This invention relates to methods and devices for analyzing waste-water and other solutions for trace organic compounds and the like, and, more particularly to a reversed permeation membrane (RPM) extraction system for concentration and subsequent determination of trace organics in sample solutions.

With the ever increasing environmental consciousness of the world, the presence of trace organics in water, waste-water, and other effluents has also become increasingly important. Because some of these substances have toxic, carcinogenic, mutagenic and teratogenic properties, their presence, and particularly their concentrations, have increasingly become the focus of both environmental and health scrutiny. Particularly, the allowable concentration of a wide variety of potentially toxic compounds in waste-water is extremely low, as many have significant adverse effects and toxic activity, even when present in only trace levels of concentration (e.g., several parts per billion, or parts oer trillion). Continuing studies have been directed towarα testing for, controlling, and determining permissible levels of a wide range of potentially toxic compounds, and much research has consequently gone into developing improved devices and procedures for testing for the presence of these various compounds. Among these adopted methods are chromatographic techniques, combined gas chromatography and mass spectrometry techniques (GC/MS), and others. In addition to each of the various classes of compounds having their own specific analytical procedure for sampling, storing, collecting, and analyzing, it is widely understood in the industry that the existing methods of analysis are often labor intensive, equipment intensive, time consuming, and, as a result, costly. ,lt is not unusual for an analysis of waste-water to take two days or more for results to be obtained. Many of these 5 processes also require the collection of a relatively large sample volume (e.g., 1 liter), utilize relatively extensive amounts of laboratory glassware, tubing, and the like, require a relatively large volume of solvent for extraction and cleaning of the glassware and apparatus, and/or require significant energy input to evaporate the solvent and concentrate the sample compounds. 0 in general, compounds of interest have been extracted from waste-water for identification and quantification by passing a relatively large volume of the waste-water through a bed of granulated charcoal. U.S. Patent 3,967,928 (Schmidt et al.) discusses a method of this type, wherein molecules are adsorbed on activated carbon product, followed by desorption. The compounds of interest are generally adsorbed into the charcoal, after which a $ volume of extractant (e.g., chloroform or the like) is passed through the bed of charcoal to release the adsorbed compounds. The extractant can be in the form of a solvent or a gas, and is generally chosen to release the compounds of interest from tne charcoal bed. Thereafter, the

extractant is evaporated to facilitate the assay of the concentrations and characteristics of the various compounds via a gas chromatograph (GC) device. While this particular procedure is widely utilized, because waste-water often contains particuiate matter, the granular adsorbent can easily become plugged or otherwise clogged, thereby interfering with the reliability and efficiency of results obtained.

Other devices and methods for recovering compounds from waste solutions include the incorporation of a two-sided membrane, such as described in U.S. Patents 4,525,278 (which issued to A. Frost) and 4,738,781 (which issued to W. Word et al.). Particularly, these references describe ultrafiltration devices wherein waste-water or the like is fed through a tubular membrane of a non-cellulosic polymer through which compounds of interest permeate for removal through a separate output. Emulsified oils and suspended particles cannot pass through the membrane, and are removed. The micro-porous membrane includes an exterior support tube which must also be porous in nature, such that it will have a low resistance to the permeating flow of the compounds of interest. Other references descri Ding the use of two sided membranes for extraction of compounds of interest include U.S. 4,775,476 (Melcher et al.) and 4,819,478 (Melcher). While a two-sided membrane arrangement can be useful in certain circumstances, it is severely limited by the size and type of compounds which can permeate therethrough, must be isolated from pressure, and/or requires a porous supporting structure which must also be designed to allow penetration, and often requires relatively lengthy concentration times to allow for the permeation of the compounds completely through the membrane. Additionally, cleanup procedures are correspondingly difficult and time consuming, and potential carryover from previous test cycles can seriously undermine reliability of results.

Other methods for concentration of organics from aqueous solutions include the passage of the sample through an uncoated plastic or metal capillary tubing. For example, the procedures reported by Zlatkis et al. incorporated the use of relatively long capillaries (e.g., 15 - 30 meters) through which aqueous solutions were passed. The organic compounds would be retained in the tubing walls while the balance of the waste-water eluted from the capillary tubes. The concentrated organics were then desorbed from the capillary using an organic solvent, generally pushed through the capillary by nitrogen gas or the like. The desorbed solution was then analyzed by gas chromatography. Zlatkis et al. specifically set forth various problems with prior extraction techniques and devices, such as the charcoal filters discussed above. However, it was also reported that the efforts to desorb the trapped organics were not satisfactory, and the use of metal capillaries was seriously questioned. Other methods for desorbing the organics, such as heating the matrix and/or utilizing a membrane dryer arrangement, were attempted to help remove the relatively large volumes of retained water required. Moreover, because long capillaries were needed, the cost, soace and equipment

requirements, and time for collection and analysis were clear negatives for the aDDiicabiiity and efficiency of these systems.

Consequently, there has heretofore not been available in the industry a relatively simple device and process for the extraction of particular compounds from sample solutions in a relatively reliable, efficient, and timely manner. Additionally, due to tne permeaple nature of two-sided membranes and their required porous supporting structures, as weil as the cumbersome length of the capillary tubes utilized for GC analysis by Zlatkis et al., the methods and apparatuses previously available have not been easily aαaptaole to a variety of separation and assaying arrangements. A device and process having the higher sensitivity reαuired for o trace component analysis, while enabling smaller sample sizes and compatibility with water as both the sample and extractant carrier system and which can be used on-line with liαuid chromatography and/or other separation/analyzing devices was needed. Moreover, the devices previously available were generally not easily used in conjunction with tecnniques requiring elevated temperatures and/or pressures. 5 The present invention is a reversed permeation membrane assembly for the analysis of one or more compounds of interest from a sample solution, said memDrane assembly characterized by: a semi-permeable membrane of predetermined thickness and length, said membrane having an inner surface and an outer surface; a membrane support adjacent said outer surface of said membrane; and an essentially non-permeable barrier 0 between said outer surface of said membrane and said support to essentially prevent permeation of compounds of interest into said membrane support.

The present invention is also a method of analyzing one or more compounds of interest from a sample solution, said method characterized by the steps of: providing a reversed permeation collection cell comprising a semipermeable membrane of predetermined 5 thickness and length, said membrane having an inner surface and an outer surface, a membrane barrier adjacent said outer surface of said membrane, said barrier being essentially non permeable to the compounds of interest; passing a volume of sample solution in contact with the inner surface of said membrane such that compounds of interest permeate into said membrane; passing a volume of extractant fluid in contact with the inner surface of said 0 membrane to extract one or more permeated compounds of interest from said membrane; and analyzing the extracted compounds of interest.

Fig. 1 is an enlarged, partially broken-out perspective view of a portion of a reversed permeation membrane cell made in accordance with the present invention;

Fig. 2 is a schematic view of an apparatus for reversed permeation membrane 5 extraction made in accordance with the present invention, and illustrated in an exDerimental set-up for use in conjunction with a liquid chromatography analytical arrangement;

Fig. 3 illustrates a series of chromatograms of extraction Drofiies for 3, 5- dichlorophenol obtained at various sampling times from an RPM extraction arrargement of the present invention;

Fig. 4 is a schematic view of an alternative conceptual arrangement incorporating the reversed permeation membrane cell of the present invention, illustrating an automatic analytical system wherein the RPM cell is connected in line with a liquid chromatograph;

Fig. 5 is a graphical representation of test results from an analytical arrangement similarto that illustrated in Fig. 4, wherein peak analyte detection or concentration of 2,6- dichlorophenol (A) and 3,5-dichlorophenol (B) are plotted against time t (in minutes) of flowing sample;

Fig. 6 is a schematic view of another reversed permeation membrane extraction arrangement of the present invention, illustrated in conjunction with a liquid chromatographic (LC) analytical arrangement;

Fig. 7 is a schematic view of an automatic reversed permeation membrane extraction arrangement made in accordance herewith, illustrated in conjunction with a gas chromatography/mass spectrometry analytical arrangement;

Fig. 8 is a reconstructed total ion current (TIC) chromatogram for an RPM/GC/MS analysis of trace organics in a water sample wherein abundance (a) is plotted against time;

Fig. 9 is a graphical representation of test results illustrating temperature effects on RPM extraction according to the present invention, where analyte signal is plotted against temperature (°C);

Figs. 10A and 10B are graphs illustrating test results comparing the effects of organic solvents on RPM extraction according to the invention, wherein analyte signal is plotted against percent acetonitrile in the water based solvent; and Figs. 1 1A and 1 1 B are graphs illustrating test results showing signal enhancement by solvent solutions, wherein Fig. 1 1 A shows signal intensities for pentachlorobenzene against various Hildebrand solubility parameter (HSP) of the sample solution, and Fig. 1 1 B shows a similar plot for fluoranthene.

Fig. 12 is a view of a highly preferred apparatus embodiment of the instant invention incorporating gas chromatography wherein the RPM cell is electrically heated.

Figs. 13 through 17 show the same apparatus as Fig. 12 operated according to a highly preferred method embodiment of the instant invention.

Referring now to the drawings in detail, wherein like numerals indicate like elements among the various views, the apparatus and method of the subject invention for analyzing trace organics with low solvent and sample requirements will be descrωed. Particularly, the present invention pertains to an apparatus and method for improved extraction of compounds from sample solutions, which incorporates a one-sided " membrane

as illustrated in Fig. 1. Membrane cell 15 comprises a membrane support 17 having at least one non-porous face 18 on which a semipermeable membrane 20 will be supported.

As illustrated in Fig. 1 , membrane support 17 can preferably comprise a length (L) of polytetrafluoroethylene tubing having a predetermined inside diameter (e.g., approximately 1.6mm). As will be described below, while the semipermeable membrane can be provided of varying thicknesses and materials, depending upon the particular application and compounds to be analyzed, it may preferably comprise a silicone rubber having predetermined inside and outside diameters for support within membrane support 17. The supported membrane 20 will provide a predetermined flow channel 22 through which sample solution and solvent will be passed, as described below.

It is also contemplated that a particular reversed permeation membrane cell 15 will further preferably comprise membrane/support seals 24 adjacent the opposite ends of the longitudinal length L thereof. Because sample solutions such as waste-water and the like will be flowed through membrane cell 15, and may often be subjected to elevated pressures and/or temperatures, a reliable connection and seal between semi-permeable membrane 20 and its support 17 will be needed at least adjacent the opposite ends of the membrane ceil 15. The manner in which seals 24 are provided in any particular membrane/support arrangement may vary depending upon the specific materials utilized for membrane support 17 and the selectively permeable membrane 20. For example, when membrane 20 is provided of a silicone rubber material, it may be preferred to utilize silicone sealant at either end to provide a reliable and compatible seal.

While it is contemplated that the membrane cell 15 can be assembled by inserting a semi-permeable membrane 20 into a corresponding membrane support 17 and sealing the opposite ends therewithin, the manner of creating membrane cell 15 is not critical to the present invention. For example, for some combinations of materials, membrane support 17 and membrane 20 may be simultaneously formed, such as by co-extrusion or the like. Membrane 20 may alternately be coated onto support 17, or formed by the chemical bonding of appropriate materials to support 17.

Fig. 2 illustrates an analytical system assembly 30 comprising a reversed permeation membrane (RPM) extraction cell 15 as discussed above attached to a liquid chromatography valve 32 for use in conjunction with an LC column 34 and detector 36. As also illustrated, assembly 30 might preferably comprise an injection syringe 38 coupled to one input of valve 32, and a liquid chromatograph eluent pump 40 connected to a source of LC eluent 41 (e.g., 35% acetonitrile/65% .01 M phosphoric acid) to provide extractant solvent to the analytical system via supply tube 43 and valve 32.

Experimental tests were run on an analytical system assembly essentially as described above, whereby sample solution was filled into RPM cell 15 and held therewithin for various predetermined time periods (i.e., 0.5, 1.0, 2.0, and 4.0 minutes) before being flushed

with water, a flushing solution, or gas (e.g., air, nitrogen or the like). As the sample solution is passed through cell 15, compounds of interest in the sample solution (e.g., chloroDnenols, benzene, chlorobenzene and polycyclic aromatic hydrocarbons (PAH)) Dermeate into semipermeable membrane 20. The experimental set-up as shown in Fig. 2 was utilized in various ways to investigate the operational characteristics of the subject RPM cell. One experiment was to determine the extraction of compounds of interest into the membrane and the reversed permeation extraction profile into the extractant eluent flowing through the cell. In this experiment, the LC column was a conventional C _{1 8} , 5 micron guard column, and the sample solution was acidified water containing 10 ppm 3, 5-dichlorophenol. The amount of sample loaded into the RPM was 0.2ml, and the extractant solvent comprised 35% acetonitπle and 65% .01 M phosphoric acid. Sample was filled into RPM cell 15 and held for various sample extraction times of 0.5, 1.0, 2.0 and 4.0 minutes before being flushed rom the system. Thereafter, the extraction eluent was directed through the membrane cell 1 5 and " injected" into the LC column 34 for detection by detector 36. Fig. 3 graphically illustrates chromatograms showing the recovery profiles of the RPM extraction tests at the respective sample extraction times.

One sample (e.g., shown as peak I on Fig. 3) was directly i njected as soon as the RPM was loaded, and no flush was used. As can be appreciated from Fig. 3, for this direct injection, most of the dichlorophenol analyte is still in the aqueous phase. The peaks for extraction times 0.5, 1.0, 2.0 and 4.0 minutes each represent only the amount of compound which has permeated the membrane, since all non-permeated material has been flushed out. Because the analyte can be carried to the detector only as it diffuses back out of the membrane, the peaks indicate the profile of the "desorption" or back diffusion of the analyte. The recovery profiles of the 1 , 2 and 4 minute tests are quite similar, indicating that analyte extraction was essentially completed in approximately 1 to 2 minutes. As can be appreciated, because the compounds of interest do not have to diffuse or permeate through the entire width of the membrane wall to a collection side (as was required with the two-sided membranes often utilized heretofore), the present apparatus and method can be quite useful for molecules which have a relatively slow diffusion rate. Moreover, since the semi-permeable membrane 20 of the present invention is effectively completely supported (e.g., via membrane support 17), high pressures and/or temperatures can be utilized in the system, and high pressure LC columns and the like can be connected directly thereto. Moreover, the requirement for a non-porous support member contacting the membrane allows a wider variety of membrane mateπals to be utilized.

Once the compounds of interest have been allowed to permeate into the membrane, the balance of the sample solution is passed from cell 15 to a waste collection container or the like. Cell 15 is then flushed out with some type of solvent, which may

preferably be water, water/methanoi, water/ethanol or metnanol, to remove all non- permeated compounds, inorganics and particulates. It should also be noted that the flushing solvent can be chosen to further "select" which compounαs are concentrated within membrane 20, as the flushing solvent can be utilized to desorb non-desired components from membrane 20 in addition to flushing residual sample matrix from cell 1 5. An example would be where five compounds readily permeate membrane 20, but only two of the compounds are of particular interest in the testing procedures, or where three of the five compounds are present in much higher relative concentrations. By appropriately choosing the flushing solution, the other three compounds which have permeated into membrane 20 can be selectively removed (ortheir concentrations reduced) during the flushing cycle.

Once the permeation or concentration step is followed by an appropriate flushing step, an extractant solvent or eluent is used to collect the permeated compounds from the membrane. Again, the mechanism for desorbtion from membrane 20 is reversed permeation of the compounds into the extract solvent. The cnoice of the extractant solvent adds additional selectivity to the overall extraction process in that different solvent properties (e.g., pH, solubility, polar characteristics, etc.) can be utilized to extract certain compounds to the exclusion of others. As discussed below, the driving force in each step of the process is maintenance of appropriate water partition coefficients between the fluid in flow channel 22 and the membrane interface. By optimizing the various RPM parameters, substantial selectivity in collection and concentration of particular compounds of interest can be achieved. As also indicated from the experimental results, the permeation and extractant times required for relatively efficient concentration and collection procedures is substantially reduced, as is the required collection column lengths. The time, space, capital investment and labor savings which result from these and the other unique features of the present extraction apparatus and method provide a new dimension in trace element analysis and related applications.

Fig. 4 illustrates a conceptual automated analytical system 150 using the reversed permeation membrane technology of the present invention. Particularly, system 150 might preferably comprise a one-sided membrane cell 1 15 similar to cell 15 described above, and an LC column 134 and detector 136 similar to that contemplated in Fig. 2. The RPM cell 1 15 serves as the sample loop for the LC system (i.e., LC column 134 and detector 136) after collection of the sample, and LC injection valve 132 coordinates the collection and injection procedures. Particularly, it is contemplated that the collected analyte from the reversed permeation membrane cell 1 15 will be extracted by the mobile phase of the LC system (e.g., 70% acetonitrile, 0.02 M H3PO ₄ ). Valve 132 serves as the base around which the system 150 is set up, and can comprise any of a variety of LC valves. As can be appreciated, sample solution is pumped via sample pump 158 through a cross-flow filter device 156 to valve 132 through cross connector 148 and a three-way flow valve 154. Similarly, flush solvent is provided to valve 132

via pump 145, on/off flow valve 146, and cross connector 148. Extractant solvent is similarly provided through pump 141 , on/off flow valve 142, and connector 148.

In operation, the automated analytical system 150 would σrovide samσle solution to valve 132 via Dump 1 58 (e.g., a pump set at a predetermined flow rate). The sample is directed to reversed permeation membrane cell 1 15 via valve 132, and from there to waste port 135. Thereafter, flushing is completed by flushing solvent such as water, or a particular solvent mixture (e.g., NaOH, or inert gas) via flush pump 145 and flow valve 1 6. Again, this flow is directed through valve 132 to RPM cell 1 15 and then to waste port 135 (which may or may not be directed to the same waste container to which the sample solution was directed). o Following flushing, extraction is completed by pumping a predetermined volume of extractant solvent into membrane cell 1 15 via pump 141 and flow valve 142. The extractant solvent is generally allowed to stand within the membrane cell 1 15 for a predetermined extraction time before it is injected into the LC column 134, such as via eluent pumo 140.

Fig. 5 illustrates results of flowing sample test results taken on a system similar to 5 that described with respect to Fig. 4. Particularly, this illustration plots the peak heights of detected compounds of interest versus the time of the flowing sample (in minutes) at a flow rate of 0.2 mL/min. The sample solution utilized was acidified water containing 10 ppm of 2, 6- dichlorophenol (line A) and 10 ppm 3, 5-dichlorophenol (line B). The sample pumo 158 was allowed to run for varying times of 2, 4, 8 and 16 minutes at the rate of about 0.5 mL per 0 minute, and an extractant solvent comprising 10% acetonitrile and 90% 0.1 NaOH was allowed to stand for 5 minutes before injection. The RPM column utilized comprised a cyli ndrical membrane support (e.g., 17) having a length of approximately 300 mm of polytetrafluoroethylene material, having an inside diameter of approximately 0.8 mm) and an outside diameter of approximately 1.6mm. Membrane 20 comprised silicone rubber tubing 5 fitted within the membrane support and having an inside diameter of approximately 0.5 mm and an outside diameter of approximately 0.8 mm.

As can be seen from a review of Fig. 5, it appears that the system begins to reach its equilibrium after about eight minutes, as the concentration curves begin to flatten out. These results also show that the concentration after about eight minutes in a flowing sample is 0 nearly three times the concentration detected after two minutes of flow. Moreover, the concentration factor after eight minutes of flow is nearly ten times the concentration factor of a non-flowing sample at two minutes, as discussed with respect to Fig. 3 above. In this regard, it can be seen that enhanced concentrating capabilities can be obtained utilizing the present reversed permeation membrane invention, wherein a relatively short RPM cell is utilized with a 5 flowing sample to collect and concentrate compounds of interest in a relatively short time period. In this way, not only can very small samples of solution be effectively extracted but much lower concentrations of compounds of interest can also be detected in a quick, automated, and efficient manner by utilizing a flowing sample.

Fig. 6 is a schematic illustration of another preferred emoodiment of an online application of the subject RPM detection cell. Particularly, Fig. 6 illustrates an automated analytical system 250 wherein RPM cell 215 is connected directly with an LC analytical system. Compounds of interest, such as listed in Tables I and II below, which have permeated into the membrane of cell 215 will be extracted by the mobile phase of the LC system (e.g., an extraction solvent comprising 70% acetonitrile, 0.02 M H ₃ P0 ₄ ). LC valve 232 is the base of the system, and can comprise a multi-port injection valve. A second multi-port valve 270 is utilized to control the flow of sample solution from a sample source 272, as well as the input of water or other flush solvent or gas from source 273. The flush water or solvent can be provided via a pump (e.g., 245). Another pump 275 can be provided for facilitating the discharge of waste fluids. For automatic sample loading, a predetermined volume of sample solution is pulled into a sample loop (e.g., 277) of predetermined volume (e.g., 2.0 mL) by pump 275. For manual sample loading, the injection valve 270 can be fitted with a syringe adapter or the like. LC valve 232 controls the injection of the compounds collected in the RPM cell 215 into an LC system (e.g., LC column 234 and detector 236). The attachment of RPM cell 215 to injection valve 232 can be accomplished by a variety of arrangements known in the industry, however, the attachment must be sufficient to withstand the high pressures of an LC system. For example, where valve 232 comprises a liquid chromatography injection valve, it was found that Vespel brand GC capillary ferrules and a standard tubing nut perform nicely for this function. It is contemplated that a variety of other arrangements will also be effective to connect the RPM cell of the present invention in-line with LC, GC and other high pressure/temperature analytical devices. Fig. 6 shows valves 232 and 270 configured so that valve 270 is in its sample collection mode (sample from source 272 being loaded into loop 277), while valve 232 is in its inject mode (extractant from source 241 being provided through cell 215 directly to column 234 for analysis). Alternate valve pathways (e.g., for sample flow to cell 215 and flushing) are indicated by dashed lines.

In tests completed utilizing an automatic loading configuration similar to that shown in Fig. 6, the LC system comprised an LC pump and detector, wherein the wavelength of detection could be either 210 or 254 nm. A C-18, 4.6 x 100 mm column was used to determine the presence of chlorobenzenes, while a C-84.6 x 10 cm column was used to determine the presence of PAH compounds. For both columns the mobile phase was 70% acetonitrile, 0.02 M H ₃ P0 ₄ , and data collection and analysis was completed using a liquid chromatography integrator. For the set-up utilizing injection valves, valve switching was completed automatically via a digital valve interface and the external control option on the integrator (these devices schematically illustrated in Fig. 6 as control 280).

In use, the RPM sampling was initiated by injection of the sample solution via valve 270. After the sample solution had flowed within RPM cell 215 for a predetermined sampling time, a water flush was completed by loading water from source 273 via pump 245

and valve 270. Thereafter, valve 232 permitted injection of the mobile phase eluent from source 241 via LC pump 240. Due to the relatively thinner membrane thickness and smail internal diameter of the capillary sized tubes utilized, it was found that foil owing sample dwell time and flush procedures, the flow of extractant could be pumped without interruption (i.e., without holding extractant within cell 215) through cell 215 and into the LC column. The ability to extract on a continuous, or non-stoo basis is believed to be a significant breakthrough in extraction/analysis procedures of this type. The mobile phase completed extraction of permeated compounds from the membrane of cell 215 ana was thereafter provided directly to LC column 234 for chromatographic separation. The data obtained from LC separation testing o provided chromatograms exhibiting sharp peaks due to the focusing of compounds of interest on the LC column. It was found that the focusing characteristics was a result of collection and concentration of the compounds in RPM cell 215, and that after extraction, collected compounds elute as a very compact sample plug into the LC column.

As mentioned above, the material from which semi-permeable membrane 20 is 5 constructed will provide a certain degree of selectivity of compounds thereby providing a first level of selecting compounds of interest to be analyzed. When the sample solution is placed into contact with the supported membrane 20, the compounds of interest dissolved in the sample solution permeate into the membrane, while the balance of the residual sample matrix is expelled from the membrane cell 15 as it is flushed out by continuing flow of the sample, 0 and, eventually, by a flush solvent. The selectivity of the membrane is a result of differing water partition coefficients between the various elements of the solution and the membrane interface. In order that a compound will preferably permeate into the membrane, the coefficient must favor the membrane, as opposed to the solution. The partition coefficient between the solution and membrane interface can be determined by the following equation: 5 ^p s = UX] Λ/ _M )/([X] ₅ /VS)

Where [X]M and [X] _s are respective concentrations of a compound X in the membrane and in the sample solution.

Vi i an V ₅ are the volumes of the membrane and sample solution phases, respectively. 0 For example, the efficiency of collection and recovery of chlorobenzenes and PAH compounds (which are two groups of compounds of particular general environmental interest) are listed below on Tables I and II for several different membrane configurations. Particularly, the OV-1 film is 100% dimethylpolysiloxane, while the DB-5 material is more polar and comprises 5% diphenyl and 95% dimethyl-polysiloxane. The SE-30 is a silicone-based oil which 5 is not crosslinked, and the DB-624 comprises cyanopropyl, phenyl, dimethyl-polysiloxane polymer film.

TABLE I

Efficiency Data in percent for Chlorobenzenes on various RPM Columns

A = benzene 3 = chlorobenzene C = 1 ,2-dιchlorobenzene

D = 1 ,2,3-trιchιoroDenzene E = ,2,3,4-tetrachlorobenzene

^: = oentacnloroDenzene

The tests of Table I were obtained on an experimental device shown and described with respect to Fig. 6 above, wherein a six minute sampling time was utilized with a 0.58-0.99 ppm concentration level of chlorobenzenes, and a flow rate of 0.20 mL per minute. Similarly, Table II shows test results from an experimental set-up utilizing a six minute sampling time with a solution of 160-240 ppb concentration of PAH compounds in 10% methanol, with an RPM flow rate of 0.2 mL per minute. The individual stock solutions of the chlorobenzenes and PAH compounds were prepared by dilution of the compound in acetonitrile (e.g., HPLC grade acetonitrile). Particular solutions of the chlorobenzenes were then prepared by dilution in water of the indicated parts per billion stock solutions of each compound, while the PAH compounds were prepared by dilution of the indicated parts per million stock solutions in 10% methanol (e.g., High Purity methanol) due to the low solubility of PAH compounds in water.

TABLE II

Efficiency Data in percent for PAHs on various RPM Columns

G = Acenaphthylene H = Phenanthrene I = Fluoranthene J = Benz(a]anthracene K = Beπzo(b]fluoranthene

.30

35

As will be understood, as the sample solution passes through RPM coiumn 215, the compounds of interest permeate into the membrane film, the diffusion coefficient and the partition coefficient of the membrane/solution interface governs the amount of each compound which is permeated into the membrane at any point along tne sample Dlug. It has been found that initially, efficiency of the permeation of the compound into the membrane film is relatively high due to the correspondingly high partition coefficient for the interface. However, eventually the system approaches a steady state and the efficiency of permeation begins to decrease or level off. This leveling off is also governed by the partition coefficient between the interface. While it has been found that slower flow rates appear to enable more o efficient permeation into the membrane as a result of longer contact time per unit area of the membrane, it has also been observed that the ability to collect sample compounds utilizing a fast flowing sample permits the collection and analysis of much lower concentration levels (e.g., parts per billion (ppb) and parts per trill ion (pptr)) in shorter periods of time.

There are two approaches to considering and utilizing sample flow rates. As has 5 been discussed, the highest efficiency of extraction occurs at low flow rates of sample through the RPM column (e.g., 0.1 mL/min). However, at lower flow rates, a corresponding relatively smaller sample volume will pass through the RPM (e.g., 0.1 mL/min x 10 min = 1 mL). As a result, the absolute amount of analyte available for extraction is reduced for a specific sampling time. It has been shown that if faster flow rates (and correspondingly larger sample volumes) are 0 used, the absolute amount of the analyte collected is greater in the same time period even though the collection efficiency is reduced. This effect was illustrated by testing the following mixture of compounds in the 10 to 20 ppb concentration range.

1. Acenaphthalene 10 ppb

2. Phenanthrene 10.7 ppb 5 3. Fluoranthene 10 ppb

4. 1 ,3,5-Trichlorobenzene 23 ppb

5. 1 ,2,3,4-Tetrachlorobenzene 18.5 ppb

6. Pentachlorobenzene 12.5 ppb

7. Hexachlorobenzene 12.5 ppb 0

A water solution containing the above compounds were pumped through a 75cm

RPM column, with a 7um thick membrane for 5 minutes at different flow rates. The collected analytes were then analyzed. The amounts collected at a 4mL/min flow rate were about four (4) times the amount collected at a 0.34mL/min flow rate, and an estimated sixteen ( 16) times the 5 amount collected at a O. l mL/min flow rate. This would allow better detection limits or shorter sampling times and/or RPM column lengths for the same detection limits.

Consequently, while the efficiency of permeation and collection increases as the flow rate decreases (i.e., as the contact time is increased), efficiency of collection is not always

critical, as more consistent sample solutions and higher sensitivity to low concentrations can be accomplished in a shorter amount of testing time utilizing a flowing sample in the present invention. In effect, it has been found that an extraction test which would normally require as much as one liter or more of sample solution, 40-50 mL or more of solvent, and 1-2 days of relatively manpower intensive testing procedures, can be automatically completed and reported within an hour or less with an automated configuration such as shown and described with respect to Figs. 6 and/or 7 above. As illustrated in the test results described above, RPM cells made in accordance herewith and having a length of one meter or less have been shown to obtain extraction results comparable or better than those previously provided by devices and procedures which required collection columns of 15 to 50 meters length.

As can be appreciated, the reduced column length required for the RPM extraction procedures of the present invention simplify preparation, use and handling of the physical equipment and procedures involved, reduce the costs of the columns themselves, and enable relatively easy storage and transportation of RPM cells to remote sensing locations for collection of samples. Particularly, it is contemplated that RPM cells made in accordance herewith, and generally having an operational length of approximately 1 meter or less, can easily be transported to a remote collection site where a predetermined volume of sample solution can be run through the columns on site. Thereafter, the columns are returned to the laboratory for extraction and separation procedures. In this way, a relatively large volume of sample solution could be tested for very low concentrations of trace elements without a need to transport large quantities of sample solution between the site and the laboratory.

It is also contemplated that the present RPM collection and extraction process can be similarly utilized in-line with gas chromatography/mass spectrometry (GC/MS) analyzing devices for trace organics. Particularly, Fig. 7 illustrates a schematic illustration of an RPM/GC/MS air-segmented interface utilizing three pneumatically actuated multi-port valves. This arrangement illustrates a preferred arrangement for integrating the present invention with the GGMS hardware, while incorporating an air segment to separate the sample solution from the organic extractant solvent.

Referring to the analysis system 350 of Fig. 7, sample solution is loaded into sample loop 377 of valve 370 from a source (e.g., 372), and a pump (e.g., 373) pushes the sample to valve 332 where it is directed through RPM cell 315. After the predetermined sample volume from loop 377 has contacted RPM cell 315, and cell 315 has been appropriately flushed, valve 332 is switched to allow extractant solvent to pass through cell 315 such as via eluent pump 340. Preceding the injection of the extractant solvent, however, it is contemplated that valve 332 will pass an air segmentthrough RPM cell 315 to limit the diffusion of organic elution front into the wash or sample solution, to minimize the proPability of water being injected into the GC device, and to facilitate flow rate measurement for timing of the valve switching events themselves. Air gap loop 385 is provided to dose the volume of air to be passed through the

RPM cell. The extractant solvent is pumped into the GC sample loop (e.g., 384) anσ injected into GC/MS analyzing device 382 via valve 380.

A number of tests were run on a device similar to that shown and described with respect to Fig. 7, wherein the first 100 uL of eluent exiting RPM cell 315 was injected into the GC device 382. In the testing arrangement, valves 370 and 332 were controlled by a liquid chromatography integrator, while valve 380 was controlled by the GC/MS hardware. Fig. 7 collectively illustrates the controlling hardware schematically as controller 391. RPM cell 315 was provided as a 100 cm long cylindrical tube of approximately .53 mm inside diameter, and having a 7 uM film membrane of poly(dimethylsilicone). The sampling time was 12 minutes at approximately 0.3 mL per minute flow rate, while the air gap volume was 10 uL. The extractant solvent utilized was 100% acetonitrile. The GC/MS hardware comprised a MS, with a GC injection loop volume of 100 uL, a retention gap of 15 meters by 0.53 mm inside diameter (intermediate polarity deactivated), and the analytical GC column was 30 meters long with a 0.25 mm inside diameter and a 0.25 um film of (5%- phenyl) - methylpolysiioxane. Additional parameters utilized for the experiment included an divert time of 1.40 minutes and a heat time of 2-10 minutes, with GC temperatures of 85°C isothermal for 7 minutes, then 20 ^C C per minute increase to 285°C. The MS conditions included scan range of 45-300 m/z at two scans/second (Emv = 1729 volts).

The reconstructed total ion current (TIC) chromatogram for the RPM/GC MS analysis performed on the device described above is shown in Fig. 8, where abundance (a) is plotted against time. The eight peaks shown on Fig. 8 represent the direct analysis of the following analytes and their concentrations:

Peak Analyte Concentration(ppb)

1. 1 ,2-Dichlorobenzene 15.0 2. 1 ,2,3-Trichlorobenzene 19.2

3. 1 ,2,4,5-Tetrachlorobenzene 13.8

4. Acenaphthalene 1 1.4

5. Pentachlorobenzene 1 1.7

6. Hexachlorobenzene 10.8 7. Phenanthrene 12.4

8. Fluoranthene 12.0

This example documents the fact that the apparatus and process of the present invention can selectively extract analytes of interest from aqueous and non-aqueous samples for direct transfer to a gas chromatograph/mass spectrometer for on-line separation, identification and quantification. The RPM extraction process provides for the transfer of collected analytes with organic solvents which are compatible with GC analysis. The sampling, transfer and GGMS analysis is completely automated for on-line and unattendeα operation if

desired. In this way, the attributes of the present RPM extraction technology and its selective analyte collection can be directly combined with the separation power, sensitivity and selectivity afforded by large volume on-column injection high resolution GGMS technology. The amount of analyte collected by the RPM cell may be selected and controlled by adjustment of the sample volume, the flow rate, the extractant solvent, etc., and an extremely wide range of analyte concentrations may be assayed with this device. Very low limits of detection (i.e., in the low parts per trillion (pptr) range or lower) are attainable utilizing the present RPM invention.

It has further been found that as a result of the relatively short column lengths, and the ability to quickly flush residual sample solution and rapidly recover analytes permeated into the RPM cell membrane, the present RPM capillary cells feature rapid recovery and essentially no carryover between successive testing operations. Consequently, successive tests can be manually or automatically performed without requiring inconvenient and laborious disassembly and cleaning operations. As mentioned above, due to the strengthening nature of the non-porous membrane support of the present invention, membranes of widely varying chemical and physical make-up can be utilized to augment the selectivity and control of the concentration and extraction process. By simplifying and speeding up the trace component analysis testing procedures, and by providing virtually unlimited opportunities for improving selectivity of compounds collected and detectable concentrations, analysis of various fluids such as waste-water and the like is improved, as is the ability to monitor the quality of industrial effluents, drinking water, ground water runoff, etc.

Additional testing has been undertaken to study the various parameters of the subject RPM extraction device and process. As mentioned, the collection characteristics of an RPM cell are dependent on parameters such as column length, thickness of the semi-permeable membrane, composition of the membrane, and diffusion rates (i.e., the solubility plus diffusion rate) of the individual compounds to be collected. The amount of each compound collected also depends on its concentration, the flow rate of the sample, the sampling time and the temperature.

For example, the collection time is one of the critical parameters that effect the compounds of interest collected within the RPM cell. During the collection of PAH compounds on a 90 cm OV-1 RPM cell (for concentrations of approximately 180-240 ppb of the PAH compounds) for example, it was found that the peak collection begins to level off at about 12 minutes of sampling time for acenaphthylene, phenanthrene, and fluoranthene. The loading of the other PAH compounds is much lower and does not level off. On the other hand, in a sampling time-dependence test concerning the collection of chlorobenzenes (0.6-1 ppm concentration) within RPM cell 233 cm in length (7 urn OV-1 ) and 0.53 mm in inside diameter, it was found that only the collection of benzene appeared to level off at about 12 minutes of sampling time, while the RPM cell continued to concentrate the other chlorobenzenes. It is

believed that the leveling effect observed for the permeation of benzene mayhave resulted from a rapid equilibrium in the film.

As the molecular weight of permeating compounds increases, the diffusion rate through the film correspondingly decreases, thereby causing equilibrium within the membrane to be reached more slowly. With regard to the collection of benzene, the limiting factor , appears to be its rapid equilibrium within the membrane. In contrast, the PAH compounds having greater molecular weights are limited by their diffusion rate within the membrane. During a particular sampling period, lower diffusion rates lead to a lower permeation of the compounds and slower equilibrium. In this respect, it has been found that an increase in film thickness tends to improve the efficiency for collection of chlorobenzene and 1 , 2- dichlorobenzene, but has little effect on 1 ,2,4,5-tetrachlorobenzene and pentachlorobenzene. Moreover, sampling time of collection through an RPM cell can also be adjusted to access a wide range of sample concentrations. For example, a highly concentrated sample can be assayed utilizing a relatively short sampling time, while a longer sampling time, or flowing sample (i.e., larger volume of sample), can better determine trace analysis on the same RPM cell.

As discussed above, once the compounds of interest are collected from the sample solution, the residual sample matrix is preferably flushed from the column with the use of a flush solvent or gas. While use of a flush solvent provides a faster and more thorough cleaning of the RPM cell, it may also cause reverse permeation of collected analytes back into the flush solvent. In order to minimize undesired loss of analyte, the partition coefficient for the membrane/solvent interface must favor the membrane. The amount of solvent should also be minimized, and the total volume will depend upon the internal volume of the RPM cell itself. As mentioned above, some selectivity as to permeated compound loss can be obtained by proper choice of flush solvent. For example, the concentration of permeated compounds which are more soluble in water (e.g., benzene, chlorobenzene, acenaphthylene, etc.) can be reduced or eliminated by utilizing a water based flush solvent, and by increasing the water flush time.

In most situations, the flush step of the present process is necessary to remove the particulates and other interferences found in the sample matrix. If these are not present, or are removed by some pre-treatment, the flush step may be eliminated. Additionally, it may be desirable not to use the flush step in setups where the RPM cell acts like an LC loop injector for compounds which are not extracted by the RPM while concentrating compounds which are extracted. In this dual mode, both types of compounds may be analyzed in one injection. Tests varying the length of the RPM cell column length show that there does not appear to be any significant effect on the rate of loss of permeated compounds due to varying column length. On the other hand, longer column lengths typically result in additional capacity and better efficiency for collection of particular analytes due to the increase in _>

membrane volume and contact area. With respect to the collection of chlorooenzenes at a fixed flow rate, it has been seen that as column length increases from about 20 cm to about 90 cm, the amount of collected compound increases for all of the priority chlorobenzene pollutants discussed above. Further increases in column length from about 90 cm to about 510 cm, however, have iess effect for most of the chlorobenzenes, as a level ing-off effect becomes evident between about 230 cm and 510 cm. Because it would be exoected that increasing column length should continue to increase the efficiency of collection up to 100% , this leveling off is indeed unexpected and is believed to be likely due to the slow diffusion of the compounds of interest within the sample solution plug as it passes through the RPM cell. Particularly, as the sample plug passes through the cell, an essentially laminar concentration profile develops, and only the compounds on the outer portion of the profile will permeate into the membrane. The comoound molecules in the inner portion of the plug must diffuse outwardly to contact the membrane. The limitation of slow diffusion would be exacerbated as the column diameter increases, and can be minimized by decreasing the column diameter and/or by increasing the temperature or turbulence of flow therewithin (e.g., adding glass beads or other inert material within the column) and/or decreasing the flow rate.

A similar leveling-off effect was discovered witn the PAH compounds acenaphthylene, phenanthrene, fluoranthene, and benz[a]anthracene with 233 cm long columns. Although some increase in efficiency is observed for a few compounds by utilizing longer RPM columns (e.g., 200 - 500 cm), it has been observed that the amount of increase is not directly proportional to the increase in length. While RPM column lengths of up to 5 meters or so may have slight advantages for a few compounds, lengths of less than about 2 meters, and most preferably about 1 meter or less, are optimum for most applications.

The thickness of the membrane of the present RPM cell can also be adjusted for selectivity and efficiency purposes. For example, for the chlorobenzene compounds being collected on the DB-5 material discussed above, a thicker membrane was found to increase amounts of collected compounds. It is believed that this increase is due to the increased volume of such thicker membrane, which results in a more favorable partition coefficient between the membrane and the analyte in accordance with the equation set forth above. The magnitude of the increase of collected PAH compounds with thicker membranes decreases for more non-polar compounds. In fact, for the compounds benz[a]anthracene and benzo[b]fluoranthene, there is actually a slight decrease of collected compound with an increase in membrane thickness. In contrast, varying the membrane thickness of the SE-30 material had no effect for collection of PAH compounds. Increased collection of compound with increased thickness of the membrane results only when the compound can diffuse through the membrane at an acceptable rate. Depending upon the diffusion rate of the analyte, further increases in membrane thickness have negligible effects as the compounds

never permeate far enough during the sampling time for the additional thickness to make a difference.

As set forth above in Tables I and II, smaller RPM cell diameters show increases in compound collection with increased sampling time. As discussed above, this appears to be the result of smaller diffusion distances within the sampie plug, minimizing the leveiing-off effects for longer length columns. For larger diameter columns (which may be required with thicker membranes) the addition of glass beads or other inert fillers should overcome any sample plug diffusion limitations which might otherwise result from the larger diameters.

Tests on collection of chlorobenzenes and PAH compounds in experimental RPM cells made in accordance herewith show that good linearity of collection of chlorobenzenes is observed from between at least about 64 ppb and about 2.75 ppm, while good linearity of collection of PAH compounds was seen between at least about 20 ppb and about 420 ppb. Consequently, a variety of RPM cell parameters can be adjusted to provide additional selectivity for optimizing the determination of trace organics in sample solutions such as water. It should also be noted that the diffusion rate of various compounds into the semi-permeable membrane will change with varying temperatures, as will the solubility of the compounds in the membrane, flush solvent, and extract solvent. Control of test temperatures can thereby provide additional selectivity for the present extraction method.

The effect of temperature on RPM extraction operations was investigated using an RPM/LC configuration similar to that shown in Fig. 6. In this experiment, the RPM cell comprised an 88 cm long, 0.53 mm i.d. fused silica capillary with a 7 urn thick 100% poly(dimethyisiloxane) (i.e., Rtx-1 ) membrane. The membrane was thermostated using a proportional band temperature controller, and sample solution was pumped through the RPM column for 6 minutes at a flow rate of 0.2 mL/min. The analytes were then extracted using an acetonitrile/water solution and analyzed on a HPLC column using UV detection.

An aqueous sample solution containing the following organic analytes at the listed concentrations was used as a test mixture. TEST MIXTURE

a acenapnthalene 144 ph phenanthrene 124 f fluoranthene 120

Experiments were performed at approximately 20°C intervals in the temperature range between about 37.5° and 100°C. The analyte signal (LC peak height normalized to

37.5°C) is plotted as a function of temperature (°C) in Fig. 9. To clearly display relative effects of temperature, the analyte peak heights were normalized to the intensity obtained at 37.5°C. The results in Fig. 9 show that the analytical response for benzene, chlorobenzene, and 1 ,2- dichlorobenzene decreases as the RPM cell temperature increases. The attenuation is most significant for benzene, followed by chlorobenzene and dichlorobenzene. The attenuation appears to scale with relative analyte vaoor pressure or boiling point. The higher molecular weight chlorobenzenes exhibited relatively minor changes in signal over the temperature range studied. In contrast, the PAH response was enhanced at higher temperatures. From this study it is clearly seen that RPM temperature may be tailored to increase the selectivity and sensitivity of certain compounds.

A study was also conducted to investigate the effects of organic solvents in the sample solution. The studies were performed at ambient temperature using an RPM/LC configuration similar to that shown in Fig. 6. In this experiment, the RPM cell comprised an 88 cm long, 0.53 mm i.d. fused silica capillary with a 7 um thick 100%o poly(dimethylsiloxane) (i.e. Rtx-1) membrane. Sample solution was pumped through the RPM cell for 6 minutes at a flow rate of 0.2 mL/min. After sampling, the RPM cell was flushed with pure water for 2 minutes at 0.2 mL/min. The analytes were then extracted using an acetonitrile/water extraction solution and analyzed on an HPLC column using UV detection.

Solvent solutions studied included acetone/water, acetonitrile/water, methanol/water and 2-propanol/water. Sample solutions were prepared having 0, 5, 10, 20, 40, 60, 80 and 100% (v/v) organic solvent content, respectively. Analyte mixtures were prepared in binary organic-solvent/water solutions with concentrations as listed above for the Test Mixture. An example of data for an acetonitrile/water solvent system is shown in Figs. 10A and 10B where the analyte signal (LC peak height normalized to 37.5°C) is plotted against % (v/v) of acetonitrile. Analyte LC peak heights were normalized to the response obtained from the mixture prepared in pure water (0% organic). As can be seen, for most analytes, the collection efficiency generally increases as the organic content of the matrix is increased from 0% to about 20% acetonitrile, where maximum response was observed. The relative collection efficiency begins to generally and rapidly decline above acetonitrile concentrations of approximately 20% . In general, the analyte response is less than that from pure water when the acetonitrile content exceeds approximately 60% .

The extent of signal enhancement, as well as the organic solvent content at which maximum signal is attained, is analyte dependent. For example, benzene exhibits a maximum

response enhancement of about 10% relative to pure water. This occurs when the sample matrix contains 10%) acetonitrile. On the other hand, the collection efficiency or response is increased approximately 200%) (relative to pure water) for 1 ,2,4,5-tetrachlorobenzene and pentachlorobenzene when the sample solution contains 20% acetonitrile. Similar ennancement effects were also observed in the acetone/water, methanol/water and 2-propanol/water solvent systems. Figs. 1 1A and 1 1 B show tne relative analyte peak heights for pentachlorobenzene and fluoranthene plotted as a function of the effective Hildebrand solubility parameter (HSP) of the sampie solution. The signal intensities are normalized to the signal obtained with a pure water matrix (HSP = 48 Pa ^{1 ,2} ). Particularly, these plots show results of the foil owing solvents:

(Symbol) Fig. 11 A (Symbol) Fiq. H B a Pentachlorobenzene (ACN) e Fluoranthene (ACN) b Pentachlorobenzene (ACE) f Fluoranthene (ACE) c Pentachlorobenzene (MeOH) g Fluoranthene(MeOH) d Pentachlorobenzene (2-Prop) h Fluoranthene(2-Prop)

For a given analyte, the signai or collection efficiency trends appear to be dependent on the sample solution's HSP. The absolute extent of the solvent effect, however, depends on the specific organic solvent present in the sample solution. For example, as illustrated in the relative fluoranthene response shown in Fig. 1 1 B, maximum collection efficiency was observed for all of the solvent systems when the effective solution riSP was in the range of between about 42 and 46. The absolute value of the collection efficiency maxima follow the order acetone > acetonitrile > methanol >2-propanol.

From the data presented on the effect of organic solvents in the sample solution, two major points may be gleaned. First, the RPM cell of the present invention is very effective at extracting the organic analytes of environmental interest from sample matrixes (solutions) which contain significant organic solvent content. This effect is unexpected, since one would always expect decreases in extraction efficiency relative to pure water as the organic nature of the sample solution is increased. These tests demonstrate that the collection efficiency can actually be enhanced in solutions which contain certain organic solvent content. Thus, an aqueous sample stream may be mixed with an organic solvent prior to contacting the RPM cell in orderto enhance analyte collection and increase sensitivity.

Second, since the collection efficiency of the analytes is dependent on the sample solution HSP, sample streams may be modified with specific organic solvents in specific concentrations whereby the RPM collection may greatly favor specific analytes resulting in increased selectivity.

Referring now to Fig. 12, therein is shown a highly preferred apparatus of the instant invention which includes a 65 centimeter length of 0.53 millimeter inside diameter fused silica capillary with a 1 micrometer thick film or membrane of (5%- phenyl) -

methyl polysiloxane as a reversed oermeation membrane cell 427. One end of the ceil 427 is connected to a port of a six port switching valve 412. The other end of the ceil 472 is connected to a port of a six port analysis valve 423. The valve 423 is electrically heated, for example to a temperature of 120 degrees centigrade, by a heating element which is not shown. Two adjacent ports of the valve 412 are plugged with plugs 413 and 414 as shown. A seventh tube 442 is connected to one of the ports of the valve 412. Adjacent ports of the valve 412 are internally connected by first, second and third channels thereof 41 5, 416 and 417 respectively. Similarly, the adjacent ports of the analysis valve 423 are internally connected by first, second and third channels thereof 424, 425 and 426 respectively. Referring still to Fig. 12, one end of a length of a 0.2 meter long, 0.1 millimeter internal diameter fused silica capillary restrictor tubing 420 is connected to one port of the valve 423. Each end of loop of a 0.2 meter long, 0.1 millimeter internal diameter fused silica capillary tubing 422 is connected to adjacent ports of the valve 423 as shown. One end of a thirty meter long, 0.32 millimeter internal diameter fused silica capillary column 431 , having a stationary phase or membrane of one micrometer thick (5%- phenyl) - methyl polysiloxane, is connected to a port of the valve 423 as shown. The other end of the column 431 is connected to a flame lonization gas chromatography detector 433. The RPM cell 427, the valve 423 and the column 431 are within a gas chromatography oven 432 of a Hewlett-Packard 5890A gas chromatograph. The detector 433 is in electrical communication with a strip chart recorder 434 by way of wires 443 and 444

The RPM cell 427 is covered with a glass fiber sleeve, not shown, and then wrapped with a 0.287 millimeter wire 428 consisting of a Nichrome brand metal alloy reported to have a composition of 60 percent nickel, 24 percent iron, 16 percent chromium and 0.1 percent carbon. The wire 428 has a resistance of 24 ohms. The wire 428 is connected to a variable voltage power supply 430 via a circuit that includes an electrical switch 429. Glass fiber insulation, not shown, is then wrapped around the RPM cell 427 and wire 428. A J-type thermocouple, not shown, is inserted between the RPM cell 427 and the glass fiber sleeve so that the temperature of the RPM cell 427 can be monitored.

A cylinder of compressed helium 418 is provided as a source of carrier gas for gas chromatography. A gas pressure regulator 419, set at one atmosphere gauge pressure, is used to regulate the pressure of the helium carrier gas. One end of a third tube 438 is connected to a port of the valve 412 as shown while the other end of the tube 438 is connected to the regulator 419. Similarly, one end of a fourth tube 439 is connected to a port of the valve 423 as shown while the other end of the tube 439 is connected to the regulator 419. Fig. 12 also shows a six port sample injection valve 403. One end of a second tube

437 is connected to a port of the valve 403 as shown. The other end of the tube 437 is connected to the the remaining port of the valve 412. The ends of an injection loop 406 are connected to opposed ports of the valve 403 as shown. The internal volume of the loop 406 is

one milliliter. A reservoir 400 is provided for containing water 401. A sixth tube 441 connects the reservoir 400 with a metering pump 402 set to deliver 200 microiiters per minute. A first tube 436 connects the pump 402 with a port of the valve 403 as snown.

A syringe 407, containing a sample to be analyzed for its compounds of interest, is connected to a port of the valve 403 as shown. A waste reservoir 404 is provided to contain waste liquid 405. A fifth tube 440 connects the remaining port of the vaive 403 to the reservoir 404. The adjacent ports of the injection valve 403 are internally connected by first, second and third channels thereof 408, 409 and 410 respectively.

Referring now to Figs. 12 through 17, a highly preferred method of the instant invention will now be discussed beginning with Fig. 12. The syringe 407 contains a liquid sample containing at least one compound of interest. When the barrel o the syringe 407 is depressed, then sample flows through the channel 408, the loop 406, the channel 410, the tube 440 and then into the waste reservoir 404. Wash water 401 is pumped from reservoir 400 through the tube 436, the channel 409, the tube 437, the channel 417, the tube 442 and then to waste in the direction of the first arrow 41 1. Helium is flowed through the tube 438, the channel 416, the RPM cell 427, the channel 426, the tube 420 and then to waste i n the direction of the second arrow 421. Helium is also flowed through the tube 439, the channel 425, the column 431 and then through the detector 433.

Referring now to Fig. 13, the channels 408, 409 and 410 of the valve 403 and the channels 415, 416 and 417 of the valve 412 have been rotated one position clockwise as shown. Now the water flowing through the tube 436 flows the sample held in the loop 406 through the tube 437, the channel 416, the inner surface of the RPM cell 427, the channel 426, the tube 420 and then to waste. When the sample is passing through the RPM cell 427, the membrane of the RPM cell extracts at least a portion of the compounds of interest therefrom. The pump 402 can be stopped when the sample is passing through the RPM cell 427 so that there is more time for a more complete extraction of the compounds of interest from the sample.

Referring now to Fig. 14, the channels 408, 409 and 41 O of the valve 403 have been rotated counter clockwise so that the water no longer flows through the loop 406. The timing of this step can be relatively short so that less than one milliliter of sample is directed to the RPM cell 427. However, the step shown in Fig. 14 is optional Thus, the valve 403 can be left in the position shown in Fig. 13 during the remaining steps of the method.

Referring now to Fig. 15, the channels 415, 416 and 417 of the valve 412 have been rotated counter clockwise so that the liquid contained in the RPM cell 427 is flushed out with helium. In this regard, the internal diameter and length of the tube 420 are best selected so that the liquid in the RPM cell 427 flows out to waste in a controlled and even manner without the formation of droplets of liquid in the RPM cel l 427.

Referring now to Fig. 16, the channels 424, 425 and 426 of the valve 423 have been rotated counter clockwise so that helium from the tube 438 flows through the RPM cell

427 and into the column 431. In addition, the electπcai switch 429 is losed so that the wire 428 can heat the RPM cell 427. The helium flowing through the RPM cell 427 is an extractant fluid for extracting the compounds of interest from the membrane of the RPM cell 427. The application of heat to the RPM cell 427 increases the rate of extraction of the compounds of interest from the membrane of the RPM cell 427 into the helium stream flowing mto the column 431. The temperature of the oven 432 is best set at a relatively low temperature during this operation so that the extracted components of interest focus at the inlet of the column 431. It will be noted that the helium flowing through the tube 439 now flows through the tubing 422 and the tubing 420 before going to waste. The internal diameter and length of the tube 422 is best selected so that the flow rate of helium through the tube 422 is not wasteful of helium.

Referring now to Fig. 17, the switch 429 is opened, the channels 424, 425 and 426 are rotated counter clockwise and the temperature of oven 423 is increased in a programmed manner so that the components of interest which were focused at the inlet of the column 431 are separated by temperature program gas chromatography in the column 431 and detected by the detector 433. The recorder 434 is connected to the detector 433 and the recorder 434 draws a chromatogram 435 of this separation and detection.

Referring now to Figs. 12-17, the valves 403, 412 and 423 and the switch 429 are best controlled by an event timer that is integrated with the gas chromatograph. In this manner, the apparatus and method can be automated. Similarly, the syringe 407 can be replaced with an autosampler so that many samples can be analyzed in an automated manner. When compounds are being analyzed that are not highly retained by the membrane of the RPM cell 427, then it is best to proceed from the step shown in Fig. 14 to the step shown in Fig. 15 as soon as the sample has been flushed out of the RPM cell 427 and it is best to proceed from the step shown in Fig. 15 to the step shown in Fig. 16 as soon as the liquid is flushed out of the RPM cell 427. On the other hand, when the components of interest are highly retained in the membrane of the RPM cell 427, then it is best to extend these operations to better flush the sample from the RPM cell 427 with water or other wash liquid and to better flush the water or other wash liquid from the RPM cell 427 with helium so that less water or other wash liquid is introduced into the column 431.

The film thickness of the membrane used in the RPM cell 427 and the temperature of heating of the RPM cell 427 are best adjusted according to the characteristics of the components of interest to be analyzed. The temperature of heating of the RPM cell 427 is best kept to the minimum needed to give good peak shapes on the chromatogram 435. If the components of interest are so highly retained in the membrane of the RPM cell 427 that an excessive temperature of heating of the RPM cell 427 is needed, then the use of a thinner membrane, such as a 0.1 micrometer thick membrane, should be considered. On the other hand, a 5 micrometer thick membrane can be considered if the compounds of interest are not

highly retained by the membrane of the RPM ceil 427. Some components of interest such as highly polar ones, will not be retained sufficiently by many types or thicknesses of memorane.

A specific example of the system of Figs 12-17 will now be discussed for compounds of interest that are relatively highly retained by the membrane of the RPM cell 427. 5 The oven 432 temperature program is set to remain at 50 degrees centigrade for 16 mm. and then to rise at 16 degrees centigrade per minute to a temperature of 280 degrees centigrade and to hold this temperature for 4 minutes. The voltage of the supply 430 is set so that the temperature of the RPM cell 427 will come to about 280 degrees centigrade after the switch

429 is closed. An event timer is used that causes the system to progress from the steps of: Fig. 10 12 to the step of Fig. 13 at time 0.01 minutes; Fig. 13 to Fig. 14 at time 5 minutes; Fig. 14 to Fig.

15 at time 7 minutes; Fig. 1 5 to Fig. 16 at time 8; Fig. 16 to Fig. 17 at time 16 minutes; and then Fig. 17 to Fig. 12 at 34.8 minutes. Test samples are prepared containing chlorobenzenes and PAH's in water and analyzed using the system. The results show a relative standard deviation of better than 15 percent relative at concentrations in the 100 parts per billion (1 ,000,000,000)

1 5 range and a recovery of from about 23 percent for fluorantnene to about 70 percent for tetrachlorobenzene, each at the 100 parts per billion concentration range. Linearity of response was generally good from 500 parts per trillion ( 1 ,000,000,000,000) to 1000 parts per billion concentration.

Another specific exam pie of the system of Figs. 12-17 will now be discussed for

20 compounds of interest that are not highly retained by the membrane of the RPM cell 427. The oven 432 temperature program is set to remain at 30 degrees centigrade for 7 mm. and then to rise at 16 degrees centigrade per minute to a temperature of 150 degrees centigrade and to hold this temperature for 4 minutes. The membrane of the RPM cell 427 is changed to a 0.32 millimeter internal diameter, 5 micrometer thick, 65 centimeter length of five percent

25 phenylmethyl silicone polymer supported by a fused silica support. The voltage of the supply

430 is set so that the temperature of the RPM cell 427 will come to about 200 degrees centigrade after the switch 429 is closed. The event timer settings are changed so that the system progresses from the steps of: Fig. 12 to the step of Fig. 13 at time 0.01 minutes; Fig. 13 to Fig. 14 at time 2.5 minutes; Fig. 14 to Fig. 15 at time 3 minutes; Fig. 15 to Fig. 16 at time 5.5

30 minutes; Fig. 16 to Fig. 17 at time 10 minutes; and then Fig. 17 to Fig. 12 at 18 minutes.

Test solutions are prepared containing compounds of interest such as propylene oxide, hexane, carbon tetrachloπde, benzene, dichloroethane, propanol, perchioroethylene, toluene, ethyl benzene and styrene at concentrations ranging from 10 to 1000 parts per billion in water. Under these conditions, propylene oxide and propanol can not be detected, probably

35 because they are not sufficiently extracted from the sample by the membrane and/or because * they extracted out the membrane too rapidly. Hexane, carbon tetrachloride and benzene are - detected but with broadened peak shapes probably because they focus poorly on the column

431 even at the lower initial oven temperature of 30 degrees centigrade. Some Daseline upset

is noticed in the early part of the chromatogram 435 prooably because tne wasn water was not completely flushed from the RPM cell 427 prior to the step cnange of Fig. 15 to Fig 16 Under these conditions dichloroethane also could not be detected which was unexpected since its boiling point is dose to that of benzene which was detected. Clearly, the partition coefficient for the component of interest between the membrane and the sample solution is more important than its boiling point. The remaining components of interest mentioned atthe beginning of this paragraph are detected essentially without difficulty.

Having shown and described the preferred embodiment of the present invention, further adaptions of the apparatus and method for reversed permeation membrane extraction described herein can be accomplished by appropriate modifications by one of ordinary skill in the art without departing from the scope of the present invention. Many of the potential modifications and adjustments have been discussed herein, and others will be apparent to those skilled in the art. Accordingly, the scope of the present invention should be considered in terms of the following claims, and is understood not to be limited to the details of structure and operation shown and described in the specification and drawings.