GB2419421A

GB2419421A - Calibration of a mass flow sensor or controller using a variable gas flow source

Info

Publication number: GB2419421A
Application number: GB0526348A
Authority: GB
Inventors: Ali Shajii; Nicholas Kottenstette; Jesse Ambrosina; John A Smith
Original assignee: MKS Instruments Inc
Current assignee: MKS Instruments Inc
Priority date: 2002-06-24
Filing date: 2003-06-24
Publication date: 2006-04-26
Anticipated expiration: 2023-06-24
Also published as: GB2404028A; GB2419422B; GB2419957B8; GB2419420A8; GB2419676B; GB2404028B8; GB2419421B; GB0526337D0; GB0526343D0; WO2004001516A1; GB2419958B8; GB0526341D0; GB2419958A8; GB2419676A; GB2419422A; GB2419421A8; GB2419958B; GB2419422A8; GB2419957A; JP4544992B2

Abstract

A mass flow calibrator 706 includes a variable flow gas source 708, a receptacle of predetermined volume 710, and a pressure differentiator 712. The variable flow gas source 708 may provide gas at varying rates to the mass flow sensor 702 being calibrated and at proportional rates to the receptacle of predetermined volume 710. The pressure differentiator 712 produces a signal representative of the time derivative of gas pressure within the receptacle of predetermined volume 710 and, from that, the actual flow into the receptacle. Given the actual flow, the proportionate flow into the mass flow sensor 702 may be determined and the flow signal from the mass flow sensor 702 correlated to the actual flow.

Description

GAS FLOW STANDARD

The present invention relates to mass flow sensing and control systems.

Capillary tube thermal mass flow sensors exploit the fact Slat lost transfer to a fluid flowing in a laminar tube from the tube walls is a function of mass flow rate of the fluid, the difference between iLe fluid tenperature and the wal] temperature, and the specific 1leat of the fluid. Mass flow controllers employ a variety of Lass flow sensor configurations. For example, one type of construction invo]>res a stainless steel flow sensor tube with one, and more typically two or snore, resistive elements in thermally conductive contact Title the sensor tube.

The resistive elements are typically composed of a material having a high temperature coefficient of resistance Each of the elements can act as a heater, a detector, or both. One or more of the elements is energized Title electrical current to surlily heat to the fluid strearr through He tube. If the 1eatcrs are supplied with conskant current, tile rate of fluid mass flow tlroup,rh the tube can be derived Torn t.cnperature differences in the elements. Fluid repass flow rates can also be derived by wearying the current through the heal.ers to maintain a constant temperature profile.

Such thermal class floNv sensors may be attached as a part of mass fluter 0 controller, with fluid from the controller's main channel feeding the capillary tube (also referred to herein as the sensor tube). The portion of the main channel to Rich the inlet and outlet of the sensor tube are attached is often referred to as the "bypass" of the flow servitor. Many applications employ a plurality of mass ilov controllers to regulate the supply of fluid through a supply line, and a plurality of the supply lines may be "tapped off' a main fluid supply line. sudden change in flow to one of the controllers may create pressure fluctuations at the inlet to one or more of the other controllers tapped off the main supply line. Such pressure fluctuations create differences between the flow rate at the inlet and outlet of an affected mass flow controller Because thermal mass low sensors measure Low - 2 at the inlet of a mass Dolor controller, but outlet flow from the controller is the criticalparameter for process control, such inlet/outlet flow discrepancies can lead to significant process control errors.

In a semiconductor processing application, a process tool may include a plurality of chambers with each chamber havin, one or more mass flow controllers controlling the flow of gas into the clamber. Each of the mass flow controllers is typically re-calibrated every two Greeks. The recalibration process is described, for example, in U.S. Patent 6,332,348 B l, issued to Yelverton et al. December 25, 2001, which is hereby incorporated by reference. In the course of such an "In Situ" calibration, covcntional methods require a technician to connect a mass flow meter in line with each of the mass flow controllers, flow gas through the mass flow meter and mass flow controller, compare the mass flow controller reading to that of the mass flow meter and adjust calibration constants, as necessary Such painstaking operations can require a great deal of tinge and, due to labor costs and the unavailability of process tools, with which flee mass floss controllers operate, can be very costly.

mass flow sensor cleat substantially eliminates sensitivity to pressure varia.lions would therefore be highly desirable. A convenient calibration method and apparatus for mass flow controllers should also be highly desirable. More flexible access to a mass flow controller would also be highly desirable.

Apparatus and method for increasing the control pcrfonnance of a mass flow controller would also be highly desirable.

Sumlllal v of the InveIllion According to the present invention there is provided a mass flow controller and a system, defining a flow path between a gas inlet and an outlet of the system as set forth in the appended claims.

Brief Description of Drawings

Figure 1 is a block diagram of a system that includes a mass flow sensor; Figure 2 is a sectional view of a mass flow controller that employs a mass flow sensor.

Figure 3 is a sectional view of an illustrative thermal mass flow sensor as used in conjunction with a pressure sensor to produce a compensated indication of mass flow through a mass flow controller; Figure 4 is a block diagram of the control electronics employed by an illustrative embodiment of a mass flow sensor.

Figure S is a flow chart of the process of compensating a thermal mass flow sensor signal.

Figure G is a conccE,tual block diagram of a web-cnablcd mass flow controller.

Figure 7 is a conceptual block diagram of calibrator such as may be employed with a mass how controller.

Figure 8 is a block diagram of a self-calibrating mass Bow controller.

Figure 9 is a graphical representation of flow and pressure curves corresponding to the process of calibrating a mass flow controller.

Figure 10 is a conceptual block diagram of a dual processor configuration such as may be used in a mass flow controller.

Figure 11 is a Cow chart of the general operation of a mass flow controller's non deterministic processor.

Figures 12A and 12B are flow charts of the general operation of a mass flow controller's deterministic processor, and Figures 1 3A through 1 3E are screen shots of web pages such as may be employed by a web server embedded within a mass flow controller.

Detailed Description of Disclosure

A mass flow sensor employs a thermal mass flow sensor to sense and provide a measure of the flow of fluid into an inlet of a fluid flow device, such as a mass flow controller. The mass flow sensor uses a pressure sensor to compensate the inlet flow measure provided by the thermal mass flow sensor to thereby provide an indicator that more accurately reflects the fluid flow at the outlet of tile associated mass flow controller.

A system 100 that benefits from and includes the use of a mass flow sensor is slown in the illustrative block diagram of Figure 1.

A plurality of mass flow controllers MFC1, MFC2, ... MFCn receive gas from main gas supply lines 102, 103. The mass flow controllers, MFC1, MFC2, MFCn are respectively connected through inlet supply lines 104, 106, ...109 to a main gas supply line 1()2, 103 and though respective outlet supply lines I 10, 1 12, ...1 15 to chambers C1, C2, Cn. In this illustrative embodiment, the tern1 "chamber" is used in a broad sense, and each of the ehambe?-s may be used for any of a variety of applications, including, but not limited to, reactions involved in the production of seniconduetor components. Generally, users of the chambers are interested in knowing and controlling the amount of each gas supplied to each of the chambers Cl, C2, ...Cr. Each chamber C1, C2, ... Cn may also include one or more additional inlet lines for the supply of another type of gas. Outflow from the chambers may be routed through lines (not shown) for recycling or disposal.

The mass flow controllers, MFC1, MFC2, MFCn, include respective mass flow sensors MFS1, MFS2, ...MFSn, electronic controllers EC1, EC2, ...Ecn and outlet control valves OCVI, OCV2, ...OCVn. At least one of the mass flow sensors is, and, for ease of description, assume all are, compensated mass flow sensors. Each mass flow sensor senses the mass of gas flowing into the mass flow controller and provides a signal indicative of the sensed value to a corresponding electronic controller. The electronic controller compares the indication of mass flow as indicated by the sensed value provided by the mass flow sensor to a set point and operates the outlet control valve to ninimize any difference between the set point and the sensed value provided by the mass flow sensor.

Typically, the set point may be entered manually, at the mass flow controller, or downloaded to the mass flow controller. The set point may be adjusted, as warranted, through the intervention of a human operator or automatic control system. Each of the inlet supply lines l 04, l 06, . . . l 09 may be of a different gauge, andlor may handle any of a variety of flow rates into the mass pow controller. A single electronic controller, such as electronic controller ECl, may be linked to and operate a plurality of mass flow sensor/outlet control valve combinations. That is, for example, any number of the illustrated electronic controllers EC2 through Ecn may be eliminated with the corresponding mass flow sensors and outlet control valves linked to the electronic controller ECl for operation.

An abrupt change of Now rate, due to a change in set point for example, into any of the mass flow controllers may be reflected as an abrupt pressure change at the inlet of one or more of the other mass flow controllers. This unwanted side effect may be more pronounced in a relatively low flow rate mass flow controller if the abrupt change occurs in a high flow-rate mass flow controller. Because the mass flow sensors iT1 this illustrative enbodin1ent are thermal mass flow sensors positioned to sense flow in the mass flow controller at the inlet to the mass flow controller, the mass flow sensed by the thermal mass flow sensor Nay not accurately reflect the flow at the outlet of the controller. In order to compensate for this discrepancy, a mass flow sensor includes a pressure sensor positioned to provide an indication of the pressure within the volume between the inlet and outlet of the mass flow controller. In an illustrative embodiment, the pressure sensor is located In the "dead volume" between the thermal mass flow sensor's bypass and the outlet control valve. An electronic controller employs the indication of pressure provided by the pressure sensor to compensate the measure of mass flow provided by the thermal mass flow sensor.

The resultant, a compensated mass flow indication, more accurately reflects the flow at the outlet of the mass flow controller and, consequently, this indication may be employed to advantage by a mass flow controller in the operation of its outlet control valve. A display may be included to display the sensed pressure. The display may be local, attached to or supported by the mass flow controller, or it nay be remote, at a gas box control panel, for example, connected to the mass flow controller through a data link.

In a semiconductor processing application, a process tool may include a plurality of chambers with each chamber having a plurality of mass flow controllers respectively controlling the flow of constituent gases into the chamber. Each of the mass flow controllers is typically re-calibrated every two weeks. The re-calibration process is described, for example, in US Patent 6,332,348 Bl, issued to Yelvcrton et al. December 25, 2001, which is hereby incorporated by reference. ln the course of such an "In Situ" calibration, conventional methods require a technician to connect a mass flow meter m line with each of the mass flow controllers, flow gas through the mass flow meter and mass flow controller, compare the mass flow controller reading to that of the mass flow meter and adjust calibration constants, as necessary. Such painstaking operations can require a great deal of time and, due to labour costs and the unavailability of process tools Title which the mass flow controllers operate, can be very costly. In an illustrative embodiment described in greater detail in the discussion related to Figure 7, a mass Bow controller includes a self-calbrating mechanism that substantially eliminates such tedious and costly chores.

The sectional view of Figure 2 provides an illustration of a mass flow controller 200 that employs a mass flow sensor 202. The mass flow sensor 309 includes a thenlal mass flow sensor 204, a pressure sensor 206, a temperature sensor 208 and an electronic controller 210. A laminar flow element 212 establishes a pressure drop across the capillary tube of the thermal mass flow sensor 204, as will be described in greater detail in the discussion related to Figure 3. In operation, a fluid that is introduced to the mass flow controller 200 through the inlet 214 proceeds through the bypass channel 216 containing the laminar flow element 212. A relatively small amount of the fluid is diverted through the thennal mass flow sensor 204 and re-enters the bypass channel 216 downstream of the laminar flow element 212. The electronic controller 210 provides a signal to the control valve actuator 2] 8 to thereby operate the outlet control valve 220 in a way that provides a controlled mass flow of fluid to the outlet 222. The pressure sensor 206 senses the pressure within the volume within the bypass channel 216 between the laminar flow element 212 and the outlet control valve 220, referred to herein as the "dead volume 216a".

As will be described in grreatcr detail in the discussion related to Figure S. the electronic controller 210 employs the pressure sensed within the dead volume 216a by the sensor 206 to compensate the inlet flow rate sensed by the thermal mass flow sensor 204. This compensated inlet flow rate figure more closely reflects the outlet flow rate, which is the ultimate target of control. In particular, a mass flow sensor is a combination sensor that employs the time rate of change of pressure within a known volume 21Ga to provide a precise measure of mass flow during pressure transients and a thermal mass flow sensor that may be "corrected" using the pressure-derived mass flow measurement. Both the thermally-sensed and pressure derived mass flow measurements are available for processing. The temperature sensor 208 senses the temperature of the fluid within the dead volume. In an illustrative embodiment, the temperature sensor 208 senses the temperature of a wall of the controller, as an approximation of the temperature of the fluid within the dead volume 216a.

The volume 216a of the dead volume is determined, during manufacturing or a calibration process, for example, and may be stored or downloaded for use by the electronic controller 210. By taking sequential readings from the pressure sensor output 206 and operating on that data, the electronic controller 210 determines the time rate of change of pressure within the dead volume 216a. Given the dead volume, the temperature of the fluid within the dead volume, the input flow rate sensed by the thermal mass flow sensor 204, and the time rate of change of pressure within the dead volume, the electronic controller 210 approximates the fluid flow rate at the output 222 of the mass flow controller 200. As previously noted, this approximation may also be viewed as compensating the mass flow rate figure produced by the thermal mass flow sensor 204.

The electronic controller 210 employs this computed output fluid flow rate in a closed loop control system to control the opening of the mass flow controller outlet control valve 220.

In an illustrative embodiment, the value of the pressure sensed by the pressure sensor 206 may also be displayed locally, (that is, at the pressure sensor) ardor remotely (at a control panel or through a network interface, for example) In a self-calibratilg process described hereinafter in the discussion related to Figure 7, the electronic controller 210 may take the time derivative of the pressure signal The the floivrate varies in the mass flow controller and thereby derive the actual floss, rate into the mass flow controller.

The actual flow rate may then be used to calibrate the mass flow controller.

The sectional view of Figure 3 provides a snore detailed view of a thermal mass flow sensor 204, such as may be employed in conjunction with a pressure sensor to produce a corngensated mass flow indication that is, in a digital implementation, a n1ulti-bit digital value. The multi-bit digital value provides a closer approximation to tile actual mass flow at the outlet of a mass flow controller than an uncompensated mass pow sensor would, particularly during pressure transients on the mass flow controller inlet lines. The thermal mass flow sensor 204 includes lar.. inar flow element 212, which rests within the bypass channel 916 and provides a pressure drop across the bypass channel 016 for the thermal mass flow sensor 204 and drives a. portion of the gas through the sensor capillary tube 320 of the thermal Nash flow sensor 204. The mass flow sensor 202 includes circuitry that senses the rate of flo;v of gas through the controlle and controls operation of the control valve 220 accordingly. The thermal mass Ilow sensor assembly 204 is attached to a wall 322 of the mass flow c.nntrcller No that for s a boundary of the hypas,q channel 2] 6 Input 324 and output 326 apertures in the wall 322 provide access to the thermal mass flow sensor assembly Q04 for a gas travelling through the thermal mass flow controller and it is the portion of this passageway between the input and output that typically defines the bypass channel. In this illustrative embodiment the mass flow sensor assembly 204 includes a baseplate 3S for attachment to the va.ll 300. The baseplate 328 may de attached to the wall and to the remainder of the sensor assembly using threaded hole and mating bolt combinations, for example.

Input 330 and output 332 legs of the sensor tube 320 extend through respective input 334 and output 336 apertures of the baseplate 39:8 and, throu,l apertures 324 and 326, the mass flow controller wall 3Q0.

The mass flow sensor assembly preferably includes top 338 and bottom 340 sections that, when joined, form a thermal clamp 341 that holds both ends of the sensor tube 320 active area (treat is, the area defined by the extremes of resistive elements in thermal contact with the sensor tube) at substantially the same temperature. The thermal cleanup also fortes a chamber 342 around the active area of the sensor tube 320. That is, Lhe segment of the mass flow sensor tube within the clamber 342 is in thermal communication with torso or more resistive elements 344, 346, each of oldish may act as a locater, a detector, or Moths One or more of the elements is energized with electrical current to supply heat to the fluid as it streams through the tube 320. The thermal clamp 341, which is typically fabricated from a material characterized by a high thermal conductivity relative to Lhe thermal cor.ductivity of the sensor tube, makes good thermally conductive contact with tile portion of the sensor tube just downstream Tom tile resistive element 344 and with the portion of the sensor tube just Upstream loom the resistive element 346. The l. herTa.l clamp thereby encloses and protects the resistive element 344 and 346 and the sensor tube 320. Additionally Lhe thermal clamp 341 thermally "anchors" tl-use portions of Lhe sensor tube with which it haloes contact at, or near, Lhe ambient temperature. let order Lo eliminate even minute errors due Lo ten1perat.re dt1-erenLials, Else sensor tube may be moved within the thermal clamp Lo insure that any difference between iDe resistance of tile two coils is due to fluid flout through the sensor tube, not to temperature gradients imposed upon the coils from the environment.

In this illustrative embodiment, each of the resistive elements 344 and 346 includes a thermally sensitive resistive conductor that is ovoid around a respective portion of the sensor tube 320. Each of the resistive elements extends along respective portions of the sensor tube 300 along an axis defined by the operational segment of the sensor tube 320 Downstream resistive e]emei1t 346 is disposed downstream oftl1e resistive element 344 The elements abut one another or are separated by a small gap for manufacturing convenience and are preferably electrically connected at the center of tle tube. Each resistive element 344;346 provides an electrical resistance that diaries as a Unction of its temperature. The temperature of each resistive element varies as a function of the electrical current flowing through its resistive conductor and tile mass flow rate within the sensor tube 390. In this way, each softly resistive elements operates as both a heater and a sensor. That is, the element acts as a 1leater that generates heat as a Unction of tle current through the element and, at the same time, the element acts as a sensor, allowin, the temperature of the element to be measured as a Unction of IO its electrical resistance. The thermal mass Dow sensor 204 may employ any of a variety of electronic circuit., typically hi a Wheatstone bridge arrangement, to apply energy to the resistive elements 46 and 344, to measure the temperature dependent resistance changes in the element and, thereby, the mass flow rate of fluid yessing througl the sensor tube 320. Circuits employed for Allis purpose are disclosed, for example, in U.S. Patent 5,46I,gl3, issued to Hinlle et al and U S. Patent 5,41O,912 issued to Suzal.i, both of which are hereby incorporated by reference in their entirety In operation, fluid flows from the inlet 214 to the outlet 222 and a portion of the fluid f lows through the estrictive lamincar flout element 9] 2. 'I'le retraining and proportional amount of fluir! flows through the sensor tube 320.

The circuit (not slovn here) causes an electrical current to blots tlrougl1 the resistive elements 344 and 346 so that the resistive elements 344 and 346 generate and apply heat to the sensor tube 320 and, thereby, to tle fluid flowing lln ough the sensor tube 320. Because the upstream resistive element 346 transfers heat to the fluid before the Quid reaches flee portion of the sensor tube 320 enclosed l:y Ll1e downstream resistible element 344, the fluid conducts more heat away from the upstream resistible element 346 rl1a.n it does fro n1 the downstream resistive element 344. The difference in the amount of heat conducted away front the two resistive elements is proportional to the mass flow rate of fluid within the sensor tube and, by extension, the total nears flow rate through the mass how rate controller 200 from the input port 214 through the output port 222. The circuit measures this difference by sensing the respective electrical resistances of resistive elements 344, 346 and generates an output signal that is representative of the mass flow rate through the sensor tube 320.

The conceptual block diagram of Figure 4 illustrates the architecture of an electronic controller 400 such as may be used in a mass flow sensor. In this illustrative embodiment, the controller 400 includes sensor 402 and actuator 404 interface Among the sensor interfaces 402, a flow sensor interface 408 operates in conjunction with a mass flow sensor to produce a digital representation of the rate of mass flow into an associated mass flow controller. The controller 400 may include various other sensor interfaces, such as a pressure sensor interface 410 or a temperature sensor interface 411. One or more actuator drivers 412 are employed by the controller 400 to control, for example, the opening of an associated mass Row controller's output control valve. The actuator may be any type of actuator, such as, 1 5 for example, a current-driven solenoid or a voltage-driven piezoelectric actuator.

The controller 400 operal:es in conjunction with a mass Hove controller to produce a digital representation of the rate of mass flow into an associated mass flow controller. A thermal mass pONV controller, slcl1 as described in the discussion related to Figure 3, may be employed to produce the mass HOW ncasurenent. Tle controller zl00 may employ a pressure sensor interface 410 to monitor the pressure of fluid within all associated mass flow controller In an illustrative embodied, a pressure sensor, sulk as the pressure sensor 706 of Figure 9, provides a measure of the pressure within the mass no\v controller.

More specifically, in this illustrative embodiment, the sensor measures the pressure within dead volume of the mass flow controller. In an illustrative embodiment, the mass flow controller pressure thus measured may be displayed, at the pressure sensor 206 or at the controller housing, for example, or song other location.

The controller 400 may convert the pressure measurement to digital form and employ it in analysis or other fullctiolls. For example, if the mass flow controller employs a thermal mass flow sellsor, tile controller 400 may use the mass flout controller pressure measurement to compensate for inlet pressure transients Although a temperature sensor interface may be used to obtain a temperature reading front a temperature sensor attached, for example, to the wall of a mass flow controller, a separate temperature sensor may not be required for each mass flow controller. For example, mass flow controllers are often employed, as described in greater detail in the discussion related to Figure 1, in conjunction with a semiconductor processing tool that includes a number of mass flow controllers and other devices that are all linked to a controller, such as a Workstation. The processing tool is operated Willie a carefully controlled environment Lest features a relatively stable temperature. Because the temperature of the fluid within the mass how controller is very nearly eclual to ] 5 that of the wall of the enclosure arid the wall of the enclosure is very nearly the temperature of Lhe room within which the tool is housed, a temperature leasuremert frown, for example, Lhe workstation that controls Lhe tool, may provide a sufficiently accurate estimate of the gas tenperature within the mass flow controller. Consequently, in addition to, or instead of, employing a separate temperature sensor on eacl mass fioNv controller, the temperature may be obtained frown another sensor Within tl-e same environment as the news flout contZ-ol]er: one located at a. workstation, for example.

lithe controller 400 includes a local user interface 416 that may be used Title one or more input devices, such as a keypad, keyboard, mouse, trachball, joy stick, buttons, touch screens, dual inline packaged (DIP) or tlunb-w1eel switches, for example, to accept input from users, such as technicians Who operate a mass flow controller. The local user interface 414 may also include one or more outputs suitable for driving one or more devices, such as a display, which may be an indicator light, a character, alphanumeric, or graphic display, or an audio o.tput device used to connunicate ilfornlation from a mass flow controller to a user, for example. A comurications interface 416 permits a mass flow controller to communicate with orate or snore other instruments, and/or with a local controller, such as a workstation Blat controls a tool that employs a plurality of mass flow controllers and/or other devices in tle production of integrated circuits, for example.

In this illustrative example, the communications interface 414 includes a DeviceNet interface DeviceNet is known and discussed, for exanp]e, in U.S. Patent No. 6,343,617 B1 issued to Tinsley et al. February 5, 9002, which is hereby incorporated by reference. Tl1e controller 400 also includes storage 418 ilk the form, for example, of electrically, erasable programmable read only n1en1ory (EEPROM) that may be used to store calibration data, mass flow controller identification, or code for operating the mass flow controller, for example.

Various other forms of storage, such as random access memory (RAM), may be employed. The storage can take rrany forms, and, for example, may be distributed, with portions physically located on a controller "chip" (integrated circuit) arid oilier portions located cff-chip. The controller 400 employs a data }processor 420, Nvl1icl, nighL Lal;:e the Corers of an arithmetic logic unit (ALU) in a general purpose microprocessor, for example, to reduce data. For example, the data processor 420 may a.vcr-ag;e readings received at the sensor inputs, determine '0 tle number of times a sensor reading has exceeded one or nrore ills eslold values, record the time a sensor reading remains beyond a tilreslold value, fir per otl1er forms of data logging.

Pressure transients on the inlet supply line to a mass flow controller 200 that employs a thermal mass tiow sensor 204 may create erroneous mass flowQ5 readings. Erroneous mass flow readings may lead, in turn, l.o improper control of a mass flow controller's outlet valve, which could damage or destroy articles being processed Title gasses under control of the mass flow controller. The digital representation of mass flow may talkie the form of one or more data values and is salt ject to fluctuations due to pressure transients on the inlet line of the mass flow sensor. 11, an illustrative embodiments the controller 400 employs data obtained at the pressure sensor interface 410 to compensate for fluctuations induced in a thermal mass flow sensor 204 by pressure transients on the mass fond sensor inlet line Q14. In this illustrative embodin1ert, tle controller 400 obtains temperature information through a temperature interface 411. The controller 400 employs the temperature, pressure, and mass foNv readings obtained from the respective interfaces, to produce a compensated mass fond reading that more closely reflects the mass flow at the outlet of the mass flow sensor than a reading from the thermal mass flow sensor alone provides. The controller 400 also provides control to sensors, as necessary, through flow sensor interface, pressure sensor interface, and temperature interfaces, 40S, 410, and 411, respectively.

The controller 400 also includes a valve actuator interface 404, which the controller 4-00 employs to control the position of a valve, such as the valve 020 of Figure 2, to thereby control the rate of fluid flow through a mass flow controller, such as the mass now controller 200, in a closed -loop control process. The valve ] 5 actuator may be a solenoid-driven actuator or peizo-electric actuator, for example. l ire co;l.roller 400 must be capable of operating with sufficient speed to read the various sensor out:puLs, compensate as necessary, and adjust the mass flow controller outlet control valve 220 to produce a predetermined flow rate.

The noNv rate is predetermined in the scuse that it is "desired" in sone sense. It is not predetermined in the sense that it must be a static setting. That is, the predeten1Jicd flow rate may be sot by an operator using a. meclianica] means, such as a dial setting, or may be downloaded fiom another controller, such as a N70rkstation, for example, and nary be updated.

In all illustrative enbodiinent, llie conLoller 100 employs rea:ling;s from o5 the pressure interface 410 to compensate floNv neasuren1ents obtained at the mass flow interface 408 from thermal mass flow sensor 204 that senses mass flow at the inlet _14 to mass flow controller 000 The compensated flow neasuren1ent more accurately depicts the pony at the outlet 222 of the mass flow controller 200.

This outlet flow is the flow being directly controlled by the mass flow controller 200 and typically is the flow of interest to end users. Emp]oyirlg a pressure compensated flow measurement in accordance with the principles of the present invention improves the accuracy of a mass flow sensor's outlet flow reading and thereby permits a mass flow controller to more accurately control the flow of fluids. That is, at equilibrium, mass flow at a mass flow controller's inlet is equal to the mass Dow at the outlet of the mass flow controller, but during inlet or outlet pressure transients, the flow rates differ, sometimes significantly. As a result, a mass flow controller that provides closed loop control using its inlet flow to control its outlet flow may commit substantial control errors.

The steady state mass flow in the capillary sensor tube 320 of a thermal mass flow sensor such as described in the discussion related to Figure 3 is generally described by the following equation: 0C = C__: P) 32IPRL Lo) Nowhere: do = capillary tube inside dianet.er Lc - -- capillary tube length pi = the density of tlc gas at the inlet pR = the density of the gra.s at standard temperature and pressure r = tl,e gas ViSCoSity Pi = the pressure at the inlet of the mass flow controller o0 Po = The pressure at the outlet of the mass flow controller P = the pressure in the dead volume of the mass flow controller The total flow through the mass flow controller is related to that through the capillary sensor tube 390 through a split ratio: a_ QBP/Qc Where QBP is the flow through tile bypass cl1amlel 216 and Qc is the flow through the capillary tube 320. The total flow Q1 at the mass flow Qcontroller inlet 214 is: Qi - QBP +Qc = (1+ a)Qc If floss remains laminar in both the bypass and capillary, the split ratio will remain constant. When the inlet pressure varies with tine, the nature of tle inlet pressure transient and the pressurization of the dead volume govern the flow at the inlet. Assigning that all thermodynamic events within the dead volume occur at a constant ternperatw-e that is equal to the temperature of the enclosure that forms a partial receptacle arid the dead volume, the mass conservation within the dead volume may be described by: Qo = Q _ TRV dP Where: PR = pressure at standard 1;eml:,erature and pressure (760 Torr.) OR = I:emperature at standard i.enperature and pressure (273:E() Ale = wall tenperal.ure (. emperatuie ofthe wall of the mass flow controller) V = volume of the dead volume Q; = inlet flow to the mass flow controller QO = outlet flow from the mass flow controller A mass flow sensor employs the relationship of equation (2) to compensate a thermal mass flow sensors mass lUow signal and to thereby substantially reduce errors in mass how readings during pressure transients.

The flow chart of Figure 5 depicts the process of compensating a thermal mass flow sensor reading.

The process begins in step 500 and proceeds from there to step 502 where a mass flow sensor's controller, such as the controller 400 of Figure 4, obtains a mass flow reading. This reading may be obtained from a thermal mass flow sensor through a flow interface, such as interface 408 of Figure 4, for example.

This now measurement reflects the rate of mass flow at the inlet of a mass flow controller and, as previously described, may not adequately represent the mass flow rate at the outlet of the mass flow controller. The mass flow rate at the outlet of a mass flow controller is generally the rate of interest for use in control applications Consequently, a mass flow controller compensates for the inaccuracy inherent in assuming that the inlet flow rate to a mass flow controller is equal to the outlet flow rate from the mass flow controller. From step 502 the process proceeds to step 504 where the sensor controller 400 obtains the temperature of the flow within the bypass channel. The temperature could be obtained through a temperature interface such as interface 412 of Figure 4, or it may be downloaded to the compensated mass flow sensor. The compensation process may safely assume that the gas temperature is equal the temperature of the enclosure of the mass flow controller. Additionally, in most applications, the temperature will remain relatively stable over a long period of time, so that a stored temperature value may be employed, with updates as necessary.

lA star obt..,iliilg the gas tenpcrattirc in stop 501 the process proceeds to step 506 sphere the sensor controller oltains the volume ofthe dead volume. This value may have been stored during rranufacturing, for example Front step 506 tile process pioceeds to step 508 where the pressure within the dead volume is obtained over a period of tine. Tile number Of ne.asurements and the time over which the measurements are made depend upon the speed and duration of transients at the inlet of the mass flow controller. In step 510 the processor employs the pressure measurements made in step 508 to compute the time rate of change of pressure within the dead voluble After computing the time rate of change of pressure within the dead volume, tle process proceeds to step 512 inhere a compensated outlet flow value is computed according to equation (2).

Simplifications may be made in the computational process. :For example, the volume of the dead volume, standard temperature, and standard pressure may all be combined into a single constant for use with the inlet flow measurement and time rate of cleanse of pressure within the dead volume to compute a compensated outlet flow approximation This simplification should yield an equation ofthe form: Qo = Qi - Cl(V/T) (dP/dt) (3) 1 0 where: Qo = the compensated sensed outlet floral rate, Qi = the sensed inlet flow rate, C1 = a normalizing constant relating the temperature arid pressure to standard temperature and pressure N = the volume between the sensor bypass and the outlet how control valve, T = the temperature of the fluid within the volume, dP/dt = time rate of change of pressure within else volume.

0 As previously noted, the;olui-ne V coulc: be folded into the constant (: I. Front step 512 the process proceeds to step 514 where it continues, with the flow sensor's controller obtaining pressure, I:ernperature, and flow readings and computing a compensated outlet flow estimate, as described. The process proceeds from step 514 to end ire step 516, for example, when tile mass flow sensor is shut down.

Returning to the block diagram of Figure 4, in this illustrative embodiment, the controller 400 includes a diagnostic interface 422 that permits an operator, such as a teclicia.n for example, to not only initiate, but conduct diagnostic tests on the mass flow controller. Furthermore, flee interface 429 permits the operator to Conduct the diagnostics in a negater that requires no input from the local system controller, which may be a workstation, Flat otherwise normally controls the mass Dow controller. Such diagnostics are transparent to the local system controller, which nay root even be made aware of to diagnostics being performed and may, eonsequertly, contiriue its operations unabated. The diagnostic interface provides access to mass flow controller sensor measurements, control outputs and mass flow controller diagnostic inputs and outputs These various inputs and outputs may be exercised and measured through the diag20stie interface Edith very little delay. In an illustrative dual processor cm'Dodime'.t described in greater detail in the description related to the discussion of Figure9, a deterministic processor mat, modify outputs and/or monitor inputs, frown sensors or test points, for example. During tle execution of online c2'iagnosties, the controller cont:iues to execute its process control functions, u2li2npeded, wllile, at the sane Lime, the controller may provide real 1> time int.eraetion with a t.eelnieia2 (i e, interactions wherein the delays are imperceptible to a. human operator) either locally or llrou,h a telecom2nuni cation s connection.

Using the diagnostic interface 422, an operator Can adjust eor.trol values, such as the set point, used to detennie Else mass 0'ov co2,L2oller's operation Additionally, the operator may modify sensor output values in order to test the mass [low contr-nller's response to specified sensor readings. That is, aid opualor can modify the sensor readings a mass how controller employs to control tile flow of gasses Ll-ougl2 its outlet valve arid, thereby, exercise the controller for diagnostic purpcrses. An operator may read all sensor and test point inputs as well S as information stored regarding control (stored by Else deterninisLic controller in the dual processor embodiment), read all sensor values, read test point values, read control information, sulk as the desired set point. Additionally, the operator may write to control outputs and test points and cver-Nvrite stored values, such as sensor readings or set point information in order to fully test the controller through the diagnostic port.

In an illustrative embodiment, a mass flow controller may include a web server. Such a web server may be included within the diagnostic interface, for example. In such an embodiment, the diagnostic interface includes a web-server that permits the mass flow controller to be used in a system such as illustrated in the block diagram of Figure 6. In such a system, a user, such as a technician, may employ a web-enabled device 600 such as a personal computer, personal digital assistant, or cellular telephone that runs a web Browser (e.g. Netscape (RTM) or Explorer (RTM) to communicate with a server 602 embedded in the mass flow controller 604. The server 602 includes web pages that provide an interface for the user to the mass flow controller 604. The discussion related to Figures 13A through]3E provide greater detail related to the web server capability embedded in an illustrative embodiment of a mass flow controller.

Mass flow sensors arc typically calibrated during Blair manufacturing process. Because a mass flow sensor is usually incorporated into a mass flow controller, this discussion will center on mass flower controllers, but the methods and apparatus discussed herein are applicable to "standalone" mass flow sensors as well. The calibration process requires a technician to supply a gas at a known flow rate to the mass flow controller and correlate the mass TIONV sensors Ilo\v sin,ol to the kno.. n flow rate. For c,ample, in tl,e case of a mass Bole sensor that provides a voltage output corresponding to flow, the technician maps the voltage output front the sensor into the actual flow rate. Allis process may be repeated for a plurality of flows in order to develop a set of voltage/flow correlations: for exanple, a revolt output indicates a JO standard cubic centimeter per ininute (scorn) Pow, a S Volt output indicates a 50 scam float, etc. Flow rates that fall between calibration points may be interpolated using linear or polynomial interpretation techniques, for example. This process may be repeated for several gases. Correlation tables Blat relate the signal front tile Class flow sensor (which may be a. voltage) to flout rates for various gases may thus be developed and stored. Such tables may be downloaded to a mass flow controller for use "in the field", or may be stored within a mass flow controller. Often, technicians calibrate a mass flow controller using a relatively innocuous gas, such as N2, and provide calibration coefficients that may be used to correlate the flout of another gas to the calibration gas. allele calibration coefficients may then be used in the field when a known gas is "flowed" through the mass flow controller to compute the actual flow from the apparent flow. Tla.t is, the apparent flock may be a flour correlated to N2 and, if Arsine gas is sent through the mass flow controller, the mass flow controller mtiltiplies the apparent flow by an Arsine gas calibration coefficient to obtain the actual flow. Additionally, once in the field, mass flow controllers may be re-calibrated Ott a regular basis to accommodate "drift", orientation, water content of a gas else flow of which is being controlled, or to compensate for other factors. U.S. Patent 6,332, 348 B1, issued on December 2S, 2001 to Yelverton et al, which is hereby incorporated by 1:S reference, discusses these factors, and flee unwieldy processes and equipment required to cany out these in-the-feld ca]ibraLions in greater:let.ail.

A calibration method and apparatus will be described in the discussion related to the conceptual block diagram of Figure 7. This calibration system and method Nay he employed in a manufacturing setting, or, in an illustrative embodiment, may be incorporated into a self-calibrating mass flow controller. The mass flow controller 7()0 includes a mass flow sensor 702 and an electronic controller 704 that receives a flow signal from the mass flow sensor 702. A calibrator 706 includes a variable flow gas source 708, a receptacle of predetermined volume 710, and a pressure differentiator 712. It should be noted that the dines separating different functional blocks are somewhat fluid. That is, in different embodiments, tile function associated with one block may be subsumed by one or more other blocks. For example, in an illustrative embodiment, the pressure differentiator 712 is implemented all, or in part, by the execution of code within the electronic controller 704. The variable flow gas source 708 provides a gas at proportional rates to both the receptacle of predetennied volume and the mass flow sensor.

The flow rate to the mass flow sensor 702 may be equal to the flow rate to the receptacle of predetermined volume 7IO: i.e., a proportionality constant of 1, for example. The mass flow sensor 702 is configured to produce a mass flow signal indicative of the clove that it senses and, in this illustrative embodiment, this signal is sent to the electronic controller 704. The pressure differentiator 712 produces a signal correlated to tile flow frown else variable flow source 708 into the receptacle of predetermined volume 710 according to the relationship of equation 4: Qo = Qi - C I (V/T)(dP/dt) (4) where: Qo -- the outlet flow rate in standard cubic cenlin?eters per minute, Qi = the inlet ilov rate in standard cubic centimeters per minute, IS Cl = a non1alizing constant relating the temperature and pressure Lo standard LeTperature and pressure V = the predetermined volume of the receptacle in liters, T = the Melvin temperature of the fluid within else receptacle, dP/dt -- time rate of change of pressure within the receptacle in Torlsecond In an illustrative embodiment, tle receptacle is closed arid gas flows into ibe receptacle until the pressure within the receptacle equals tl-at of gas supplied by the variable flow source 708 In such an illustrative embodiment, the variable flow source may be a constant-pressure source that, as pressure within the receptacle builds, supplies gas at an exponentially decreasing flow rate. In such a case, the outlet Bow Qo - 0, and flee inlet flow, Qiis given by: Qi = C (V/T)(dP/dt) (5) The pressure differentiator 712 takes the time derivative of the pressure within the receptacle 710 and, given the normalizing constant C1, the predetermined volume V, and the gas temperature vithill the receptacle, the differentiator (and/or the electronic controller 704) may determine the actual floor into the receptacle 710. Because the flow into the receptacle is proportional to the flow into the tlerma] mass flow sensor 70:, the actual flow into the thermal mass flow sensor 702 may also be determined by a multiplying the actual flow into the receptacle by a proportionality constant (e.g., the proportionality constant is 1 if the flows are equal). The signal from the mass flow sensor is then correlated, by the electronic controller 704 for exar1ple, to the actual flow, determined as just describecl. Such correlation relates one or more signal levels from the mass flow sensor to tilde actual flows. The pressure diferentiator 712 may include analog differentiator circuitry, for example, that takes the time derivative of the pressure signal. The clifferentiator output signal, a signal representative of the Lime derivative of the pressure within the receptacle dP/dt, may be scrupled by an analog-l:o-digrital co.verLer (not sloven) to permit the electronic controller 704, which maN', include a microprocessor, DSP chip, or dual processors, for example to operate on else time derivative signal Alternatively, the pressure differertiator 712 may convert the pressure signal to digital form for processing by the electronic comptroller 70zl, Chicle salves the Lime derivative of the pressure signal.

In such an enbocliment, the electronic controller, in conbiation with diffcrctiator code, operates as the differentiator. The controller employs at least two pressure differences divided by corresponding time intervals to cognate the derivative The gas may be supplied in parallel to the receptacle and mass flow sensor, or it may be supplied in series, as will be described in greater detail in the fling discussion related to a self-calibratig mass flow controller.

In operations, a mass flow controller may be calibrated as just described, using a plurality of gases, Nvit|l tle con-elation values (mappings of sensor output to actual IONS) stored in tables Calibration coefficients, relating flow measurements of one gas to another may also be developed and stored. The tables and/or coefficients may be downloaded to a mass flow controller in the field for use by the controller in controlling else flow of a gas. Various letdown interpolation teclmiques, such as linear or polynomial interpolation may be employed in conjunction with the calibration tables and/or coefficients. Additionally, such stored calibration tables and/or coefficients may be used as default values in a self-calibrating Class flow controller in accordance with the principles of the present invention A self-calibrating mass flow controller in accordance its the principles of the present invention includes a calibrator 706 and a mass flow sensor 702 which may be employed to calibrate the mass flow controller in a ] O manner as just described In the case of a self-calibrating mass flow controller, though, the calibration can De performed, In Situ, in the field just as readily as in a mauLacttnin', scenting.

Once istal]ed in else field, Ott a sem.iccJnductor processing tool as in the system 100 of Figure 1, for example, the mass flow controller can calibrate itself IS using the gas that is to be used during the semiconductor processing By Losing the gas that is to be used in processing, the mass flow controller may provide a more accurate flow neasurenent, because it will autorratically accommodate variations, such as moisture content' for example. Additionally, a new processing gas Nay be used JUSL as readily as a conventional gas, since the self-calibraLin' 0 mass flow controller play calibrate itself (tllal is, correlate mass flow signal levels to actual fluid levels det:eimined by the pressure differertiatcJr), on the gas to be used, not in relation Lo another, standard gas, such as No. Because the mass flow controller is calibrated in else orientation in which it will be used, discrepancies due to reorientation of the mass flow controller in the field relative to the position in which it Alas calibrated during manufacturing Still be substantially eliminated. All the mass flock controllers within a system such as system 100 of Figure I may be calibrated automatically and simultaneously, within n1onents.

This is in contrast to the cumbersome, painstaking process employed for conventional mass flow controllers, Plaice are typically individually calibrated by a teclllician eniploying multiple mass flow meters, going from mass flow controller to mass flow comptroller. As will be described in greater detail in the description related to the discussion of' Figure 8, a mass flow controller that includes a thermal mass flow sensor and a pressure transducer may shut its outlet valve to create a varying gas flow into it's dead volume. By taking the time derivative of the pressure the actual flow into the dead volume receptacle may be determined. The mass flow controller's correlation of the actual value of the flow to the thermal mass flow ser.-sor signal acts as the mass flow controller's calibration.

Figure 8 is a conceptual block diagram of a self-calibrating mass flow controller 800. In this illustrative, series-flow, embodiment, a gas flows through a thermal sensor 802 into a receptacle of predetermined volume 804, then through an outlet valve 806.

The outlet flow Qo would normally be a controlled flow into a cllanbcr, such as a chamber within an integrated circuit processing tool. An electronic controller lS SOB, which, in this illustrative embodiment, executes code to perfonn the differentiation required to obtain actual flow, as described in the discussion related Lo Figure 7, is in connur. icatio-l with the thermal sensor 8025 pressure sensor of,O5 and the outlet valve 806. In an illustrative process, the electronic controller Bloc, operates in conjunction with Ll1e outlet valve 806 to fond a Q0 N7ariable-LioNV gas supply. That is, Else electronic controller shuts the outlet valve, which causes the flow to decrease exporeiLial]y. Tile p, essu,e -withiri the dead volume increases, arid the electronic controller differentiates this signal a number of' times in order to obtain actual flow readings to correlate to the mass flow sensor signal values over a relatively broad range of flows. Additionally, in order to extend the period of tTine during which the flow is varying and to obtain actual flow values for correlation with tile thermal mass flow signal values over a broad range, tl,e electronic controller may open the outlet valve to a fully open position before closing it.

The pressure and flow profiles associated Edith such a process are il]ust. rated conceptually in the graph of Figure 9 At an initial time to the pressure difference between gas at the inlet to tJl.e mass flow controller Pin and the pressure Pr downstream in the receptacle 804 forces gas to floor through the mass flow controller at a rate Qin. Ire this example, the inlet pressure P.,, pressure within the receptacle PR, and flow tlrougl the input of the mass flow controller Q,, are constant At time tSo the controller shuts the outlet valve, thereby reducing outlet flow (go to zero. Gas continues to Zion into the receptacle as long as there is a pressure difference between the receptacle and the inlet. As the pressure PR withir. the receptacle rises exponentially toward an equilibrium state of equality witl1 the inlet pressure Pi,,, the inlet floNv Qi,. decreases By salving the derivative of the pressure change within the receptacle (also referred to herein as "dead volume" in association with an illustrative embodiment of the invention), the electronic corLrol]er may determine the actual floNv into the receptacle, as previously clescribed.

The electronic controller may correlate a plurality of simultaneous readings produced by the thermal mass flow sensor, lo thereby calibrate the mass flow sensor. I hat is, once this process is completed for a specific gas, flow readings from the Lliermal mass flow sensor may be conela.tcd Lo actual iTlONV rates file results may be employed by the electronic controller 808 to control the opening of tile valve 806 in a closed loop control system in order to deliver a selected ICONS down. sLream. In order to increase Else period of Lime Go drool Allen the controller shuts the valve, to the time at which the 110NAT beCOnJeS undetectable, and to thereby increase the number and precision of pressure measurements Blat may be made, the controller may open the valve completely before shutting it at time tso. Additionally, one or more flow restrictors may be placed in Llle flow path between the inlet to the thermal mass flow sensor and Lle inlet to the receptacle S04.

Tile conceptual block diagram of Figure 10 illustrates the architecture of a dual-processor embodiment of an electronic controller 1000 such as may be used in a mass flow sensor in accordance Nvith the princly]es of the present irNention In this illustrative embodiment, the controller includes two processors 1002, 1004. One of the processors 1002 is dedicated to "real time" processes and the other processor 1004 is dedicated to non-real time processes. By "real time" we mean processes that require a specified level of service withir. a bounded response tinge. In this sense, the processes are deterministic and the processor 1009 will be referred to herein as the deterministic processor. The objective of the dual processor architecture is to reduce the number of interrupts and manage asynchronous event r espouses in a predictable way. The non-deterninistic processor 1004 may handle event-driven interrupts, such as responding to input from a user. The deterministic processor 1002 handles only frame-driven, that is, regularly scheduled, interrupts. In an illustrative embodiment, the non- deterministic processor is a general purpose processor 1004, suited for a variety of tousle, such as user-ii1terface, and other, miscellaneous tasks, rather than a specialized co-processor, such as a math- or communications - coprocessor. In particular, a TMS320VC547], available from Texas Instruments, Inc., may be employed in a dual-processor enbodirnent in accordance with tle principles. The TMS390VC547l is described in a data manual, ava.ilab]e athLtp://v- s.ti.com/sc/ds/l:ms320vc5471 pdf, which is hereby incorporated by reference.

A processor interface 1006 provides for inter-processor communications.

l lie deterministic processor 1002, includes sensor and actuator interfaces. Among the sensor inLei-ces, a Gore sensor interface 1005 operates in conjunction with a mass flL'v:;uiSL' LL) puLluc;e a digital ICI: esenl:atioii of the rate of mass flu-,- in an associated mass flow controller. One or more actuator interfaces 1010 are employed by the deterministic processor 1002 to control the opening of An associated mass flow controller's output control valve or drive a diagnostic test pL>int, for example. The a.c;tulor nay be a. cur-ent-driven solenoid or a voltagedriven piezo-clectric actuator, for example. As will be described in greater detail in the discussion related to the flow chart of Figure 9, after initialization, the deterministic processor 1002 loops through a control sequence, gathering sensor data, gathering, setting inforinatior (for example, a desired mass flow setting), 30:>rordinu, status information, and coiollin&; the state of flee outlet \ralve Because non-determinstic tasks are offloaded to the non- deterministic processor 1004, the deterministic processor's control loop may be very compact.

Consequently, control tastes nary be executed within a minimal period of time and control readings and drive signals may be updated more frequently than possible if time were set aside for servicing non- deterministic tasks.

The controller IOOO operates in conjunction Fitly a thermal mass flow sensor as generally described in tle discussion related to Figure 3 to produce a diita.l representation of the rate of mass flow into an associated mass flow controller. The digital representations may take the Horns of one or more data values and its subject to fluctuations due to Pressure transients at the input of the mass flow sensor. The controller 1000, and mot-e specifically, flee deterministic processor 1002 nay employ data obtained at the pressure sensor interface 1006 to condensate for fluctuations induced in tile thermal mass flow sensor by pressure transients on the mass flow sensor inlet line. In Willis illustrative embodiment, the detemlinisLic processor 1002 employs the temperature, pressure, and Glass flow readings obtained from the respective 1008, 1007, and 1005 interfaces, to produce a compensated mass flow reading cleat 1nore closely reflects the mass flow at the outlet of the mass flow sensor than a reading from the Lhe'-na.l mass flout sensor alone. The deterministic processor 1002 also provides control Lo 0 sensors, as necessary, Llu-ougl thermal ilov:1005, pressure 1007, and temperature l()OC, sensor intuf.aces. Tlu t;ompeTsatiOn process will be described In greater deta.i] in tile discussion related to Figure 1]. The deterministic processor] 002 also includes a valve actuator interface lOlO, wick the deterministic processor employs to control the position of a valve, such as the valve 29() of Figure 9, to thereby control the rate of fluid flow through a mass flout controller, sulk as the mass flow controller 200, in a closed -loop control process.

The deterministic processor 1000 is devoted to the closed-loop valve control process, and, consequently, must be capable of operating with sufficient speed to read the various sensor outputs, compensate as necessary, and adjust the valve to produce a predetermined Ilow rate. Tile flow rate is predetermined in the sense that it is "desired" in some sense, and it need not be a static setting. That is, the predetermined flow rate may be set by an operator using a mechanical means, such as a dial setting, or may be downloaded from another controller, such as a workstation, for example, and updated frequently. Typically, gas 'flow control, and in this case, compensated gas Dow control, requires relatively hiah-speed operation. Various types of processors, such as reduced instruction set (RISC), Loath coprocessor, or digital sic,na] processors (DSPs) may be suitable for such high-speed operation The computational, signal conditioning, and interfacir.g capabilities of a DSP make it particularly suitable for operation as the I O deterministic processor 1002. As will lee described in greater detail in tile description of the control process related to tile discussion. of Figure 9, the function performed by the deterministic processor 1002 is deterministic in the sense that certain operations are completed in a timely and regular manner in order to avoid crrois, and possible instabilities, in the control process. The deterministic 1002 and nor.det;erministic 1004 processors communicate via the inl-er-pi-ccessor interface 1006 in a manner bleat does not impede else deterministic cooperation of the deterministic processor 1002. Inter-processor conunicat:ions are discussed in greater detail in tle discussion related lo lFig;ure 9.

The no-detcrn1inistic processor 1004 includes a local user inLeface 1016 tl1a.t may he used vit.h one or more input devices, sucl1 as a keypad, keyboard Denise, trckball, joy stick, buttons, touch screens, dual inlirie paclagcd (my') or thunb-wl1eel switches, for exanple, to accept input from users, such as t.ecimicians who operate a mass Ilow controller associated with the non- deterninistic processor 1004. rlle local user iterfa.ce 1016 also includes one or more outputs suitable for driving one or more dcdccs, such as a display, Rich man be a character, alphanumeric, or graphic display, for example, indicator light, or audio output device used to communicate inforn1atioi front a mass flow controller to a user. A communications interface 1018 permits a mass flow controller to communicate with one or snore other instruments, and/or Judith a local cnutrolle.r, sT]ch as a vorkstation that controls a tool that employs a plurality of mass flow controllers and/or other devices in the production of integrated circuits, for example. In this illustrative example, the communications interface ISIS includes a DeviceNet interface. A diagnostic interface 1090 provides an interface for a technician to rul1 diagnostics, as previously described in relation to the diagnostic interface 422 of Figure 4. In an illustrative embodiment, the diagnostic interface includes an Ethernet interface and a web server.

The compactness of code forthe deterministic processor 1002 permits the deterministic processor to be highly responsive to input changes and to quickly modify actuator signals in response to those changes. This partitiorung of operations between deterministic and on-detern1inistic processors also eases the initial development of code, for both the deterministic and non-deterministic processors. For example, the deterministic code needn't respond to unscheduled extents, such as "mirroring" a Serfs requests on a display at a user interface, and the non-det:erlinist:ic code needn't break away Irom providing such user feedback in order to adjust an outlet valve control setting every fifty bus cycles. The partitioning between deterministic and non-dctenninistic also permits relatively simple revisions and upgrades If the Noble Air one processor must be revised or upgraded, the code for the other may require no revisions or only minor revisions.

In particular, the code for the deterministic processor may be more "mature", or fixed than that for the non-deterninislic processor; user interfaces, ;'mmunicaions and other similar [unctions tend to be upgraded more frequently t:hai1 the deLerrilinistic, mass flow control, Unctions.

Using, this illustrative dual-processor embodiment, a user interface may be updated viLliout any impact on the control f:urction code, for example. Revision and n3itenanc.e of mixed-node code (deterministic and nondeterministic code) would be a much more complicated and costly proposition than code partitioned in a mariner in accordance with the pricipies of the present invention. In an illustrative embocliinent the dual-processor controller 1000 nary by a hybrid processor that incorporates two processors on ogle integrates circuit. An integrated circuit such as the TMS320C5471 hybrid processor available from Texas Instrumer.ts (RIM) may be employed as the dual processors in accordance with the principles of the present invention The digital signal processing (DSP) subsystem of the chip, due to its math capabilities would be more suitable as the deterministic processor in such an application. The IC's dual-ported memory may be employed as the inter- processor interface, with the processors writing to and reading from memory locations set aside to act as "mail boxes" for the transfer of information, including data, commands, and command responses.

Such an inter-processor interface pennits the deterministic processor tot continue operating in a frane-driven mode while, at tle same tinge, allowing the deterrmnistic processor to play a role in diagnostics and calibration. Any request for sensor data front the nor.-deterrinistic processor may be picked up from the mailbox on one pass of the deterministic processor's control]ODp, with the readings deposited in the mailbox tire very next time through the loop. Diagnostic outputs marl be modified similarly. The deterministic processor may also operate in other, non-process oriented modes. For example, during a self-calibratior process such as previously described, the deterministic processor would no longer operate to maintain a set flow through the mass flow controller. In such a mode the deterministic processor would be occupied by shutting the mass flow controller-'s outlet valve, talking a plurality of time derivatives of tle pressure within the dead volume, computing the corresponding actual flow in the mass flow controller, and correlating the actual flout to the flow. signal produced by a Hernial mass flow sensor.

{he flow chart of Figure ll outlines the process of sensing and controlling the flow of gas through a dual processor mass flow controller.

The process begins in step I 1 O() and proceeds from there to step 11()2 where the controller is initialized. This initialization step may include the uploading of calibration values or a calibration sequence itself. Additionally, operating code for both the deterministic and nondeterministic processors 1004 may be uploaded at this point. In an illustrative embodiment, the non-deterministic processor 1004 may upload its own code and begin operating, then upload code for the deterministic processor 1002. In the process of uploading code for the deterministic processor 1002, the non-deterministic processor 1004 may select among a plurality of executable code sets to upload ' to the deterministic processor 1002, thereby tailoring the operation of the deterministic processor 1002. The non-deterministic processor I 004 may base this selection on switch settir.gs, corrmands from a local controller (e.g., a;vorkstation controlling the operation of a semiconductor process tool), or settings stored in non-volatile storage, for example. Such a selection permits a mass flow controller to be tailored to different flout control l O operations. For example, a techniciar may, by selecting among code sets, choose to operate the controller in a "pressure cotrcller" mode rather than a "mass flow controller" mode, and this selection may be made locally or remotely (i e., through a telecommunications linl;).

In step 1104 the non-deterrministic processor 1004 passes operating code IS and filial control settings to the deterministic processor 1002 Rich then begins operating in a manner described generally in connection with the:flov chart of Figure 12. Fron1 step 1104, the process proceeds to step 1106 where the nor.- dete.rnir.'istic processor 1004 services the local input/out interface. Such servicing may include reading various inputs, including keyboard, switch, or mouse inputs, and displaying inforrnc'Lion locally, Llrough LEDS, alphanumeric displays, or graluLical displays. Frorn step l1Ob the process proceeds to step 11()s inhere the non-deterministic processor 1004 services the conTunication.s interface. This servicing may include the steps of uploading control and sensor data to a vorlcstation that operates as the loca] controller of a semiconductor process tool, for example. Additionally, the non-deterministic processor 1007 Inky download updated settings Mom tile local controller.

I7rom step 1 I 08 the process proceeds to step I 110 where the nondeterministic processor 1004 services the diagnostic interface. Various diagnostic operations, such as set forth in the description related to else discussions of Figure 4, may be performed ifs this step. In an illustrative embodiment, tle mass flow controller includes a web server, which permits an operator to run diagnostics through a net;-vork such as the "world wide web. " Frorn step 1110 the process proceeds to step 1112 where the non-deterministic processor 1004 services the inter-processor interface 1005. :[:)uring "normal", non-diagnostiG operation, the:ondeterministic processor 1004 obtains readings from the deterministic processor 1002 and passes control information, such as a flout setting obtained through the communications interface, to the determir.istic processor. From step 1112, the process proceeds to continue the processes just set forth in step 1114. The process proceeds to end in step 1116 when the mass flow controller is turned off, for example.

As previously noted, the steps set forth in this and other flow charts herein need not be sequential and, in fact, a number of fimctions performed by the non- deterministic processor 1004 may be event-internpt-drixen and no predictable sequence may be ascribed to the non-detern1inistic processor's operation. Other processes, su<:;h as data-logging may be performed at regular intervals. The noncleLerminist.ic processor can support a two-vay socket connection to the deterministic processor through an Ethernet nei:vvorlc interface, for example, to provide a relatively direct connection between a remote user and the deterministic pl ocessor.

The flow chart of I;igure 12A-12B depicts the operation of the deterministic processor of a dual processor mass flow controller. In the context of this flow chart, it is assumed that an initialization process has taken place and that the deterministic processor is cycling through its control loop. The process begins in step 120(), Figure 12A, and proceeds from tl1ere to step 1202 where the deterministic processor determines whether it is to operate in its ''nomlal'' control capacity or whether it is to operate in an alten1ativc mode, such as a manual diagnostic mode or an automatic diagnostic mode, for example The deterministic processor bases this decision on information it obtains from the inter processor interface 1006. The deterministic processor services frame-driven, rather than even-driven interrupts, consequently, it regularly polls the inter-processor interface to obtain information sucl1 as this.

If the deterministic processor is to operate in its normal mode, the process proceeds from step 1202 to step 1204, where the deterministic processor obtains information Mom the inter-processor interface regarding the desired control settings. This information may be in the form of a desired flow rate received from a local controller, from a front pane] user interface, or through the diagnostic port 1020 for example. The deterministic processor may also transfer information, such as sensor data, for example, to tile non-deterministic processor through the inter-processor interface during this step. groin step 1204 the process proceeds to step 1206 Were the deterministic processor gathers data, from a variety or sensors for example. The sensors from which the deterministic processor obtains data may include a mass f10NV sensor (thermal or other type), a lengera.ture sensor, or a pressure sensor, for example.

] 5 Frorn step 1206 the process proceeds to step 1208, where the deterministic processor computes Lhe flow rate of material through the mass iloNv conLrol]er. In;m illustrative embodiment, the mass f1ON;r controller includes a thermal mass flow sensor and a pressure sensor configured to measure tle pressure wit tile clead volume between Lhe thermal mass flow sensor's bypass and Lhe mass flONv controller outlet valve, step 1212 In this embodiment, the deterministic processor may employ the nctlod described in relation to the discussion of Figure 5 to compensate a flow rate measured by a thermal mass flow sensor aL the inlet of Lhe controller to more closely approximate the O.oNv rate at the outlet of tile controller. In an embodiinent in whittle tile flow rate obtained fi om the sensor is not compensated. the process should proceed directly from step 1906 to step 12] 0, skipping the computational process of step 1208.

In step 110 the deterministic processor determli1es vhetl1er the flow rate computed in step 1208 (or read in step 1206) is equal to the desired flow rate indicated by the setting information obtained front tle nondeterministic processor via the inter-processor interface in step 1204. If the values are equal the deterministic processor continues the operation as just described, as indicated by the "continue" block 1214 (i.e., the deterministic processor retunes to step 1709 and continues to cycle tlrougl the loop). If the values a're not equal, the deterministic processor cormpL'tes an error signs] and employs the error signal to adjust the drive signal to the mass ilovr controller's outlet valve. From step 1212 the process proceeds to continue in step 1214. The process proceeds from step 1214 to end ire step 1716 veer. the mass flow controller is shut down or reset, for

example.

10If, in step 1202 the deterministic processor concludes that it is not to operate in the normal naode, the process proceeds through co1ecting box A to step 1218, Figure 12B, where the deterministic processor determir.es whether it is to operate in a diagnostic mode. The deterministic processor may obtain this information from the inter- processor interface. If the deterministic processor is to IS operate in a diagrostic mode, the process proceeds to step]220. In step 120 tile deterministic processor deLernies Which diagnostic mode it is to operate in.

Once again, this inforr.ation may he passed to tle deterministic processor through the inter-processor interface. In an ''automatic" mode, the deterministic processor acquires a sequence of diagnostic values from the inter-processor interface The sequence calf values is available at the interface for acquisition by the deterministic processor The dia.cr,non.tic valucs may be control outputs, [ult setting the opening of tle mass flow controller outlet valve or for setting test point drive values, for example, or the diagnostic values may indicate desired sensor readings or readings fi-om test points, for example. The diagnostic values may also indicate tile sequence in which else values are to be employed, in order to set test point driver values, then read test point outputs, for example. In a manual mode, diagnostic values are made available to the deterministic processor through the inter-processor interface one at a time. In an embodiment in Chicle the mass Ilow controller includes a web server, a technician may use a web enabled worlstatio:1 to contact the server ill tile mass Cow c ontroller. Once lillleci to the server, the technician may enter a valve setting command, by typing, selecting from a pull down Menu or cliclong on icon, for example. This single, setting, command would be received by the no- deterministic processor through its diagnostic port and passed to the detenni1istic processor through the inter processor interface.

In tle ma nual diagnostic mode the deterministic processor executes through whatever diagnostic values are available at the inter-processor interface, then returns to it's normal control loop. This could "override" a single control loop cycle if, for example, a single diagnostic value, such as a test point drive value, is presented to the deterministic processor or, if a sequence of diagnostic Revalues is presented to the deternir.istic processor, a number of control loop cycles may be overridden. In the automatic diagnostic mode a number of diagnostic values may be exchanged through the inter-processor interface in a period corresponding to a few control loop cycles, with a substantial number, on the ]5 order of at lest Len times as many, control loop cycles intervening between automatic diagnostic exchanges Diagnostic modes may be combined, for example, to produce an automatic active on-line diagnostic mode, for example. In an illustrative embodiment, a mass flout controller in accordance with the prircip]es of the present invention operates on a onemillisecond control loop cycle, during vlich it provides one percent of full-sca.le accuracy.

Keeping the various diagnostic modes in mind, and Icceping in mind that processes illustrated throu:,l1 the use of note charts may not be strictly linear processes and alLei-i1ative florins n1ay be inplenented within the scope of the invention, the diagnostic process will be described generally in rclaticn to steps 1220 tl-ough 1226. In step 1220 the deterministic processor acquires diagnostic statues froth the inter-processor interface. As previously noted, these values nay be for the deterministic processor to use as control outputs or they may indicate data that is to be acquired by the deterministic processor, fi-om a sensor, for example. Prom step 1220 the process proceeds to step 1202 where the deterministic processor processes the values acquired in step 1000, by changing an outlet valve actuator drive signal or transferring a sensor reading to the interprocessor interface, for example.

From step 1222 the process proceeds to step 1224 where the deterministic processor determines whether it has completed its diagnostic tasks. If it has not completed its diagnostic tasks, for example if it is operating in the automatic diagnostic mode and there are more values in a sequence of values to be retrieved from the inter-processor interface, the process returns to step 1202 and from there as previously described. If, in step 1224 the deterministic processor concludes that it has completed its diagnostic task, the process returns through connecting 10box B to step 1Q14 of Figure 12A. If the deterministic processor determines that it is not to operate in a diagnostic ode, the process proceeds from step 1218 where processor performs functions such as routine background operations, then proceeds to return t.hrougl connecting blocl: B to step 1214 and from there as previously described.

15The screen shots of Figures 13A L]-ough 13E illustrate a user interface SUCH as may ire made assailable for access to a mass flow comptroller in accordance with the principles of the present invention that includes a web server interface, such as the iu.erface 608 of Figure 6. In an illustrative embodiment the repass flour contrcller includes a web server, such as the server 602 of Figure 6. A user may employ the server locally, tl-ougl1 a local controller, or remol:cly, frown a web- eabled device, such as the device coo of Figure 6. In this namer, the same user interface may be employed for both remote and local interactions Keith tile news Lloyd controller. Detailed information regarding a mass fiov controller, such as model number, range, and manuLcturin=, setup parameters, may be displayed to a IS user and user- changeable setup paramctci-s may be displayed as well. Different display techniques nosy be employed If there are only a limited number of acceptable values, they may be displayed and chosen from a pulldown noes, for example As previously described, a user, sulk as a technician can change set point values, open or close a valve, or monitor flow output, for example, through this interface. Additionally, while the mass f10NV controller is operating under a process control application, a user may induce the server to plot arid log pa'rameter values obtained Prom the mass flow controller.

The screen shot of Figure 13A illustrates the display a user may encounter when first accessing a mass flow controller in accordance with the principles of the present invention over the wed. The display prompts the user to choose a communications protocol through use softly pulldoNvn window 1300. The "query devices'' link- 1302 allows the riser to initiate a process whereby his browser attempts to locate all devices that it recognizes.

Basic information navy be downloaded through tile server. Information related to the mass flow controller are displayed in the screen of figure 13B. Such screens may be expanded or collapsed. A user mans choose to view information related to a subset of the displayed mass flow controllers. Based on the model nunber, serial number and internally stored codes' product specifications for the mass flow controller are displayed along with user-selectable parameters, which may lye displayed in a list, for e;:ample. A user may employ this screen to download calibration clal.a to or Morn a mass flow controller and to enter calibration tables. A user may also alter set points through this interface and monitor the reported flOw through the con esponding mass flow controller-.

Additionally, a user may override setti2=s and open or close a mass flow controller's outlet control valve. Each mass Llov controller's specifications may be viewed as illustrated by the screen of Figure 13C Illustrative user-selectable parameters are displayed in the screen shot of:Figure 13D and calibration data.

sulk as a user may download fro rn a mass flow controller is i11ustra.ted in the screen shot of F;,GL;re 13E.

25, A software implenenta.tion of the above described embodiment(s) may comprise a series of computer instructions either fixed on a tangible medium, such as a computer readable media, e.g. disl:.et:te, CD-ROM, ROM, or fixed disc, or transmittable to a computer system, via a modem or other interface device, such as communication adapter connected to else net:'or1c offer a medium . Medium can be either a tangible 1nedium, including but not limited to, optical or analog eonununieations lines, or may be implemented with wireless techniques, including but not limited to mierovave, infrared or other transmission teehriques.

The series of computer instructions embodies all or part of the functionality previously described herein Title respect to the invention. Those slilled in the art will appreciate that surly computer instructions can be written in a nunber of programming languages for use faith many computer arcliteetures or operating systems Further, such instructions may be stored using any memory technology, present or future, including, but not limited to, semiconductor, magnetic, optical or other memory devices, or trailsnitted using any Communications technology, present or future, including but not limited to optical, infrared, microwave, or other transmission technologies It is contemplated that such a computer program product may be distributed as a removable media with accompanying printed or electronic documentation, e g., shrink Wrapped software, preloa.ded with a computer system, e g, on system ROM or fixed disc, or distributed fro1n a server or electronic bulletin board over a netu ork, e.g., the Internet or World Wide Web A]tlough various exemplary embocliine1ts of the invention have been clisclosed, it will be apparent to those skilled in the art that various changes and nodific;ations can be made which Will a.cllieve some of the advantages of tile invention without departing from the scope of the invention. It will he apparent to those reasonably skilled h1 the art that other components perforning the saline functions may be suitably substituted. Further, the netllods of the invention may be achieved in either all software implementations, using the appropriate object or processor instructions, or in hybrid implementations that utilize a combination of hardware logic, software logic and/or firmware to achieve the same results Processes illustrated through the use of flow charts may not be strictly linear processes and alternative flows may be irnpiemented within the scope of the invention. The specificconfiguration of logic and/or instructions utilized to achieve a particular function, as well as other modifications to the inventive concept are intended to be covered by the appended claims.

The foregoing description of specific embodinacuts of the invention has been presented for the purposes of illustration and description. It is not interceded to be exhaustive or to limit the invention to the precise forms disclosed, and many modifications and variations are possible ilk light of the above teachings. The embodiments were chosen and described to best explain the principles of the invention and its practical application, and to thereby enable others skilled in the art to best utilize the invention. it is intended that the scope of the invention be limited only by the claims appended hereto.

Claims

1. A gas flow standard comprising: a variable gas flow source; a receptacle of predetermined volume configured to receive gas from the variable gas flow source; and a pressure differentiator configured to produce an electronic signal representative of the time derivative of gas pressure within the receptacle of predetermined volume.

2. A gas flow standard comprising: a receptacle of predetermined volume configured to supply a variable gas flow; and a pressure differentiator configured to produce an electronic signal representative of the time derivative of gas pressure within the receptacle of predetermined volume.

3. The standard of claim I, wherein the differentiator includes: a pressure transducer configured to produce an electronic signal representative of the pressure within the receptacle; analog differentiator circuitry configured to produce an electronic signal that is representative of the time derivative of said electronic signal representative of the pressure within the receptacle; and an analog to digital converter (ADC) configured to convert one or more values of the analog time derivative signal to digital samples of the time derivative.

4. The standard of claim 3, further comprising: storage for storing one or more samples of the signal representative of the time derivative of gas pressure.

5. The standard of claim 1, wherein the differentiator includes: a pressure transducer configured to produce an electronic signal representative of the pressure within the receptacle; an analog to digital converter (ADC) configured to convert one or more values of the pressure transducer signal to digital form; and a digital differentiator configured to produce a plurality of digital values representative of the time derivative of the pressure signal.

6. The standard of claim 5, further comprising: storage for storing one or more samples of the signal representative of the time derivative of gas pressure.

7. The standard of claim 1, wherein the gas source is configured to provide an exponentially variable gas flow.

8. The standard of claim 7, wherein the gas source is configured to provide an exponentially increasing gas flow.

9. The standard of claim 7, wherein the gas source is configured to provide an exponentially decreasing gas flow.

10. The standard of claim 2, wherein the gas source is configured to provide an exponentially variable gas flow.

11. The standard of claim 10, wherein the gas source is configured to provide an exponentially increasing gas flow.

12. The standard of claim 11, wherein the gas source is configured to provide an exponentially decreasing gas flow.

13. A method of producing a gas flow standard signal comprising the steps of: (A) providing a variable gas flow to or from a source; (B) flowing the variable gas flow into or out of a receptacle of predetermined volume at a variable rate; and (C) producing an electronic signal that is representative of the derivative of the gas pressure within the receptacle of predetermined volume as the variable gas flows into or out of the receptacle.

14. The method of claim 13, wherein the step (C) comprises the steps of: (C1) measuring the pressure within the receptacle of predetermined volume at least three times during the variable rate flow; (C2) computing at least two pressure differences; and (C3) dividing each pressure difference by the time between the two pressure measurements used to compute each respective pressure difference.

15. The method of claim 13, wherein the step (C) of producing an electronic signal that is representative of the derivative of the gas pressure within the receptacle of predetermined volume includes the steps of: (C4) a pressure transducer producing en electronic signal representative of the pressure within the receptacle; (C5) analog differentiator circuitry producing an electronic signal that is representative of the time derivative of said electronic signal representative of the pressure within the receptacle.

16. The method of claim 15, wherein the step (C) of producing an electronic signal that is representative of the derivative of the gas pressure within the receptacle of predetermined volume includes the step of: (C6) an analog to digital converter converting one or more values of the analog pressure time derivative signal to digital samples of the pressure time derivative.

17. The method of claim 16, further comprising the step of: (D) storing one or more samples of the signal representative of the time derivative of gas pressure.

18. The method of claim 13, wherein the step (C) of producing an electronic signal that is representative of the derivative of the gas pressure within the receptacle of predetermined volume as the variable gas flows into or out of the receptacle includes the steps of: (C7) a pressure transducer producing an electronic signal representative of the pressure within the receptacle; (C8) an analog to digital converter converting one or more values of the pressure transducer signal to digital form; and (C9) a digital differentiator producing a plurality of digital values representative of the time derivative of the pressure signal.

19. The method of claim 18, further comprising the step of: (E) storing one or more samples of the signal representative of the time derivative of gas pressure.

20. The method of claim 13, wherein the gas source provides an exponentially variable gas flow.

21. The method of claim 20, wherein the gas source provides an exponentially increasing gas flow.

22. The method of claim 20, wherein the gas source provides an exponentially decreasing gas flow.