduced surface-to-volume ratio results in a slower response time to matric potential changes in soils. Such slower response times are still generally acceptable in projected applications of the device. Transducer-mounted sensors maintain a uniformly high response rate nearly independent of the surface-to-volume ratio of membrane to osmoticum, because of the superior sensitivity of electrical transducers to small movements of water into and out of the osmoticum.

This invention provides advantages of being able to measure soil matric potentials directly by an osmotic method, providing a reliable basis for scheduling irrigation events. In addition, this invention provides a stable, nonbiodegradable, permselective membrane means for long term soil immersion. This invention, particularly in its preferred cylindrical, annular- chambered embodiment, is furthermore characterized by a rapidly responsive change to changes in soil matric potential. Reduced biofouling of the membrane's soil-contacting surface is believed to be additionally enhanced by surface treatment with a gas plasma polymerizate wherein the resulting plasma polymerizate is moderately hydrophilic. Brief Description of the Figures

FIG. 1 is a schematic sectional view of a soil moisture sensor.

FIG. 2 is a perspective view of a tubular soil moisture sensor employing a cylindrical membrane. FIG. 3 is a cross-sectional view of the device depicted in FIG. 3.

FIG. 4 is a cross-sectional view of a tubular soil moisture sensor employing an annular chamber.

FIG. 5 is a cross-sectional view of a simplified tubular soil moisture sensor without an annular chamber. Best Mode for Carrying Out the Invention

Reference is now made to FIG. 1 , wherein a soil moisture sensing device in accordance with the present invention is depicted in its most basic form. The device, generally indicated as 10, includes a housing whose walls 11 define an enclosed chamber 12, a portion of the housing consisting also of a rigid, inorganic, permselective membrane 13. The enclosed chamber 12 is filled with an osmoticum, which is defined herein as an osmotically active fluid. The chamber is fitted with at least one outlet port 14 so as to allow connection or insertion of a means 15 for sensing fluid pressure within the enclosed chamber. The chamber of the device may be filled with the osmoticum through this same port 14, or the device may be fitted with at least one other port (not shown) for filling or draining the enclosed chamber.

The materials and construction of the device are preferably governed by a need for overall rigidity, so that changes in fluid pressure within the

device resulting from water movement through the permselective membrane into or out of the osmoticum are not significantly affected by movement or distention of chamber walls (including the membrane). The walls 11 of the chamber (which together with the rigid, inorganic, permselective membrane 13 define the enclosed chamber 12) may be composed of metallic, ceramic or polymeric materials. A steel alloy, particularly stainless steel, is advantageous as a wall material. Aluminum, particularly when anodized on at least its exterior surface, is also useful as a wall material. Polymeric materials are also advantageous, and are preferred relative to metallic materials in reference to ease of construction and processability, along with resistance to corrosion. Rigid polymeric materials, that is, plastics with a favorable combination of high modulus and high tensile strength, are preferred. Examples of such polymers include, but are not limited to, polysulfones, polyimides, polyetherimides, polyacetals, polycarbonates, poly(dimethylphenyleneoxy) and its alloys with polystyrene (Noryl ^® resins, General Electric Company), polyesters, and aromatic polyamides. Rigid resinous materials such as cured epoxy, phenolic or polyurethane resins may also be used, though these are not generally as preferred. These polymeric materials may be strengthened additionally by additives such as, for example, glass fibers or beads, carbon fibers, ceramic fibers, or metallic fibers or leaflets. In addition to rigidity, all such materials of construction that would be in contact with the osmoticum are preferably chosen so as to be compatible with the particular composition of the osmoticum to be employed. Selection of such materials will also be generally governed by their suitability for long term burial in soil, and for their ability to be sealingly affixed with semipermeable membrane and outlet port members.

The rigid permselective membrane 13 is formed from an inorganic material that is preferably permeable to both water and small inorganic salts. This rigid, inorganic, semipermeable membrane is preferably impermeable to polymeric species of 100,000 molecular weight, more preferably impermeable to polymeric species of 10,000 molecular weight, most preferably impermeable to species of 5,000 molecular weight or greater. Semipermeable membranes used in accordance with this invention are

advantageously formed of a finely microporous silica glass, a finely microporous ceramic, or a composite material having a semipermeable separation layer consisting of a finely porous ceramic or glass. A semipermeable glass membrane useful in this regard may be formed by casting a glass from a composition containing Na ₂ O, B ₂ O ₃ , and SiO ₂ , with or without an additional heavy metal oxide such as niobia or germania, then exposing the glass to a temperature of about 400-700 °C to effect phase separation into a silica-rich phase and an alkali-rich phase, then removing the alkali-rich phase by dissolution into hot water. Pores opened by dissolution of the alkali-rich phase are preferably in the 10-100 Angstrom range, more preferably in the 25-75 A range for application in the rigid semipermeable membrane of the sensing device. Semipermeable glasses of this type having pore sizes typically centered at around 40 A have been available commercially from Corning Glass Works, under the trade name Vycor ^® . Glass-ceramic composite membranes employing finely microporous glass separation layers are also useful in the invention. Such membrane compositions are disclosed in U.S. Pat. 4,689, 150, which is herein incorporated by reference. The thickness of the rigid semipermeable membrane may vary in the range from about 0.1 mm to about 8.0 mm, more preferably in the range of about 0.3-3.0 mm, most preferably in the range of about 0.5-2.0 mm. Thinner cross-sections are advantageous for faster water permeation and quicker response of the sensor to changes in soil matric potential. On the other hand, thicker cross-sections are advantageous for durability and resistance to breakage during handling or pressurization, but may be characterized by slower water transport or interchange. These rigid, inorganic, permselective membranes, especially examples having cross-sections at the minimums of the above-described ranges, may optionally be supported on the soil side by screens or grids for additional strength, but this complication may generally be avoided through choice of membrane thicknesses not at the minimums. In the case of Vycor ^® microporous glass tubing, strength may be increased by heat treatment at temperatures below the fusion temperature of the microporous silica network, such as by briefly heating to a temperature between 600 and 1 ,000 degrees Celsius in a muffle furnace.

Particular useful as permselective membranes are microporous glass tubing supplied under the above-mentioned tradename Vycor ^® . These membranes are nonbiodegradable, but attachment and growth of biological matter adherent to the surfaces of these membranes may be experienced. It has been found advantageous to coat the outer surface of these membranes with a moderately hydrophilic polymeric material deposited thereon from a low temperature gas plasma. In particular, surface treatment with a radiofrequency-generated gas plasma containing acrylic acid monomer has been found to provide a modified membrane surface less prone to development of adherent biogrowth. Plasma polymerization of acrylic acid is well known in the art of gas plasma.

The osmoticum consists of an aqueous solution containing one or more osmotically active species. Preferred as osmotically active species are polymeric solutes having molecular weights in the range of about 5,000-50,000, more preferably in the range of about 10,000-25,000.

Examples of such polymers are polyacrylamide, polyethylene glycol, polyvinyl alcohol, polyvinyl pyrrolidinone, and polyethylenimine. Water-soluble copolymers are also useful as osmotically active species, such as, for example, a copolymer of ethylene glycol with bisphenol A diglycidyl ether. Water-soluble polymers having ionic sites may also be employed as the osmotically active species. Examples of such polymers include sodium carboxymethyl cellulose, poly(diallyl dimethyl ammonium chloride), potassium poly(acrylamido-2-methyl-1 -propanesulfonic acid), and sodium polyacrylate. Copolymers such as poly(acrylamide-co-acrylic acid) and their salts may also be employed. Similarly, acid salts of polyethylenimine or other polyamines may be employed as osmotically active polymers having ionic sites. In the case of ionic polymers such as these, the preferred range of molecular weight is greatly expanded on the higher end, that is, molecular weights in excess of 50,000, even in excess of 500,000, may be advantageously used. However, in the choice of such ionic polymers as osmotic species, fluid pressure response to soil matric potential may exhibit some slight variability due to pH effects associated with soil acidity and/or due to some possibility of interchange of salt ions between the osmoticum and the adjacent soil. Such

effects are generally within the capability of a person skilled in the art to observe and determine without undue experimentation.

The concentration of the osmotically active species in the osmoticum may vary between 1 % and 40% by weight. In the case of nonionic polymers, the preferred concentration is in the range of approximately 10% to approximately 30% by weight. Higher molecular weights of the nonionic polymeric solute generally require higher concentrations in order to achieve the same fluid pressure response to soil moisture uptake as would be experienced with lower molecular weight nonionic polymers. In the case of ionic polymers, the preferred concentration is generally in the range of approximately 1 % to approximately 15% by weight. The relative population of ionic sites in the structure of the ionic polymers will have an effect on the preferred concentration in the osmoticum. Higher numbers of ionic sites per equivalent molecular weight among ionic polymeric candidates generally results in lower concentration levels necessary to achieve advantageous fluid pressure responses to soil matric potentials. The choice of polymer, its molecular weight and its concentration may also be governed by solution viscosity. An excessively high osmoticum viscosity presents problems in handling and filling the device chamber. The actual concentration level that is chosen for any one specific solute in the osmoticum will depend on the upper limit of fluid pressure sought in accordance with the soil matric potential sensor and its design. For most soil moisture monitoring applications, a maximum fluid pressure of 1.5 MPa within the enclosed chamber is more than sufficient, such a pressure maximum being easily determinable by immersion of the device in pure water. A preferred range of operational fluid pressures within these devices is generally about 0-1.0 MPa. The minimum fluid pressure reached within the enclosed chamber will, in most cases, be essentially zero, and will correspond to a situation where soil matric potential falls below a value measurable by the osmoticum contained within the device.

Osmotic pressures can generally be estimated in the case of ionic polymers by means of the equation fl = m-R ^Λ T/V wherein π is the osmotic pressure, m equals the number of moles of ions present in a volume V of the

solution, 7 equals the temperature of the solution, and R is a mathematical constant associated with this equation. This same equation is generally applicable to the case of the nonionic polymers, but may require some empirical determinations of solute osmosity. It is convenient, however, to measure the osmotic pressures of such solutions with an osmometer, of which several are commercially available, and this is easily within the capability of one skilled in the art to do.

Other constituents may be present in the osmoticum. In particular, the osmoticum may contain bactericides, slimicides, or other additives for inhibition of biological growth in the enclosed chamber.

The sensor is preferably filled with the osmoticum in a manner that does not leave air bubbles entrapped within the enclosed chamber. Filling may be achieved by one of several methods or their combinations, such as vacuum-fill, pressure-fill, displacement of air with a water soluble gas before filling with the aqueous osmoticum, or centrifugation of the sensor while wholly immersed in osmoticum. Centrifugation works particularly well with sensors of short to moderate length.

The combination of the wall materials and the rigid semipermeable membrane that form the housing defining the enclosed chamber, along with any seals and fittings, are advantageously chosen so as to withstand an internal fluid pressure of at least 10 bar, and preferably at least 15 bar, for a length of time of at least one day, preferably at least one month, more preferably at least one agricultural growing season. Most preferably, the combination of materials chosen in this regard will withstand an internal fluid pressure of at least 15 bar for a plurality of years. In addition, this combination of materials is preferably quite rigid to the extent that any net amount of water osmotically imbibed from soil adjacent to the device through the semipermeable membrane into the osmoticum results in a corresponding pressure increase in the enclosed chamber, and that any net amount of water released from the osmoticum into soil adjacent to the device results in a corresponding pressure decrease in the enclosed chamber. In an ideal device constructed in accordance with this invention, any water osmotically imbibed into the osmoticum will result directly in a predictable and reproducible

increase in fluid pressure within the enclosed chamber, and any water released from the osmoticum into the adjacent soil will result directly in a predictable and reproducible decrease in fluid pressure within the enclosed chamber. The outlet ports from the enclosed chamber are preferably located on the nonmembrane wall portion of the housing so as not to weaken the membrane member. The sensing of fluid pressure within the chamber is conveniently done by attaching a pressure sensing means to an outlet port or through an outlet port into the osmoticum, or otherwise bringing such means into fluid contact with the osmoticum by way of a pressure-tight connection. Such a pressure sensing means may consist, for example, of a strain gauge pressure transducer, of a bourdon tube device fitted with a dial gauge, or of a spring-loaded device. Pressure sensing means that are operable in the range of approximately 0-1.5 MPa are preferable in most applications. The pressure sensing means may further be connected to a data logging device for recording of pressure levels on a continuous basis. Alternatively, the pressure sensing means may be connected via a servo-mechanism to another device or system such as an irrigation system. Alternatively, the means for sensing fluid pressure may be connected to a means for transmitting a signal to a remote receiving device such as by a wireless mode of transmission.

Reference is now made to FIG. 2 showing a schematic depiction of a preferred embodiment of the invention. A soil matric potential sensor generally indicated as 20 is shown to include a cylindrical rigid semipermeable membrane 21 having attached at one end a first end member 22 and at the opposite end a second end member 23. The first end member 22 is generally rod-shaped and contains an internal conduit that is accessed by a fitting 24 for attachment of a pressure sensing means (not shown). The second end member 23 also contains an internal conduit, which is accessed by another fitting 25. A cross-section of this device is depicted in FIG. 3, wherein a chamber 26 is shown, enclosed by the cylindrical membrane 21, the first end member 22, and the second end member 23. A sealant 30 is shown for purpose of sealingly bonding the cylindrical membrane 21 to the two end members 22,23. Attachment may be effected by mechanical means rather

than by a chemical sealant, though use of the sealant 30 is generally preferred. Shown in cross-section in the first end member 22 is an internal conduit 27, which makes contact with a threaded insert portion of the fitting 24. This fitting is herein depicted as a capillary fitting such as commonly used in high pressure liquid chromatography. Such a fitting is convenient and advantageous for attaching a pressure sensing means thereto. The second end member 23 is shown in cross-section to contain an internal conduit 28, which makes contact with a threaded portion of a fitting 25, depicted herein as a luer-type fitting. Such a fitting is convenient and advantageous for attaching a syringe for filling of the chamber 26, and for attachment of a plug 29 when the chamber 26 is filled with osmoticum.

In the use of this device, one or both conduits are open as the chamber is filled with an osmoticum. The chamber is filled in such a way as to avoid presence of residual air bubbles therewithin. One end is then connected to a means for sensing fluid pressure within the chamber and the other end is plugged. In operation, the closed device is then placed into contact with a soil, and the fluid pressure of the osmoticum within the chamber is measured as an indication of soil matric potential.

In FIG. 3, the enclosed chamber 26 is depicted as being an open space within the device, with the presumption that it be entirely filled with the osmoticum in operation. A portion of this open space may be filled with an inert insert to displace a corresponding portion of the osmoticum that would otherwise be present. In fact, it is advantageous to displace some of the osmoticum by placement of inert beads, rods or other solid bodies within the enclosed chamber, in that this increases the ratio of membrane surface to osmoticum volume. The sensitivity and response time of the device to soil matric potential is enhanced by so increasing the me brane- surface/osmoticum-volume ratio.

Reference is now made to FIG. 4, which shows a schematic cross- section of a sensor device generally designated as 40. This design shows a rod-shaped member 41 encased along a middle portion of its length by a rigid cylindrical semipermeable membrane 21, these two members defining an annular space 42 between them. The rod-shaped member 41 is preferably

axially centered within the lumen of the cylindrical membrane 21, such as by means of spacers 43 situated at each end of the annular space 42. This is conveniently achieved with O-rings, though other means for maintaining the annular spacing may be employed. The rod-shaped member contains an internal conduit 44 extending from one of its ends to a location adjacent the annular space 42. An aperture 45 is provided whereby fluid communication is established between the annular space 42 and the internal conduit 44. The cylindrical membrane 21 is sealed at both ends to the rod-shaped member 41 by means of a sealant 46. An endcap 47 is attached at an end of the rod-shaped member 41 opposite form the end containing the internal conduit 44. This endcap 47 may be tapered, as depicted in FIG. 4, for easy insertion into soil. The rod-shaped member 41 and the endcap 47 may optionally consist of a single member, such as might be machined from rod stock or injection-molded as a single piece. A collar 48 is optionally attached to the rod-shaped member 41 at the opposite end of the cylindrical membrane 21 , and is useful for defining an indentation for application of the sealant 46 to that end of the cylindrical membrane 21. The sealant 46 may be applied at or near the two ends of the cylindrical membrane 21, but is preferably applied and located at the ends thereof, rather than near the ends, that is, offset inwardly. The rod-shaped member 41 extends beyond the collar such that this projection provides a platform for attachment of a threaded port 49. The annular space 42 constitutes an enclosed chamber which can be filled with an osmoticum through the threaded port 49, using evacuation if needed to remove trapped air bubbles from the chamber. A fluid pressure sensing means is then conveniently attached to the device through the threaded port 49.

Reference is now made to FIG. 5, which shows a schematic cross- section of an alternate embodiment with a simplified design. This sensor device, generally designated as 50, consists of a cylindrical membrane 51 which is sealed at its soil-insertion end by means of a tip 52 formed by embedding an end of the rigid, cylindrical membrane into a curable potting compound, such as an epoxy or polyurethane resin composition, and curing the resin in place on the membrane end. The opposite end of the cylindrical

membrane 51 is sealed to a hollow shaft 53 which extends from the cylindrical membrane, and which may have a tapered or machined section at its opposite end suitable for attachment of a threaded port or pressure sensing means. The hollow shaft 53 may include a hollowed-out end for direct insertion and attachment of the cylindrical glass membrane 51 , particularly if such a hollowed-out end can be conveniently provided in an injection-molded shaft. But for hollow shafts that would instead be machined to receive the cylindrical glass membrane, it has been found advantageous to attach the membrane 51 to an intermediate connector such as an injection- molded "male" luer lock adapter 54, then assemble the luer lock adapter 54 to the hollow shaft 53. The luer lock threads provide a mechanical means in addition to surface bonding forces for maintaining a tightly adherent adhesive bond (such as with a cured epoxy or poyurethane resin), that will contain the high fluid pressures to be encountered within the device. The microporous nature of the rigid inorganic membrane naturally promotes strong adhesive bonding between the membrane and any adhesive sealant. The "male" extension of the molded adapter 54 is inserted and bonded into the hollow shaft 53, an opening being drilled into the end of the shaft to receive the male extension. Such molded adapters are commercially available, particularly adapters molded from clear acrylic resins. While commonly available epoxy or polyurethane adhesives may be employed to cement the adapter to the shaft, anaerobic adhesives have been found to be particularly good for bonding molded acrylic adapters to hollow shafts made of Noryl ^® resin. Instead of a cured potting resin, the tip may consist of a pre-formed piece which is adhesively bonded to the cylindrical membrane. Such a tip may be machined or molded, and may be polymeric or metallic in composition.

The length of the cylindrical membrane may be varied widely, from as little as one centimeter to as high as 50 centimeters, and the outer diameter of the cylindrical membrane may vary from 3 mm up to about 10 centimeters. The amount of membrane area thus exposed to soil will vary accordingly. A preferred range for the outer diameter of the cylindrical membrane is from 5 mm to about 4 cm. Larger outer diameters tend to

require thicker membrane walls to provide sufficient wall strength for pressure containment. If the sensor is to be deployed in a vertical position in soil, it is practical to limit the length of the cylindrical membrane so as to measure soil matric potential only in a selected soil depth of interest, usually in the root zone of a particular plant crop. Generally, a membrane length in the range of 1 -10 cm is preferable for vertically deployed sensors. If the sensor is to be deployed horizontally at a selected soil depth, longer lengths may be suitably employed. In practice, vertical positioning can normally be anticipated. When short sensors are deployed at considerable depth in soil, the fluid pressure sensing means may be remotely connected by means of capillary tubing, such as stainless steel capillary tubing commonly used in high pressure liquid chromatography, thereby allowing the fluid pressure sensing means to be physically located at or above the soil surface. Alternatively, the fluid pressure sensing means may be located subsurface along with the sensor, but be connected by signal transmission wires to an above surface receptacle. Other variations of these themes will be readily apparent to one skilled in the art. The invention will now be further described through the following examples. Example 1 A sensor having a design as depicted in FIG. 2 was constructed using a 7.6-cm-long section of 10 mm outer diameter Vycor ^® 7930 (Corning Glass Works) glass tubing as the cylindrical membrane, to which were affixed anodized aluminum end members at each end. Effective surface area of the device was approximately 20 cm ² , and volume of the enclosed chamber was 3 cm ³ . This sensor was evaluated for osmotic pressure response using a series of aqueous 10,000 mol. wt. polyethylene glycol solutions ranging from 0% to 30%. An electronic pressure transducer was attached to the sensor, and signals form the transducer were sent to a data acquisition system connected to a computer. At each concentration, the sensor was immersed in deionized water until fluid pressure within the chamber reached a stable value. The internal pressure stabilized at 0.1 1 , 0.34, 0.67, 1.21 and 1.79 MPa respectively for osmoticums containing 10, 15, 20, 25 and 30% aqueous polyethylene glycol. These results were in generally in good

agreement with calculated values of 0.12, 0.34, 0.74, 1.25 and 1.80 MPa based on measurements of polyethylene glycol solutions in an osmometer.

The sensor, filled with a 30% aqueous solution of polyethylene glycol, was then cycled by immersing the sensor in water, followed by immersion in 20% aqueous polyethylene glycol in place of water, then water, then 15% PEG, then water, then 10% PEG, and finally water again, each time until fluid pressure in the enclosed chamber stabilized. The stabilized pressure readings were 1 .58 MPa (water), 1.09 MPa (20% PEG), 1.51 MPa (water), 133 MPA (15% PEG), 1.52 MPa (water), 134 MPa (10% PEG), and 1.48 MPa (water). The sensor thus showed an ability to be cycled through different fluid pressure levels, returning to approximately the same final level of maximum pressure. Total time of the cycling test was 7,285 minutes. Example 2

A sensor fashioned generally in accordance with FIG. 2 was evaluated for an extended period of time covering four months, wherein the osmoticum was 22.5% polyethylene glycol (10,000 mol. wt.) and the external solution was distilled water. Fluid pressure was approximately 0.90 MPa in the first days of the test, declined to about 0.83 MPa during the first month, further declined to 0.76 MPa in the second month, and appeared to reach an asymptote at 0.70 MPa at the end of the fourth month. Thus, this sensor exhibited an ability to maintain a continuous internal fluid pressure of 0.70 MPa or greater for a period of four months, though with some decline to a final stable value. Example 3 A sensor was constructed generally as depicted in FIG. 2 but containing an inert insert in the form of a solid rod. The membrane consisted of 10 mm o.d. Vycor ^® 7930 bonded to anodized aluminum end members, and the insert was a 0.95-cm-diameter plastic rod machined from polycarbonate rod stock. This sensor was filled with an osmoticum consisting of a 29.4% aqueous solution of polyvinyl pyrrolidinone and 0.05% sodium azide

(bactericide). This sensor was cycled through a series of immersion baths to demonstrate its response to changes in external osmosity, simulating changes in soil matric potential. Results are shown in Table 1 on the next

page. The data in Table 1 demonstrate the ability of the osmoticum in the sensor to respond to varying osmotic potentials in the surrounding medium.

TABLE 1

Time Fluid Pressure External Comments (hours) (MPa) Solution

1 1 .00 distilled water

21 0.93 distilled water move to 17% PEG

45 0.50 17% PEG move to dist. water

69 0.93 distilled water move to 18% PEG

93 0.43 18% PEG move to dist. water

165 0.93 distilled water move to 19% PEG

189 0.34 19% PEG move to dist. water

213 0.93 distilled water move to 19.5% PEG

237 0.31 19.5% PEG move to dist. water

333 0.88 distilled water move to 21.5% PEG

357 0.16 21 .5% PEG move to dist. water

381 0.91 distilled water move to 22% PEG

405 0.14 22% PEG move to dist. water

430 0.91 distilled water

Example 4

Four sensors, constructed as in Example 3, were placed into a bucket of soil along with a tensiometer. The soil was prepared so as to have an intended matric potential of -0.05 MPa. Three of the sensors were filled with an osmoticum containing 30% polyvinyl pyrrolidinone. These three measured soil matric potentials at approximately -0.028 to -0.035 MPa. the fourth

sensor contained 20% PEG. This sensor gave a soil matric potential reading of approximately -0.040 MPa. The tensiometer gave a soil matric potential of only -0.015 MPa, which was believed to be an inaccurate response by the tensiometer. Example 5

Two sensors, constructed as in Example 3, were filled with an osmoticum consisting of a 29% aqueous solution of polyvinyl pyrrolidinone and were positioned horizontally in a soil sample, alongside a tensiometer. The soil sample was prepared so as to have a calculated soil matric potential of -0.033 MPa. The sensors both displayed initial soil matric potentials in the range of about -0.045 to -0.05 MPa during the first 12 hours of soil immersion, then rose to stable values of -0.037 MPa and -0.036 MPa, which were held over a five day period. The tensiometer went to an initial reading of -0.030 MPa, then dropped to a stable reading of about -0.036 MPa, which it also held for the five day period. Thus, good agreement was found between the tensiometer and the sensors at soil matric potentials accessible by the tensiometer. Examples 6-8

Three sensors were constructed having a design generally as depicted in FIG. 4, using a pressure transducer as a fluid pressure sensing member. The sensor of Example 6 was filled with an osmoticum consisting of 29% polyvinyl pyrrolidinone (24,000 mol. wt.) in water. The sensor of Example 7 was filled with 26% polyethylene glycol (10,000 mol. wt.) in water. The sensor of Example 8 was filled with 20% of a 15,000 mol. wt. copolymer of ethylene glycol with bisphenol-A diglycidyl ether (CA Reg. No. 37225-26-6) in water. All three osmoticums also contained 0.05% sodium azide as an added biocide. The three sensors were immersed in water with the pressure transducer of each above the water surface. Internal fluid pressure was measured over a period of approximately seven days, along with water temperature and air temperature. Results are shown in Table 2. Based on the fluid pressures present at the end of the seventh day, the sensor of example 6 had a capability of measuring soil matric potential from 0 down to about -0.6 MPa (-6 bar). For Example 7, this range was extended to

TABLE 2

Day Ex. 6 Ex. 7 Ex. 8 Air Temp. H ₂ O Temp.

(MPa) (MPa) (MPa) ° C ° C

1 0.68 1.03 0.47 22.8 22.5

2 0.65 0.95 0.46 23.5 22.4

3 0.63 0.94 0.46 22.5 22.6

4 0.61 0.92 0.45 23.2 22.8

5 0.61 0.93 0.44 23.8 22.8

6 0.61 0.92 0.44 23.8 22.9

7 0.59 0.89 0.43 23.4 23.3

about -0.9 MPa (-9 bar). For Example 8, soil matric potentials of 0 to about -0.43 MPa (-4.3 bar) were measurable. Example 9

A sensor was constructed generally as depicted in FIG. 4, using a 7.6- cm length of 10 mm o.d. Vycor ^® 7930 cylindrical glass tubing mounted onto a Noryl ^® plastic member and bonded to it with epoxy resin seals. Osmoticum in the sensor consisted of a 29% aqueous solution of polyvinyl pyrrolidinone also containing 0.05% sodium azide. A 1 .1 MPa (160 psig) Ashcroft pressure gauge was attached. The sensor was immersed in water for a period of 90 days. Internal fluid pressure rose to a high of 0.83 MPa during the first day, then settled to about 0.71 MPa in about 30 days. Over the following 60 days of continuous immersion, the sensor maintained a steady internal fluid pressure in the range of 0.67-0.71 MPa. Example 10 Two sensors were constructed generally as depicted in FIG. 5, using a

7.6-cm length of 7 mm o.d. Vycor ^® 7930 cylindrical glass tubing. One end of each of the tubes was potted with a room-temperature-curing epoxy resin composition. The other end was adhered to a male luer lock adapter with the

same epoxy resin composition, and the luer lock adapter was cemented into a hollow 7.6-mm length hollow Noryl ^® shaft. One of the sensors was filled with a solution of 24.5% aqueous polyvinyl pyrrolidinone, which was calculated to provide a matric potential range of 0 to -0.4 MPa (0 to -4 bar), and was assembled to a 0-0.4 MPa dial gauge by means of nylon fittings. The other sensor was filled with a 32% polyvinyl pyrrolidinone solution, which was calculated to provide a matric potential range of 0 to -1 .0 MPa (0 to -10 bar), and was assembled to a 0-1 .1 MPa dial gauge by mean of stainless steel fittings. These were tested for pressure stability by immersion of the glass membranes in water for a period of 33 days. The 0.4 MPa- gauge sensor stabilized at an internal fluid pressure of 0.39 MPa within five days and maintained fluid pressure at 0.37-0.39 MPa during the ensuing 28 days. The 1 .1 MPa-gauge sensor stabilized at 1 .03 MPa within 7 days and maintained fluid pressure at 0.98-1 .03 MPa during the ensuing 26 days. Both of these sensors constructed generally as in FIG. 5 performed in satisfactorily and in accordance with expectations.

The sensors described in the above examples and in the description preceding those examples may be variously used in agricultural applications. Gauge-mounted designs may be placed in gardens, in potted plants, in vineyard rows, or adjacent to orchard drip irrigation lines, for example, to provide visual indication of moisture status of soils. Transducer-mounted designs may be buried at any depth and connected through wires to data recording stations or to data signal transmitters transmitting to remote receivers. These sensors, in combination with servomechanisms or with data analysis devices, may be used to automate irrigation schedules on farms, golf courses and nursery plots in direct response to soil drying cycles. The description and examples given above are believed to fully disclose the best mode of carrying out the invention at the time of this application. Additional modifications and alternative embodiments of these designs and examples, which would reside within the scope of this invention, will naturally be evident to one of skill in the art, and the scope of the invention should thus be regarded and governed in accord with the claims that are appended to this disclosure.