Background of the Invention "Diesel particulate matter,"as specified by the Environmental Protection Agency (EPA) procedures and most other standards and regulations worldwide, is sampled by filtering diluted diesel exhaust at temperatures not higher than 125°F (52°C). Devices which are used in the laboratory to produce the mixture of air with diesel exhaust gas are known as dilution tunnels. The intention of this procedure is to simulate conditions at which diesel particulates are released from vehicles into the atmosphere. Fiberglas" filters, which are used for diesel particulate matter (DPM) sampling, capture the solid particles, as well as liquid droplets or mist, which may condense from exhaust gases during the dilution process. In effect, the definition of DPM extends to all solid and liquid material present in diluted (and cooled) diesel exhaust.

Diesel particulates are composed of elemental carbon particles (CP) which may agglomerate and adsorb other species, most importantly, raw fuel and incomplete lube oil combustion products, to form structures of complex physical and chemical properties. Diameters of the aggregated particles are typically in the range of 0.08 to 1 um. By chemical analysis, DPM is commonly divided into three fractions:

1. solid fraction--primarily elemental carbon (CP); 2. soluble organic fraction (SOF)--organic material derived from fuel and lubricating oil and extractable in a hydrocarbon solvent; and 3. sulfate particulates (SO9), hydrated sulphuric acid.

Diesel particulate matter sampling, according to the EPA and most other regulations, is done from diluted exhaust gas.

This is achieved by turbulent mixing of exhaust gases with air in a dilution tunnel. The dilution air is drawn through a polished and heated stainless steel tube, typically 250 to 300 mm in diameter, at a velocity of about 10 m/s. The exhaust gas from the engine enters the upstream end of the tube (the dilution tunnel) and turbulent mixing takes place, causing cooling of the exhaust and formation of condensates.

The full flow dilution tunnel is a bulky and expensive device.

Partial-flow mini-dilution tunnels have been developed to address this problem. Commercially available mini-dilution tunnels have been certified to be fully equivalent with the U. S. EPA full dilution systems under steady-state particulate sampling conditions. For transient particulate sampling, the full flow tunnels are still the only available alternative.

The Constant Volume Sampling (CVS) method has been adopted by most countries for exhaust emission testing. In the CVS method, the exhaust gases are diluted with filtered air to maintain a constant total flow rate (air + exhaust) under all running conditions. This is achieved either by a positive displacement pump or by making the air-gas mixture flow through a critical flow venturi nozzle.

The complexity of current sampling and monitoring procedures creates a need for a highly-sensitive, on-line, particulate monitor and method, useful for research purposes

by engine manufacturers (certainly diesel, and possibly gasoline internal combustion), government, university and private research laboratories and owners of large diesel fleets (bus authorities, trucking companies, etc.).

Further, an instrument with moderate cost could be used in vehicle inspection stations. Legal authorities could require vehicle emissions testing beyond that presently used for gaseous contaminants, and especially for particulate emissions from diesel engines in cars and trucks.

Use of RF Sensor Technology for Measurement of Particulate Loadings in Diesel Filters Diesel particulate filters, based on the use of ceramic monoliths, were first developed about 20 years ago. They operate by collecting DPM within the passageways of the wall- flow monolith filter. These filters must be periodically regenerated to avoid excessive pressure drop and the subsequent detrimental effects of exhaust backpressure on the diesel engine. Thermal regeneration of the filter requires some external source of energy to ignite the CP collected in the inlet side of the filter-usually with the filter isolated from the engine exhaust to minimize heating requirements. Once ignition of the CP has been accomplished, a flow of air is supplied to provide oxygen for the CP combustion process as it proceeds to burn like a wick down the filter passageways. Sufficient air must also be provided to remove the heat released by the highly exothermic oxidation of the CP. Without effective removal of the heat, thermally- induced stresses will fracture the ceramic filter or, in severe cases, cause local melting of the ceramic. The hot gasses passing down the filter passageways, ahead of the CP burn front, vaporize the SOF collected in the filter and transport the SOF out of the filter. As a result, the SOF

does not contribute significantly to the heat load in the filter during regeneration. Thermal regeneration of a monolith filter, therefore, requires a reliable means of determining the CP content of the filter in order: (a) to efficiently remove the DPM; and (b) to prevent damage to the ceramic filter due to thermal overload.

It has been proposed that the CP content of a ceramic monolith filter can be determined by measuring changes to the effective dielectric properties of the filter, caused by the amount of the DPM collected in the filter. The complex permittivity of a material is comprised of two components: a real component called the"dielectric constant"and an imaginary component called the"dielectric loss factor."The cordierite used to manufacture most monolith filters has a relatively low dielectric constant and a very small loss factor. This makes the cordierite filter virtually transparent to RF energy. In contrast, the CP has a relatively high dielectric constant and very high loss factor.

Based on these latter observations, three RF-based measurement methods have been proposed. 1t2t3 In the first method1, the metal containment housing the filter forms a RF wave-guide, which can be periodically excited by RF energy. This metal containment forms a resonant RF structure. The resonance characteristics of this structure are a function of the effective permittivity of the material contained within the ceramic filter and the DPM collected in the filter. Reflected RF power measurements are used to assess the resonance characteristics of the filter. This methodl claims to be able to correlate the amount of reflected power to the amount of DPM in the filter.

A second method2 also uses the metal containment housing the filter to form a RF wave-guide. However, this second

method uses an antenna at the inlet of the filter to transmit RF energy down the wave-guide and through the ceramic filter to a receiving antenna at the outlet end of the filter. The ceramic filter has a very low dielectric loss factor and hence very little attenuation of the RF power occurs (-3 dB) when the filter is clean. However, CP has a very high dielectric loss factor and readily adsorbs RF energy. This second method2 claims to be able to correlate the attenuation of the RF power transmitted through the filter with the amount of DPM collected in the filter.

A third method3 uses the same fundamental RF transmission measurement as above, except that both the transmitting and receiving antenna are inserted into the filter passageways.

The antennae are parallel to each other and are axially positioned within the metal containment. These antennae are spaced apart, and the RF energy is transmitted between the sections of the antennae where they axially overlap. Where the antennae overlap, adjacent axial segments of a ceramic filter can be isolated by the method3 for RF interrogation.

This isolation provides greater measurement sensitivity in areas of the filter where DPM accumulation levels are most critical during thermal regeneration (i. e., typically, near the outlet end of the filter where thermally induced radial stresses are highest).

Ceramic-Membrane-Coated Particulate Filter The configuration of the CeraMemsceramic-membrane-coated filter is similar to other diesel filters. 1llll The filter uses a cordierite wall-flow monolith with square passageways, which extend from one end face to the other. The passageways are Goldsmith, et al., USP 5,114,581,"Eack-Flushable Filtration Device Method of Forming and Using Same", May 19,1992.

typically 0.075" (2mm) to 0.165" (=4mm) in size. For filters used for industrial and vehicular applications, the filter diameter is 6 to 12 inches; the filter length is 6 to 15 inches, and individual filter areas range between 25 ft2 and 70 ft2.

The major improvement associated with the CeraMem filter is that before plugging the passageways in the usual checkerboard configuration to form a dead-ended filter, a key additional step is taken. The surfaces of the"inlet" passageways of the filter are coated with a thin, fine-pored, ceramic particulate layer, which is fired to high temperature to bond the layer to the monolith. This"ceramic-membrane" layer functions as a surface filter (ca. 0.1-pm pore size) and keeps particulate from penetrating into the monolith wall pore structure. This surface filtration function allows the filter to be effectively regenerated by back-pulsing to remove captured particulate matter. But, because the membrane is very thin, added pressure drop is low.

Extensive testing for particulate removal from diesel engine exhaust for has been performed by Northeast University in the laboratory of Professor Yiannis Levendis. This includes multiyear testing of the system in over-the-road testing of a Volkswagen diesel Rabbit. The results have shown excellent filter regenerability and substantially complete soot removal in contrast to all other non-membrane-coated diesel filters evaluated. The results of these tests are reported in a series of SAE papers from the annual SAE conferences on diesel exhaust (some co-authored

111111 SAE #920567,"Development of a New Diesel Particulate Control System with Wall-Flow Filters and Reverse Cleaning Regeneration" SAE #930367,"Design of a Diesel Particulate Trap-Incinerator with Simultaneous Filtration and Compressed Air Regeneration (CAR)" SAE #930368,"On-Road Testing of a Reverse Air-Flow Cleaning, Soot-

with CeraMem) and a joint US patent between CeraMem and Northeastern.1111111 Separate, more quantitative tests at CeraMem on a 75 kW diesel genset have shown a soot removal efficiency >99.97%. 11111111 These tests were limited by the sensitivity of the gravimetric measurement (weight uptake of sampling filter).

In summary, the Northeastern and CeraMem tests have demonstrated the following performance characteristics of the CeraMem filter for filtration of diesel engine exhaust gas: The filter shows substantially complete removal efficiency for all particulates in diesel exhaust, including particulates <0.1 um.

The filter is readily regenerated (i. e., regeneration of clean filter pressure drop) by simple short (ca. 50-100 ms) backpulses with compressed air.

The filter can be operated hot (at exhaust engine manifold temperature) or cold, without adversely impacting soot removal efficiency or backpulse regenerability.

For the instrument disclosed herein, a small-size specimen of this filter is employed. It is encased in a metal

Oxidizing Diesel Particulate Trap System" SAE #940460,"Control of Diesel Soot, Hydrocarbon and NOx Emissions with a Particulate Trap and EGR" SAE #950370,"Diesel Vehicle Application of an Aerodynamically Regenerated Trap and EGR System" SAE #950371,"An Aerodynamically Regenerated Diesel Particulate Trap Coupled to an Electric Soot Incinerator with Dual Wall-Flow Filters" SAE #960473,"An Optimization Study on the Control of NOx and Particulate Emissions from Diesel Engines" SAE #970477,"An Integrated Diesel Engine ART-EGR System for Particulate/NOx Control Using Engine Sensory Inputs" 1111111 Levendis et al., USP 5,426,936,"Diesel Engine Exhaust Gas Recirculation System for NOx Control Incorporating a Compressed Air Regenerative Particulate Control System", June 27,1995.

Final Report,"Reducing Diesel NOx and Soot Emissions Via Particle-Free Exhaust Gas Recirculation", submitted by CeraMem to EPA under Grant No. 68D30118, June 1994.

sleeve, as is normally done for diesel filters. These components must be sized and of a design to satisfy the needs of a RF system to interrogate the filter such that good sensitivity in measuring diesel particulate emissions is realized.

Summary of the Invention The invention relates to an improved RF-based particulate measuring instrument and method for particulate- containing gases. More particularly, the invention concerns a real-time, sensitive, on-line RF-based measuring instrument and method for exhaust gases for carbon particulates, condensable organics, and sulfates.

The invention comprises a method of measuring the concentration of carbon particulates from an exhaust gas stream, which method comprises: extracting a sample of an exhaust gas; filtering the exhaust gas sample with a high efficiency particulate filter at a substantially isothermal temperature, above the condensation temperature of condensable organic liquids in the exhaust gas sample, to provide for a substantial, complete collection of carbon particulates in the filter and a filtrate gas; employing a RF-based sensor using a selected RF frequency, with transmitting and receiving antennae, in close proximity with the collected carbon particulate; integrating the changes in attenuation of the RW signals of the RF sensor over a time measurement intervals; and correlating the changes of attenuation to provide a determination of the concentration of carbon particulates, with time, in the exhaust gas.

The invention also comprises a real-time measuring system for the measuring of the concentration of carbon particulates from an exhaust gas containing carbon particulates, sulfates, and organic volatiles, which system

comprises: a means to extract a sample of the exhaust gas to be measured; a high efficiency carbon particulate filter to filter from and to collect in the filter, substantially all of the carbon particulates from the sample, during a measured time, and at an isothermal temperature of about 500°C, to provide collected carbon particulates and a filtrate gas; and a RF-based sensor with antennae in close proximity with the collected carbon particulates in the filter and employing selected RF frequencies; and also a means to correlate the changes in attenuation of the RE signal, to provide a measurement of the concentration of carbon particulates, with time, in the sample.

An underlying assumption in the prior art cited above is that the changes in the dielectric properties of the ceramic filter are dominated by the amount of CP collected in the filter and are not significantly effected by the amount of SOF also accumulated by the filtration process. This assumption is erroneous as will be demonstrated below.

The rate of DPM production and its composition are a complex function of number of parameters including: engine design; engine condition; fuel and lube oil composition; and engine operating conditions. The rate of accumulation and composition of the DPM in a ceramic monolith filter are a function of all the preceding parameters, plus filtration efficiency, and most importantly, the thermal history of the filter. To illustrate this latter point, samples of DPM were collected from the exhaust of a 2.3-L, 4-cycle, 4-cylinder Cummins Series A diesel engine under a range of steady-state engine load conditions. The weight percent of SOF and CP in the DPM were then determined by standard methods.

The percentage of the SOF in the DPM is plotted in Fig. 1 as a function of the internal temperature of the

monolith filter. As can be seen from Fig. 1, the relative amount of SOF in the DPM increases exponentially as function of decreasing filter temperature. While specific amounts of SOF will vary widely with engine and filter design and system operating conditions, the trend shown is representative of a widely observed phenomenon in engine exhaust filtration systems.

It is well-known in the art that SOF includes a wide range of materials. While diesel fuel has a boiling point of about 150°C, incomplete fuel and lube oil combustion products can have boiling points up to 400°C and 500°C. The ratio of SOF to CP will, therefore, depend not only on the time- integrated range of engine operating parameters, but most importantly, on the average and most recent thermal history of the filter.

Elemental carbon is used commercially as a RF energy adsorber. While its permittivity is an intrinsic physical property, its apparent permittivity is a function of its apparent density. In the case of engine exhaust CP, its permittivity depends on how densely packed the CP is on the filter ceramic membrane surface, and whether the individual carbon particulates are in intimate contact (i. e., the particles act like a series of interconnected carbon resistors). This concept is illustrated schematically in Fig. 2. Idealized carbon particulates 20 are shown collected on the surface of a ceramic filter membrane 21 in Fig 2a. In this illustration, surface coverage of the ceramic filter membrane will influence the degree of contact between adjacent carbon particulates. When carbon particulates 20 become coated with SOF 22, as illustrated in Fig. 2b, the SOF acts as an electrical insulator (e. g., like transformer oil) inhibiting electron transfer between adjacent particulates.

The apparent permittivity of the accumulated CP in the filter will be a function of the extent of the SOF surface coating on the carbon particulates. For example, keeping the total weight of DPM constant, but varying the SOF to CP ratio from 0.01 to 0.1, the transmission of RF energy through a cordierite monolith filter increases by 100%. This variation in CP dielectric properties, with changes in SOF concentration in the DPM, makes all three methods1 2, 3 cited as prior art unreliable as a means of determining accumulated levels of CP or DPM in diesel engine exhaust filters. Reliable measurements using these methods1 2 3 can only be made under conditions where the SOF can be eliminated from the CP or where the SOF to CP ratio remains constant.

The invention shall be described for the purposes of illustration only in connection with certain illustrated embodiments; however, it is recognized that various modifications, additions, improvements, and changes may be made by persons skilled in the art without departing from the spirit and scope of the invention as disclosed and claimed.

Brief Description of the Drainas Fig. 1 is a graphical plot of the percentage of SOF as a function of the internal filter temperature under constant diesel engine load conditions; Fig. 2a is a schematic drawing showing a deposit of idealized, spheroidal carbon particulates 20 deposited on the surface a ceramic filter membrane 21; Fig. 2b is a schematic drawing showing a deposit of idealized, spheroidal carbon particulates 20, coated with a layer of condensed organic material 22 deposited on the surface a ceramic filter membrane 21; Fig. 3 is a schematic drawing showing a longitudinal cross-section of a typical sample filter and metal housing

assembly, and the position of the antenna within the outlet filter channels, which is parallel to the inlet CP sampling channels, is also depicted ; Fig. 4 is a block diagram illustrating one embodiment of the RF sensor instrument for DPM measurements; Fig. 5 is a conceptual graphic plot of RF signal attenuation (TRANS) as a function of carbon particulate (CP) loading in a sample filter, and the impact of various concentrations of condensed organic material (SOF) on TRANS is also depicted; Fig. 6 is a simplified block diagram of a preferred embodiment instrument of the invention; Fig. 7 is a simplified block diagram of another instrument of the invention; Fig. 8 is a schematic drawing of RF energy propagating through the air and striking the surface of a solid; and the drawing depicts RF energy being reflected from the surface of the solid, attenuated as it passes through the solid, and phased shifted in space and time as it passes through the solid; Fig. 9 is a block diagram showing the main operations in a transmission type RF sensor, and, as indicated, the source RF signal is split into a reference and transmitted signal, and the amount of transmitted RF signal attenuation is measured relative to the reference signal in the comparator; Fig. 10 is a schematic drawing showing a radial cross- section through the sample filter, illustrating the position of the RF antenna relative to the sampling and gas outlet filter channels; Fig. 11 is a block diagram showing the interconnections between the network analyzer, the transmission/reflection test

set, the RF prototype sensor, and the PC controlled/data logger; Fig. 12 is a graphical plot of the measured signal-to- noise ratio over a range of frequencies between 0.3 MHz and 3000 MHz, as a function of the average transmitted power (TRANS) through the RF sensor; Fig. 13 is a graphical plot of transmitted power (TRANS), through the copper prototype sensor at 109°C, as a function of RF frequency, and separate plots for a clean filter, when the filter contained a carbon black load of 34.4 mg, are compared; Fig. 14 is a graphical plot of reflected power (REF), through the copper prototype sensor at 109°C, as a function of RF frequency, and separate plots for a clean filter, when the filter contained a carbon black load of 34.4 mg, are compared; Fig. 15 is a graphical plot showing the difference between TRANS frequency response curves, such as those illustrated in Fig. 13, and curves are plotted to show differences for various filter loadings with carbon black; Fig. 16 is a graphical plot of transmitted and reflected power at a frequency of 1.2 GHz and a temperature of 109°C, as a function of carbon black accumulated in the copper prototype sensor filter; Fig. 17 is a graphical plot of transmitted power (TRANS) as function of sampling time for three difference carbon particulate (CP) concentrations in an engine exhaust; Fig. 18 is a graphical plot of the rate of change of the transmitted power with accumulated weight of carbon black (dTRANS/dCB) as a function accumulated carbon black (CB); Fig. 19 contains a chart and a table showing the operating envelope for the RF sensor differential mode; and Fig. 20 is graphical plot of the transmitted power (TRANS) for the stainless steel prototype, at 510°C, as a function of the accumulated weight of carbon black.

Description of the Embodiments The invention disclosed in this patent application comprises a real-time particulate measuring instrument and utilizes a suitably sized, ultra-high-efficiency filter, such as, the CeraMemO ceramic-membrane coated filter. An example of an antenna configuration in such a wall-flow monolith filter is shown in the cross-sectional view in Fig. 3. These components, filter and antenna, must be sized and of a design to satisfy the needs of a RF system to interrogate the filter, such that good sensitivity in measuring engine particulate emissions is realized.

The application of RF sensing to analytical measurement of particulate mass differs in significant ways from the prior art1^2t3. These differences include the following: The prior art RF-based measurement methods were intended to measure CP accumulation in diesel engine exhaust systems.

As such, they had to measure CP accumulation under a variety of engine exhaust and filter temperatures. As discussed immediately above, this wide range of temperatures results in a highly variable range of time-variant SOF concentrations relative to the CP in the filter. This latter condition prevents a reliable measurement of CP concentration in the exhaust filter using the methods described'. In the present invention, the CP is collected at a constant temperature above the condensation temperature of SOF, thus allowing a reliable measure of the CP concentration in a ceramic filter.

The antenna configurations used in the prior art measurement methods2, were designed to decrease measurement sensitivity in order to measure large concentrations of CP in the exhaust filters. In contrast, the present invention uses antenna configurations designed for maximum sensitivity, in

order to measure extremely low concentrations in a small sample filter used in the present invention.

The filters used in the prior artl-2, 3 were not intended as absolute filters, such as the ultra-high-efficiency CeraMems filter.

A block diagram of one possible arrangement of the system and method of the present invention is shown in Fig. 4. A sample of exhaust gas is extracted from the engine exhaust and split and filtered in two separate filter systems containing identical filters. Flow through each loop is controlled at equal levels by a mass flow controller on each loop.

The gas flow to the upper system is heated, and the collection filter is maintained isothermal and hot (e. g., 500°C), so that only CP is collected. The RF signal attenuation from this filter is correlated with the CP mass collected. By monitoring"differentially" ; that is, measuring change in attenuation over a measurement time interval, one obtains the mass of CP collected in a fixed volume of sampled gas for that time interval. Thus, using the electronics of the instrument, one can convert collected mass in the filter to CP concentration in the gas volume sampled to obtain a CP concentration versus time curve. At the end of a filtration run, the hot filter is isolated and regenerated, for example, by backpulsing with compressed air, through the valves shown, to be readied for the next measurement cycle.

The cold filter operates in a similar mode. The RF signal attenuation is for DPM containing both CP and SOF. This signal, as a raw signal, cannot be directly correlated with DPM mass as the ratio of CP and SOF is unknown. In order to make a correction, one uses a correlation, which adjusts RF absorption as a function of DPM composition. More

specifically, this is a correlation of the type illustrated conceptually in Fig. 5.

The two filter systems will simultaneously measure RF absorption as a function of time. By detecting increases in RF absorption over short time segments, one has two measurements corresponding to CP and DPM masses collected.

One knows the CP mass from the hot filter, as it will track the unique correlation with RF absorption at a fixed temperature and RF frequency. The cold filter will have exactly the same mass of CP, but the RF absorption will be reduced due to the presence of SOF. From a quantitative correlation like that of Fig. 5, the amount of condensables present is determined. The total DPM is calculated by adding the CP and SOF masses. Since the exhaust gas flow is known, therefore DPM concentration, as well as those of the SOF and CP fractions, is determined.

At the end of an engine cycle test and filtration cycle, the cold filter is heated to 500°C by flowing hot nitrogen.

This will drive off the SOF leaving only the CP fraction. A reading at this point should confirm the reading obtained on the hot filter. Finally, at this point, the filter will be regenerated (while hot), for example, by backpulsing with back-compressed air, to prepare for another filtration.

At the end of a test, the instrument will give CP and DPM sampled over the engine operating cycle, and this will be an integral measurement of CP and DPM concentrations. In addition, all of the differential measurements will provide data on CP and DPM concentrations versus time during the engine test.

A variant of the above system is to utilize three filters. The third filter would operate at an intermediate temperature, e. g., 250°C, where lower temperature condensables

remain as vapours (e. g., uncombusted fuel). By developing appropriate algorithms between RF response and particulate matter collected under this condition, it is possible to use this method to measure DPM, SOL, the fractions of volatiles which condense at higher temperatures (incomplete combustion products of fuel and lubricant), and the fractions of volatiles which condense at lower temperatures (uncombusted fuel).

The above sampling and analysis scheme has been configured for a steady state test cycle, that is, one in which constant sampling flows are employed. However, this system can be readily adapted to determine CP and DPM emissions accurately during a transient test cycle. This can be achieved by the following means: the engine RPM will be monitored and a correlation exists between engine RPM and exhaust gas flow. Thus, by monitoring engine speed, exhaust gas flow can be determined. An analog signal of RPM can be used to modulate the mass flow controllers for the instrument, to maintain flows through the instrument directly proportional to exhaust gas flow. By this means, the exhaust gas sample can be extracted at a flow rate proportional to total exhaust gas flow. Thus, if the instrument sensitivity is such that real-time measurements are achievable, then the instrument will be able to do so during transient tests.

The instrument disclosed herein is capable of measuring the following: 'DPM; CP; All condensables (including unburned fuel); Low-volatility condensables (heavier fuel and lubricant partial combustion products, excluding unburned fuel);

Real-time measurements of the above; Integral measurements of the above; and Valid results for steady state and transient test cycles.

The instrument described above can utilize one or more filters, depending on the measurements to be made, with the complexity (number of filters) related to the need to determine DPM, CP, and various condensable fractions.

Further, the instrument described above has the filters regenerated by backpulsing with compressed air.

Alternatively, the CP can be removed by flowing an oxygen- containing gas through the filter at a temperature of about 650°C through the filter to oxidize the collected CP.

Fig. 6 is a simplified, block flow diagram of a preferred embodiment of the instrument of the invention, wherein an exhaust gas sample is filtered at 500°C, CP measured by a RF sensor, and total hydrocarbons in the hot gas filtrate then measured with a hydrocarbon vapor measurement method, such as flame ionization detector (FID).

Fig. 7 is a simplified, block flow diagram of another embodiment of the instrument of the invention. As in Fig. 6, an exhaust gas sampled at 500°C is filtered, and the CP measured by a RF sensor. An exhaust gas sample is also filtered at 52°C and then measured for organic volatiles, while another sample is filtered at 191°C and then measured for sulfates. In all the systems, the filters may be regenerated periodically between measurement cycles by gas backpulsing or oxidative thermal regeneration.

The instrument and method described above was conceived to monitor particulate emissions in diesel engine exhaust gas, but can be used to monitor particulate emissions in other

combustion gas sources, such as gasoline-fuelled, internal combustion engines.

The present instrument system comprises: one or more substantially RF transparent, wall-flow monolith particulate filters capable of retaining particulate matter emissions in a combustion gas; RF antennae suitably located to monitor RF energy absorption by particulates collected in such filters; electronic equipment capable of generating a suitable RF signal and measuring its energy absorption by the particulates; and electronics suitable for converting such RF signal attenuation to a measurement of particulate loading.

While a preferred embodiment of the filter will contain membrane coatings on a monolith filter passageway walls, this will not be required if the filter, without such membrane coatings, is suitably retentive of the particulates in the exhaust gas and is reasonable regenerated by backpulsing or thermal oxidation. Further, a preferred embodiment of the filter is in a wall-flow monolith structure, but other suitable filter element configurations can also be employed.

Further, analysis of the exhaust gas passing through the filter can be performed by gas phase analytical procedures, such as flame ionization detector (FID), gas chromatography (GC), and mass spectroscopy (MS) to measure organic volatiles or total hydrocarbons or sulfates, at selected temperatures, for example, organic volatiles at 500°C and sulfates at 191°C.

The invention provides a real-time, exhaust gas analysis instrument. The integration of a RF-based measurement technology for CP with other instrumentation to measure sulfates and SOF and volatile organic and materials provides a comprehensive analysis of engine exhaust emissions.

CARBON PARTICULATE MEASUREMENT Overview of Carbon Particulate Measurement Concept The CP measurement concept consists of two stages, i. e., CP concentration by hot filtration and measurement of the accumulated CP by a RF-based measurement method. CP concentration takes place in a ceramic-membrane-coated filter previously developed by CeraMem for quantitatively complete particulate removal from diesel exhaust. This latter filter is essentially transparent to RF energy, whereas, CP is not.

A RF-based measurement method can, therefore, continuously measure changes in the CP concentration in the ceramic filter at 500°C. For a constant gas-sampling rate, measured changes in the CP accumulation rate can be related to changes in CP concentration in the exhaust gas.

RF Measurement Fundamentals RF-based measurement methods are based on detecting changes in the dielectric properties (permittivity) of a material. Permittivity is a complex parameter consisting of a real component, the dielectric constant, and an imaginary component, the loss factor. For example, when a RF signal is broadcast through air and strikes a solid, three phenomenon are observed (see Fig. 8). Firstly, part of the signal is reflected if the dielectric constant of the solid differs significantly from that of air. Secondly, if the loss factor is significant, then the signal is attenuated (i. e., part of the RF energy is converted to heat-the basis for microwave heating). In the present application, the loss factor for the filter is very low, while that for the CP is relatively high.

The amount of attenuation as the RF signal passes through the CeraMems filter is, therefore, primarily a function of the amount of CP collected in the filter. Thirdly, the

propagation velocity of the RF energy varies as the square root of the dielectric constant. A RF-signal passing through two media with different dielectric constants, will undergo a phase shift. In general, RF-based sensor technology can be based on measuring one or more of the following: 1. the relative amount of power reflected (REF) by a material; 2. the relative amount of power transmitted (TRANS) through a material; or 3. the phase shift in a RF signal transmitted through a material.

From a technical point of view, the simplest and the cheapest RF-based sensors are based on measurements of transmitted or reflected power.

Description of RF-Based Measurement Concept The block diagram, in Fig. 9 provides the major functions in a RF-based sensor for measurement of particulate concentrations in diesel exhaust, after-treatment filters.

In this type of sensor arrangement, the transmitted RF power is measured and compared to the reference incident power measurement. The signal is then correlated with the particulate concentration levels in the ceramic filter. This type of circuit arrangement has been retained in the instrument of the present invention.

For a small filter specimen to be used in an instrument, interactions between the metal enclosure and the RF antenna can create signal distortions. It is necessary to utilize antenna configurations that eliminate of control such interactions. An example of a useful antenna array in a monolith sensor filter is shown in Fig. 10.

The ceramic filter used in this device is of the wall- flow monolith-type. Only two of the inlet channels are used

to collect the CP, but other configurations are possible.

Transmitting and receiving antennae are inserted into unused adjacent channels as shown in Fig. 10, This type of antennae arrangement concentrates the RW power transmission through the CP collection channels, maximizes measurement sensitivity, and minimizes metal wall distortions.

EXPERIMENTAL MEASUREMENTS Measurements were made using both low-temperature and high-temperature prototype filter cavities. All RF measurements are made using a Hewlett Packard Model 8753C vector network analyzer and a Hewlett Packard Model 85055A transmission/reflection test set. The network analyzer was interfaced to a PC for recording and processing of measurement data. This set-up allowed evaluation of REF1 and TRANS2^3 measurement methods described in prior art. A block diagram is provided in Fig. 11 illustrating the instrumentation set- up. Simplified versions of this instrument have been built and previously used on diesel vehicles to measure CP concentrations in exhaust after-treatment filters1llllllll.

Weighed aliquots of carbon black (CB) were used as a substitute for engine CP in these preliminary experiments.

Carbon particulate diameters were on average 30 nm and the SOF content was 1.5% by weight.

Measurements Using the Low-Temperature Prototype The following discussion presents experimental results obtained using a low-temperature (109°C), copper prototype filter cavity. m""" Walton Frank B., et al., SAE Paper 910324,"On-Line Measurement of<BR> <BR> Diesel Particulate Loading in Ceramic Filters", presented at 1991 SAE Conference, Detroit, MI, February 25-March 1,1991.

Determination of Electronic Measurement Limitations Measurements were undertaken to determine the useful dynamic range of the measurement electronics. This involved making repeated measurements (n = 15) of the sensor response over a wide range of frequencies and signal levels. The signal-to-noise level for a given frequency was then calculated from the ratio of the sample average to the sample standard deviation. The results of these determinations are shown in Fig. 12.

Sensor Response to Varying Carbon Black Load Experiments were conducted to determine sensor response to varying CB load. Figs. 13 and 14 illustrate typical changes in the transmitted (TRANS) and reflected (REF) frequency response of the sensor to the accumulation of CB in the sample filter. In the plots shown, the response spectrums prior to CB addition (i. e., a clean filter) are compared after a number of sequential additions of CB totally about 27 mg.

This measured response is roughly equivalent to sampling an exhaust with a CP concentration of 100 mg/m3 for 30 min. As can be seen from Figs. 13 and 14, the measured response is a function of frequency.

Selection of Measurement Frequency Analysis of the aforementioned broad-spectrum curves, yields information used to select frequencies that provide maximum measurement sensitivity as the quantity of CB collected (load) in the sample filter increases. Subtracting TRANS curves at different CB loads allows a determination of the difference in transmitted power at each frequency.

Maximum measurement sensitivity occurs when the absolute difference in the two curves is greatest. Fig. 15 illustrates this process for three measurement ranges. As can be seen

from Fig. 15, frequencies in the range of 1.1 GHz to 1.2 GHz give the maximum measurement sensitivity under the conditions of these measurements.

Sensor Response at a Fixed Frequency The TRANS and REF response curves for the sensor, as a function of CP load at 109°C, are shown in Fig. 16. As can seen from this plot, both TRANS and REF are quite linear at CB loads less than about 20 mg (about 20 minutes of engine exhaust sample time). Above 20 mg, the sensitivity decreases with load and will require a separate correlation. The high sensitivity (linear) portion of the curve (<20 mg) would be best suited to engine transient response measurements. The lower sensitivity portion of the response curve (>20 mg) would still be suitable for longer-term, time-integrated emission measurements. Since TRANS measurements show the highest measurement sensitivity, all further discussion focuses on this method of measurement.

Assuming an exhaust gas-sampling rate (V) of 160 cm3/s, an equivalent CP accumulation rate in the sample filter can be calculated for a range of CP concentrations in an engine exhaust gas. Accumulated CB in Fig. 16 can thus be converted into total sampling time (6dt). Fig. 17 shows the measured TRANS as a function of total sampling time for three different CP concentration CP ranges. As can be seen from Fig. 17, the shape of the TRANS response curve remains the same as shown in Fig. 16. The TRANS curve is shifted in time with particulate concentration, because accumulated CP (6CP) in the sample filter is 6CP = V [CP] 6dt.

Determination of Sensor Measurement Sensitivity Estimates were made of the operating envelope of the sensor to determine whether the RF sensor concept was

achieving its target performance objectives. The measurement concept has two modes. The first mode is the time-average mode, where the average CP concentration is CP =OCP/ÒVdl.

Figs. 16 and 17 define the operating envelope of the sensor with respect to both total accumulated CP and total sampling times. That is, the sensor prototype has a maximum 6CP of about 30 mg. The total sampling time will be a function of [CP] or bdtmaX = (30 mg)/6V [CP] dt. From Fig. 17, it can be seen that total sampling times range from about 18 minutes at a CP of 200 mg/m3 to over 70 minutes at a CP of 50 mg/m.

The second mode is the differential mode. In this mode, a desired objective is to reduce 6dt to 5 seconds or less.

The measurement sensitivity of the sensor changes as CB or CP accumulates. That is, dTRANS/dCB varies as a function of 6CB, as illustrated in Fig. 18. As can be seen from Fig. 18, dTRANS/dCB decreases by a factor of 10 as the 6CB increases from 1 to 10 mg.

In the differential mode, the sample interval (At) will be defined by the instrument signal-to-noise ratio. At 1.2 GHz, the measured standard deviation for TRANS is about 0.02 dB. For the purpose of defining the operating envelope of the sensor in the differential mode, the minimum change in TRANS (ATRANS) has been conservatively set to 0.1 dB or 5 times the electrical noise level. That is, the operating envelope is bounded by a set of conditions where ATRANS > 0.1 dB for a specified At. Minimum CP and maximum differential mode operating times (ÒdtmaX) are summarized in Fig. 19 as a function of sampling interval (At).

The objective of the differential measurement mode is to minimize the time interval over which the particulate

concentration is averaged. In the limit, as At-0, the calculated concentration approaches a real-time measurement.

Because differential measurements must be made under conditions where the measurement sensitivity is highest (see Fig. 19), total operating or run time times are considerably shorter than in the time-average mode. As can be seen in Fig. 18, there is a trade-off between At and OdtmaX. The relationship between bdtmaX is a function of dCP/dt, At and dTRANS/dCP at a given OCP. Low [CP] requires a high dTRANS/dCP, because dCP/dt is low. Conversely, a higher CP can take advantage of a lower dTRANS/dCP, because dCP/dt is higher. bdtmaX is proportional to ATRANS. Reducing ATRANS from 0.1 dB to 0.05 dB (about 2 times the noise level at 1. 2 GHz) will double bdtmaX- High-Temperature Measurements To verify the basic measurement concepts developed using the low-temperature copper prototype filter cavity described above; additional measurements were made using a stainless steel, prototype filter cavity with Inconel antenna of the same fundamental design as that illustrated in Fig. 3.

Aliquots of CB were injecte into the sample filter at a temperature of 510°C. A typical set of TRANS measurements is illustrated in Fig. 20. It should be noted that commercially available copper and stainless steel pipe differ in their internal diameters for the same nominal outside diameter, hence, it was not possible to reproduce filter cavity dimensions exactly in the construction of the low-temperature and high-temperature prototypes. Also, since permittivity varies with temperature, differences in measured TRANS response were anticipated, due to the-400°C difference between the low-temperature and high-temperature experiments. Not withstanding these differences, it can be seen from Fig. 20, that the trend of TRANS versus accumulated CB in the sample filter is substantially the same as that for the copper prototype illustrated in Fig. 16.