EP1325304A1

EP1325304A1 - Afterglow emission spectroscopy monitor

Info

Publication number: EP1325304A1
Application number: EP01972976A
Authority: EP
Inventors: Inc. Verity Instruments
Original assignee: Verity Instruments Inc
Current assignee: Verity Instruments Inc
Priority date: 2000-09-11
Filing date: 2001-09-10
Publication date: 2003-07-09
Also published as: AU2001292603A1; EP1325304A4; WO2002023160A1

Abstract

Afterglow spectroscopy allows observing light emission of gaseous species in absence of direct plasma light. This absence avoids the creation of a background sprectrum obscuring weak emission from trace species. The geometry of the invention prevents direct plasma light (3) from reaching the observation window (10) while a lens (10) focues afterglow emission light on the entrance of a photo multiplier (11). The invention describes a flowing afterglow version monitoring in-situ the cleanup of vacuum tools during pump/purge cycles. The invention also describes an intermittent afterflow version siutable for trace gas analysis at atmospheric pressure.

Description

Patent Application of

Jacob Mettes

for

AFTERGLOW EMISSION SPECTROSCOPY MONITOR

Background-Description of Invention

The invention relates to semiconductor manufacturing, monitoring of vacuum chamber cleanup and trace gas analysis.

Background-Description of Prior Art

The semiconductor manufacturing industry achieves higher and higher levels of integration of integrated circuits dealing with contamination and yield problems. Frequent pump/purge cycles with high purity purge gas in between reactive process gas cycles are hereto applied. These pump/purge cycles clean out process gas from a prior reactive process cycle as well as species like oxygen and moisture that might have entered the tool's chamber. In situ, real time process monitors of all kind of process variables are considered critical for current and future developments. An example of a new type of monitor is the measurement of trace concentrations of moisture in the exhaust of a semiconductor-manufacturing tool by diode laser based absorption spectroscopy. Traditionally, monitoring the pressure during pumpdown can reveal gross problems with leaks or excessive outgassing. A pressure gauge however is not specific for a particular species. Moreover, most information is in the very tail part of a drydown curve and in a normal operating procedure there is simply no time to wait that long. The purge and pump cycles are often done simultaneously, opening wide to the pump while entering a stream of purge gas with a total cycle time in the order of 10 or 20 seconds. Species specific monitoring can be done with a mass spectrometer that is then called a residual gas analyzer (RGA). Such RGA operates at 10^"3 - 10^"6 Torr requiring differential pumping when probing a purge gas at a higher pressure. Mass spectrometers have a residual spectrum at the lower masses that prevents measuring water and oxygen at levels below a part per million.

Some of the above and other in-situ measurement techniques for pump/purge cycles can detect the removal of priory introduced reactive process gas and rare intrusions of excessively high contamination levels. Even to accomplish this limited objective, in-situ techniques, such as RGA, are expensive, maintenance intensive and require elaborate calibration. Moreover, none are sensitive and fast enough to follow the actual progress of mentioned 10-20 seconds pump/purge cycles at partial pressures of relevant contaminants in the order of 10^"12 atmosphere.

Optical (none afterglow) emission spectroscopy finds practical applications in the semiconductor manufacturing industry. The observation of the spectrum emitted by the process plasma during a process and its evolution over time provides information about the completion of certain reactions such as in end point detection. The presence of a broad background spectrum however obscures any weak emission from trace species.

Non in-situ qualification of cylinder gas or bulk supply of electronic grade nitrogen and argon by metastable transfer emission spectroscopy (MTES) is investigated for oxygen-containing impurities by J. W. Mitchell et al., Analytical Chemistry 1986, 58, 371-374. The required bulky and maintenance intensive vacuum technology in such a flowing afterglow application of MTES prevents it from being applied in any practical gas analytical instrumentation.

Non in-situ gas analysis at parts per billion levels is routinely performed on line for the purge gas prior to it's entering into the process tool. At that point the purge gas is at higher than atmospheric pressure while the analysis can be performed in the order of minutes. The very sensitive and contaminant specific instruments for such purge gas analysis are used downstream of the large purifiers that provide gas to a semiconductor plant's gas distribution manifold. Such instruments are still too complex and expensive to be used as simple end of life detection for point of use purifiers that are used at the purge gas inlet of an individual tool.

Such non in-situ analysis prior to entering the tool can sometimes also be performed for the reactive process gases although, e.g., highly desirable hygrometry of corrosive gases can only be done economically per cylinder, offline, in a laboratory environment.

Even non in-situ applications requiring only ppm level detection limits can be troubled by the lack of speed, reliability, robustness and ease of maintenance of current state of the art instrumentation. Such applications include a 10 percent increased efficiency of the recovery of argon from an air separation system described in patents US 4784677 and US 544889 requiring ideally an analyzer with a 1 second response time. Patent US 4801209 describes an analyzer for the maximization of argon recovery, but lacks the robustness and reliability needed for application in a process environment. Objects and Advantages

Accordingly, besides the objects and advantages of monitors described above, several objects and advantages of the present invention are to provide a device and method for sensitive and rapid in-situ gas contaminant monitoring during the pump/purge cycle of vacuum tool equipment where such monitoring includes among others:

(a) contaminants including oxygen containing species like moisture and molecular oxygen. The device and method according to the invention enables detection at very low levels unattainable by existing prior art technology,

(b) the clean up of reactive process gas introduced prior to the present pump/purge cycle,

(c) the moisture level in reactive process material introduced prior to the present pump/purge cycle. Other objects and advantages are to provide devices and methods to perform non in-situ gas analysis in a simple, robust and economical fashion for applications such as end of life detection for gas purifiers and in air distillation applications, requiring respectively very low detection limits and very fast response times.

Further objects and advantages of the invention will become apparent from a consideration of the drawings and ensuing description.

Description of Drawings

Figure 1 : flowing afterglow monitor with RF plasma excitation on the tool exhaust.

Figure 2: basic flowing afterglow monitor with DC plasma excitation.

Figure 3: flowing afterglow monitor at the tool's purge gas inlet.

Figure 4: flowing afterglow monitor with a zero reference.

Figure 5: afterglow monitor with intermittent plasma and lock-in detection.

List of Reference Numerals

flowing afterglow inlet 39 additional shield tools mechanical fore pump's inlet 40 beam splitter plasma 41 first interference filter quartz tube 42 first photo multiplier first RF external ring electrode 43 second interference filter second RF external ring electrode 44 second photo multiplier shield for RF emissions 45 light observation area center leg tee 50 tool tee housing observation area 51 afterglow monitor's fore pump

10 window / lens combination 52 tool's turbo pump 11 photo multiplier 53 turbo pump's outlet

12 interference filter 54 tool's fore pump

13 light observation area 55 isolation valve

24 DC central electrode 56 zero gas MFC

25 electrical isolation 57 tool's purge gas MFC

26 HV connection 58 reactive process gas MFC

27 feedthrough flange 59 pressure gauge

30 purge gas inlet 70 Venetian blinds

31 connection to tool 71 square wave generator

32 first ring shaped electrode 72 square wave function

33 second ring shaped electrode 73 lock-in amplifier

35 ring shaped electrical isolator 74 electrical activation

36 ring shaped electrical isolator 75 RF generator

37 plasma 76 AM modulated RF

38 gap for purge gas and metastables 77 RF amplifier

Description of Invention

First preferred embodiment: afterglow monitor on tool exhaustFigure 1 shows the first preferred embodiment. The afterglow monitor is positioned, e.g., in the gas stream from the turbo pump exit 1 of the tool to the inlet 2 of the mechanical forepump. The monitor creates it's own plasma 3 in a quartz tube 4 equipped with two external circular RF electrodes 5 & 6. The quartz rube 4 conducts mentioned gas stream to the mechanical forepump. Also shown in figure 1 is a shield 7 around the plasma 3 preventing the device from becoming a source of RF radiation. The plasma 3 will create metastable species that will be taken by the gas flow into the central arm 8 of a tee piece 9. One of the remaining arms of the tee piece 9 connects to the inlet 2 of the mechanical forepump while the other arm features the vacuum tight afterglow observation window/lens 10. This geometry prevents direct plasma light from reaching the observation window 10 while a lens 10 focuses afterglow emission light on the entrance of a photo multiplier 11. The setup of the lens 10 is such that indirect plasma light reflected by the walls is not projected on the photo multiplier's 11 entrance. In all of this patent's figures the lens 10 and the window 10 are shown combined as one single component 10. An interference filter 12 is mounted in between the lens 10 and the photo multiplier 11 to select the desired emission line wavelength.

More arms can be added to the tee piece 9 to observe afterglow emission at additional wavelengths through additional windows and interference filters. The geometry for mounting such additional arms should avoid direct plasma 3 light from reaching the arm's observation window while a lens is used to focus light from the afterglow on the corresponding multiplier.

The inside of aluminum vacuum pieces such as the tee 9 can be black anodized to absorb direct plasma light thus reducing the light reaching the observation window through reflections. Figure 1 also shows the use of an additional elbow vacuum piece in between the plasma 3 and the afterglow observation area 13 to provide an additional barrier for direct plasma light to reach the observation window 10.

The quartz tube 4 in between plasma 3 and the electrodes 5 & 6 avoids potential contamination from contact of the metal electrodes with the plasma.

Optionally, the first preferred embodiment can, instead of using it's own plasma 3 source, use the plasma of the process tool itself. This is achieved by igniting (after an impedance match) the tool's plasma in a purge gas stream of inert gas like nitrogen or argon.

Second preferred embodiment: afterglow monitor with DC plasma The second preferred embodiment, shown in figure 2, differs from the first preferred embodiment as far as its plasma source is concerned. The quartz tube 7 and the coils 5 & 6 of the RF plasma in figure 1 are replaced by center electrode 24 that is connected outside the tool's chamber to a high voltage DC power supply through an electrical feedthrough in flange 27. The connection 26 of the feedthrough is electrically isolated from flange 27 by isolation 25. The walls of the electrically grounded metal chamber surrounding the center electrode 24 form the second electrode. The plasma 3 is now created in between the first, center, electrode 24 and the second electrode being the chamber's wall. The simpler second preferred embodiment is subject to contamination such as evaporating traces of metal generated by the contact of the metal electrodes with the plasma. However, it is unlikely that such traces of metal contamination will emit light downstream of the gas sfream as the result of a collision with metastable species. Moreover, the wavelength and width of the interference filter 12 can be used to block such unwanted light emission.

Third preferred embodiment: afterglow monitor at the tool's purge gas inlet The third preferred embodiment is shown in figure 3. The afterglow monitor is positioned in between the purge gas stream 30 downstream of, e.g., a mass flow controller and flange 31 connecting to the tool's vacuum chamber. The plasma 37 is created in the space between two similarly sized coaxial ring shaped electrodes, 32 & 33, that face each other. Each electrode 32 & 33 is mounted by means of a coaxial ring shaped insulator respectively 35 & 36. The gas stream first passes through the space with the plasma 37 where after it flows through a ring shaped slot 38 to flow into the tool's vacuum chamber. The afterglow emission is observed through a vacuum tight window 10 that is shielded from direct plasma light by the geometry of shield 39 around the slot 38. Shield 39 prevents direct plasma light to reach the observation window 10 while providing limited wall contact to minimize loss of metastables by wall collisions. A beam splitter 40 enables to observe different wavelengths of afterglow light with multiple interference filter & photo multiplier combinations 41, 42 and 43, 44 using one and the same observation lens/window combination 10.

The third preferred embodiment allows to observe, besides impurities already present in the purge gas itself, species from inside the tool's process vacuum chamber that back diffuse into the stream of metastables that exits the monitor. Such backdiffused species will react with these metastables and resulting light emission can be monitored by the afterglow monitor. These backdiffusing species are the indicators of the quality of the ongoing purge.

Apart from purge gas, also part of the process gas constituents, like argon, can be made to enter the tool passing through the afterglow monitor during reactive process cycles. This will protect the monitor's observation window 10 from process tool plasma deposits that will now have to diffuse against a "counterflow".

Optionally not shown in the figure 3, a mechanical shutter, e.g. of the Venetian blind type, can be positioned in between the afterglow monitor's process tool connection flange and the mating process tool's connecting flange. This blind can, on the one hand, protect the observation window during active, non-purge, process cycles. On the other hand, during purge cycle monitoring, shutting the blinds can drastically reduce the influx of species backdiffusing from inside the process tool's chamber into the afterglow monitor's observation zone. A Venetian blind kind of shutter consists of an array of parallel metal strips that can rotate around their longitudinal axis and is shown as component 70 in another context in figure 5. The blind does not shut in a vacuum or leak tight manner at all, but when shut it will restrict the gas stream somewhat and force it locally into flowing at an increased speed. This speed increase through narrower openings will effectively reduce backdiffusion. This way, shutting the blinds will show the emission spectrum from the purge gas in absence of impurities added by the tool. Such emission spectrum can serve as a reference or zero as it corresponds to the ultimately achievable end result of purging a perfectly cleaned tool, similar to that discussed hereunder in the fourth preferred embodiment. Eventually, an additional L piece can be added in between the Venetian blind and the afterglow monitor's connection flange 31 to assure equal observation areas for opened and closed blinds.

Fourth preferred embodiment: afterglow monitor with zero reference The fourth preferred embodiment is shown in figure 4. This embodiment is basically that of the first preferred embodiment with added provisions to monitor purge gas that bypassed the tool 50, thus providing a zero reference. The zero reference's afterglow emission corresponds with the asymptotically approached ultimate end result of a perfectly purged tool 50. Contributions on top of the zero reference are tool 50 contributions, which can be separated by subtracting the zero reference from a regular tool 50 measurement. Such subtraction can be done most easily for measurements done in the linear part of the calibration curve. The afterglow monitor in preferred embodiments 1 & 2 uses the mechanical fore pump from the tool itself. This fourth preferred embodiment has an afterglow monitor with it's own mechanical fore pump 51. A stream of purge gas, exiting the tool 50, is teed off from the gas stream between the tool's turbo pump 52 exit 53 and the tool's mechanical fore pump 54. An isolation valve 55 can block this teed off stream which is then replaced by a gas stream of purge gas entered directly through a Mass Flow Controller, MFC, 56 in between the afterglow monitor's inlet and the isolation valve 55. Purge gas to purge the tool enters through MFC 57 while process gas enters through MFC 58. When the isolation valve 55 is closed, e.g. during active non-purge process cycles, mentioned zero reference can be obtained while a feedback loop tunes MFC 56 to reproduce the same pressure gauge 59 reading that is present during regular purge gas monitoring conditions with the isolation valve 55 opened. In fact, with an opened isolation valve 55 during a tool 50 purge, MFC 56 can be tuned over a range of flowrates while MFC 57 is adapted to maintain a fixed pressure at the indicated pressure gauge 59. This range for MFC 56's flowrate can vary from, on one extreme, a minimum to prevent a stagnant flow at the MFC 56's outlet connection. This corresponds to monitoring undiluted purge gas exiting the tool 50. On the other extreme, MFC 56's flowrate can be made large enough to provide not only a flow through the afterglow monitor into the afterglow's monitor's mechanical forepump 51 but also an excess flow streaming away from the afterglow monitor mixing in with the purge gas exiting the tool 50 toward the mechanical forepump 54 of the tool 50. This situation would correspond to monitoring undiluted purge gas without contributions from the tool 50 that would have to diffuse against the excess purge gas flow. In between these extremes, mixtures of pure purge gas and purge gas with tool contributions can be generated where the ratios can be determined from the MFC 56 flowrate and a measurement of the flowrate at the exit of the afterglow monitor's mechanical forepump 51. Such setup enables to dilute very contaminated purge gas exiting the tool 50 thus assuring measurement on the linear part of the calibration curves. Optionally this last described setup could function also without isolation valve 55 or with a Venetian blind shutter instead of an isolation valve 55 to avoid potential contamination during active process cycles.

The isolation valve 55 setup however enables a single afterglow monitor with a single corresponding MFC 56 to be connected to several tools of, e.g. a cluster tool, serving one at the time, where each cluster tool has its own isolation valve 55.

Fifth preferred embodiment: intermittent afterglow monitor Figure 5 shows the intermittent afterglow setup combined with a phase sensitive, lock in, detection scheme. The mechanical shutter is shown as a Venetian blind 70 with activator 74 but can be a perforated rotating wheel, a butterfly structure etc. The phase sensitive detection allows in principle to measure signals smaller than the background signal fluctuations that otherwise set the limit to the achievable detection limit. Hereto, however, the shutter 70 created optical isolation from the plasma must be perfect, leaving only dark counts as the signal from the photo multiplier 42 (and 44).

A low frequency square wave generator 71 's signal 72 is offered to the lock-in 73 as reference signal while it is also offered to the actuator 74 of the mechanical shutter 70 and to the amplitude modulation input of a RF signal generator 75. In case of, e.g. a mechanical rotating chopper wheel, such low frequency square wave signal can be taken directly from the chopper and offered to the lock-in's reference input and the RF generator 75's AM input.

The RF generator 75 will produce periods of RF signal 76 in phase with the periods of a closed shutter (chopper). The RF generator 75's output will be amplified by amplifier 77 and offered to the electrodes 5 & 6 of the plasma. Periods of open chopper 70 will result in afterglow signal from the photo multipliers 42.& 44. Closed chopper photo multiplier signal will correspond, as discussed, with the dark count of the multiplier. The lock-in 73 will produce an output signal proportional to the average signal contribution from the afterglow emission during a chopper 70 period.

Alternative methods of data treatment consist of curve fitting a down-going exponential to the signal obtained as a function of time during the individual open chopper periods. Other options would be to ignore the closed shutter signals and produce a signal proportional to the average emission received during the open shutter periods using gated electronics. If the shutter adequately isolates plasma light, one could also output the averaged received light from the photo multiplier(s).

It necessary, any of the described afterglow monitor embodiments can be isolated from vibrating components such as the mechanical forepump by, e.g., bellows in order to protect more delicate parts such as a photomultiplier.

Operation of Invention

Background spectrum Optical plasma emission spectroscopy is typically not suited to measuring trace concentrations of species. Even high-resolution spectroscopy is of limited use in case of trace concentrations because of the presence of a broad background spectrum. The invention is based on the avoidance of this background spectrum made possible by the realization that it is mainly caused by fluorescence of impurities in the (quartz) observation window/lens 10 irradiated by strong main (UV) lines of the plasma. Apart from this fluorescence, the presence of strong plasma lines and possibly other plasma emission make it also difficult to detect weak signals from trace concentrations in the plasma.

Shielding of the window The invention shields the observation window/lens 10 from direct plasma emission by creating a "dark" area in front of it. This is done, on the one hand, by geometrical shielding with one or more screens blocking a straight path from the plasma to the window 10. Additional blackening of the screens and parts of the chamber can be used to reduce plasma (UV) light reaching the window 10 after one or more reflections. Additionally, dark zones created in the plasma, e.g., near the electrodes in a DC plasma or with grounded plates, can also be used to limit direct emission from irradiating the window. These forms of shielding are applied for a plasma that is continuously ignited during, e.g., purge cycles of a semiconductor manufacturing process, in contrast to the hereafter discussed intermittent plasma.

On the other hand, the observation window or lens 10 can be shielded from plasma light by means of a mechanical shutter 70 positioned in between the window 10 and the plasma. Extinguishing the plasma and opening the shutter 70 allows the observation of the afterglow emission decreasing over time. The shutter 70 can be closed again and the plasma reignited whereupon the cycle is repeated. Such intermittent cycles can be repeated many times, e.g., during one semiconductor manufacturing purge cycle. This form of shielding is applied in the invention's intermittent plasma afterglow monitoring.

Afterglow Long lived high energetic species, such as metastables, created priory in the plasma end up in front of the observation window 10 either by travel in case of the flowing afterglow or by the removal of the chopper barrier in case of the intermittent afterglow. In the flowing afterglow mentioned travel into the invention's shielded " dark " area, can take place either by moving with a stream of gas, by diffusing into a stagnant flow or even by moving against a stream of gas (back diffusion). The mean free path without collision of a species in a low-pressure environment can become relatively long while, e.g., an excited argon atom cannot lose its energy efficiently colliding with another argon.

The afterglow spectrum observed in isolation of the main plasma has a much less complex emission spectrum than the main plasma itself. The dominant mechanism of excitation of atom or molecular species by high energetic electrons is replaced by recombination and energy transfer by other atoms, ions or molecules. This results in a much less dense matrix gas spectrum where trace impurity emission appears more clearly in absence of mentioned broad background spectrum.

Interpretation of the spectrum It is typically difficult to extract quantitative information regarding the type and concentrations of contaminating species present prior to their entrance into the plasma from the information regarding the resulting species observed in the plasma. The highly energetic plasma environment decomposes molecular species that recombine into new species. As an illustrative example a plasma of argon with 2 ppm H₂ and 1 ppm 0₂ will have similar OH emission as a plasma with 2 ppm H₂0.

Another difficulty is that quantitative emission spectroscopy not only involves the concenfration of emitting species but also the intensity of the source of excitation. The invention realizes that a plasma created in ultra high purity purge gases, such as argon or nitrogen used in semiconductor manufacturing tools, addresses both difficulties. On the one hand, such gases decompose in a very limited number of fragments while there is only a handful of impurity types. On the other hand, such plasma provides a stable source of excitation little impacted by the presence of the very low trace concentrations of impurities. As a result, calibration curves will be linear at low concentrations.

The invention can use chemometrics to address remaining non-specificity, basically solving n unknown trace concentrations from n, or more, quantified emission line intensities. The invention can use additional distinguishing characteristics that are presented hereunder.

Additional distinguishing characteristics Moisture and oxygen are main contaminants in most semiconductor manufacturing processes and typically have a negative impact on the process yield. Removal of these two impurities is also the main goal of the pump/purge cycles performed in between active process cycles. In contrast to a molecule like oxygen, a moisture molecule is very "sticky" and has a tendency to reside on the walls of a process chamber. This characteristic makes that during an evacuation, or purge, oxygen is removed relatively fast while moisture is removed much slower.

The way impurities like moisture and oxygen are introduced into the process chamber also results in different behavior. Oxygen is typically introduced as a leak from the atmosphere into the chamber. Moisture typically outgases from chamber components like O-rings or from materials from, e.g. wafer trays, brought into the chamber. This makes that besides a difference in pumpdown speed, there is also a difference in asymptotic value approached during prolonged purging. An oxygen leak will create a non-zero plateau, while moisture will eventually reach zero. Presented additional distinguishing characteristics allow mathematical treatment of data obtained as a function of time. Such mathematical treatment can be, but is not limited to, checking if higher derivatives approach zero or, e.g., curve fitting with one or more downgoing exponentials etc.

Indirect detection of wet process gas Hygrometry in corrosive gasses is typically problematic because it exposes delicate sensor parts to corrosion. The invention allows monitoring the wetness of the process gas used in the cycle prior to a purge cycle. Moisture, as a very sticky molecule, entered into the process chamber by, e.g., wet HC1, will stick long enough to the walls of the chamber to be noticeable by afterglow monitoring in the following purge cycle. Means to distinguish such moisture from that of other sources can be: enter process gas without ignition of the process plasma, vary the introduction time or flowrate of the process gas and differences with and without a stack of wafers present in the tool chamber.

Pump/purge efficiency monitoring As mentioned, the purpose of the pump/purge cycles in a semiconductor manufacturing process is typically to remove such trace impurities as moisture and oxygen. In absence of real time data regarding the efficiency of an ongoing pump/purge cycle, a tool operator has to rely on prior experimentally determined purge times that proved adequate under most "normal" operating conditions. These purge times will be applied plus an additional safety margin. In situ metrology will not only cut excess purge time that will become available as process time. It will also catch contamination problems as soon as they occur, avoiding time and material costs related to the manufacture of unusable product runs. In situ monitoring can also cut startup time after maintenance and save costs running test wafers.

Even when used only as a qualitative monitor, the invention allows following the evaluation of characteristic emission line intensities during the pump/purge cycle, which provides a measure for the reproducibility of this part of the process. It can also give an indication that further purging is worthwhile or not.

Calibration Calibration of the monitor can be done by the introduction of a purge gas with known concentrations of, e.g., oxygen and moisture. A stable flowrate of such gas can be created at various pressures in the range of interest.

Using the linearity of the monitor's responds at low concenfrations and the absence of a background spectrum, one can also add calibrated concentrations to an unknown purge gas. Extrapolation allows determining the unknown purge gas concentration.

Using the large volume of the tool chamber upstream of the afterglow monitor, exponential dilution can be used to generate a large range of calibrated concentrations as a function of time. Hereto, a known concentration of, e.g. oxygen in high purity purge nitrogen, is flowing until stable afterglow emission is reached in an equilibrium situation. Then a switch is made introducing only a flow of high purity purge gas into the tool whereupon the monitored concentration at the outlet of the tool will follow an exponentially down going function of time. This function can be calculated from the volume of the chamber, the initial concentration and the size of the flowrate of the purge gas. Window depositions The transparency of an observation window in a vacuum chamber used to perform active process steps like chemical vapor deposition, might degrade over time. Solutions preventing window deposition range from the creation of a counter gas flow away from the window, to window shutters and "Venetian blinds", to an isolation valve to protect the window during active process cycles.

Compensation for window depositions can be done in the hereunder further described setups that allow creating a zero or reference emission spectrum. This zero or reference specfrum can be used to "normalize" other measurements. Results can be based one the ratio of line intensities and the zero spectrum line intensities thus compensating for changes in the transparency of the observation window.

Normalization As mentioned, quantitative emission spectroscopy not only involves the concentration of the emitting species but also the intensity of the source of the excitation. In case there are emission lines available that are a measure of the intensity of the source of excitation, such lines can be used to "normalize" a quantitative measurement. This allows correcting for fluctuations in the intensity of the source of excitation. An example according to the invention is the use of afterglow emission caused by transitions from N₂ (C³II_U) to N₂(B³IT_g) in a nitrogen plasma. Two N₂(A³Σ) metastable nitrogen molecules can react with each other leaving one of the molecules in the (C³π_u) state. This makes the concentration of metastable N₂(A³Σ) proportional to the square root of the intensity of mentioned afterglow emission (N₂ (C³IT_U) to N₂(B³π_g)).

Metastable production Plasma conditions such as DC voltage & current or AC frequency, voltage & current can be tuned to generate high concenfrations of metastable species in the plasma. Metastables can loose their energy either by wall collisions or by collisions with frace impurities. The rate of losses of metastables through wall collisions depends on the diameter of the tubing, the flowrate of the gas, pressure, diffusion coefficient, concenfration of metastables generated in the plasma and the distance downstream of the plasma. Energy loss through collision with frace impurities depends on the concentration of the impurities and the reaction rate. The metastable concentration will be stable at low impurity concentrations resulting in a linear calibration curve. Such curve will eventually flatten at high impurity concentrations that deplete the metastable concenfration making less metastables available to generate afterglow emission.

Emission spectra The afterglow emission is typically generated by highly excited diatomic species which can be molecules or radicals mostly created from recombining fragments of species decomposed in the plasma. The emission specfrum associated with a particular transition between specific molecular states appears in the form of bands containing peaks each associated with a specific vibrational initial and final state. For the purposes of the invention, a particular wavelength has to be selected, e.g. when use is made of an interference filter. In principle, any major peak, or group of peaks, from a band can be chosen for the monitoring as long as it is not in the vicinity of possible other afterglow emission wavelengths from another species.

Argon purge gas The metastable state of Argon ³P₂ is typically the main source of excitation of the impurity molecules and in a much lesser extend metastable Argon ³P₀. Emission that can be seen in an Argon afterglow originate from species like, e.g., N₂ (C³π_u)-(B π_g) and radicals like OH (A²Σ)-

(X²π) and CH (A²Δ)-(X²π).

As discussed above, the main concern when purging a vacuum tool is the removal of moisture and entered atmospheric components like oxygen. Using ultra high purity Argon as purge gas the afterglow's OH, N₂, and eventually NO emission enable to monitor for moisture and air leaks. Leaks from atmospheric air will bring in a fixed well known ratio of nitrogen and oxygen, making de nitrogen emission a good tracer for the presence of the corresponding oxygen. The NO emission can provide additional information.

Nitrogen purge gas The N₂(A³Σ) metastable is the main source of excitation of NO formed in the nitrogen plasma out of oxygen containing species such as moisture and 0₂. NO emission is caused by transitions from NO(A²Σ⁺) to NO(X²II). Other emission lines in a nitrogen purge gas can be OH (A²Σ)-(X²π) and CN (B²Σ) - (X²Σ). See also Normalization.

Bypass the tool A valve arrangement could send a purge gas stream directly into the afterglow detector bypassing the tool. This allows comparing the difference in emission from the afterglow with and without the contribution from the tool. Such arrangement can be combined with the use of an isolation valve to protect the afterglow monitor during the actual process cycle. Emission spectra from purge gas bypassing the tool, hereinafter called a zero or reference spectrum, can be obtained during the time periods that an active (non purge) cycle takes place in the tool. In some arrangement, see preferred embodiments, the afterglow plasma can stay on continuously.

Performance Linear calibration curves were obtained up to the parts per million range, with a darkcount rate of the photomultiplier corresponding to one part per billion (moisture in argon). Interference filters instead of the monochromator used to obtain above-mentioned sensitivity allow to catch a much larger fraction of the emitted light, not only in terms of space angle but also in terms of a wider frequency band. This and an optimized design should easily enable 0.1 ppb detection limits with responds times in the order of seconds. In a 2 mBar purge gas such detection limit represents a partial pressure of 2 x 10^"13 Bar.

Afterglow monitoring of the removal of reactive species introduced prior to the purge/pump cycle

The first option to do this is simply monitoring afterglow emission associated with remaining reactive species as a function of time during the purge cycle. Option two is to use besides the emission of the first option also emission related to the concentration of purge gas metastable species such as caused by transitions from N₂(C³II_U) to N₂(B³π_g) in a nitrogen plasma. Similarly as described above, see "Normalization", this last emission can now also be used to normalize the reactive species related emission. The third option is to monitoring just the mentioned emission related to the concentration of purge gas metastable species such as caused by transitions from N₂(C³I1_U) to N₂(B³π_g) in a nitrogen plasma. Apart from losses through wall collisions, the metastable species will loose their energy through quenching. Quenching is caused by metastables colliding with other contamination molecules and with molecules from the remainder of reactive species introduced prior to the purge/pump cycle. The better the reactive species are cleaned out by the purge the stronger the emission from, e.g., a N₂(C³IT_U) to

N₂(B³π_g) transition. This third option can result in a very simple device where metastables diffuse out of a shielded plasma area into the observation of a photo sensor with an interference filter. The fourth option monitors the same emission as the previous third option but uses the additional information about the strength of this emission at "zero" gas conditions as a reference. Zero gas conditions can be obtained either from the end state of a previous purge/pump cycle or by above described purge gas bypassing the tool.

A fifth option is to operate the flowing afterglow setup in a pulsed fashion. Such option would require a detection optics that gathers emission over a certain length of the path of the afterglow. Such optics can, e.g., consist of a photosensor in the focal point of a lens with the observation direction along the flow path of the afterglow. The "flowing" of the afterglow can, as mentioned above, in it's simplest form be caused by diffusion of metastables away from a shielded active plasma zone. Monitoring the dying out in time of any afterglow emission following a plasma pulse will reveal information about the presence of quenching species such as the remainder of priorly introduces reactive process gas. A sixth option combines the fifth option with information of dying out characteristics of zero gas species following a plasma pulse. This zero gas info can again be used as a reference corresponding to an ideally cleaned tool.

Intermittent afterglow arrangement An alternative to the hitherto discussed continuous afterglow arrangements is the intermittent afterglow arrangement. In the intermittent afterglow arrangement the plasma is alternately ignited and extinguished. During plasma ignition, emission is blocked from reaching the observation window by a chopper provision placed in between the window and the plasma. When the plasma is extinguished the chopper provision clears access to the window. This way, afterglow emission can be observed in absence direct plasma light and in absence of fluorescence of impurities in the observation window.

The plasma is created in a volume in front of the chopper provision and might be sustained during most of the blocking period of the chopper. Losses of metastables by wall collision are limited while phase sensitive lock-in detection can be applied allowing to measure afterglow emission smaller than the darkcurrent of the photomultiplier. For in-situ monitoring in a vacuum tool however, the mechanical shutter operates under relative difficult conditions being placed in a vacuum and exposed to a plasma.

However, the intermittent afterglow using, e.g., a fast enough rotating perforated disk chopper can observe the very short lived afterglow associated with operation at higher gas pressures such as atmospheric pressure. Such device can perform non in-situ gas analysis on gases like argon, nitrogen and hydrogen at pressures around atmospheric pressure making it unnecessary to use any vacuum equipment such as a vacuum tight window, pumps etc. The plasma in this case should be very short lived and can just be created by, e.g., an arc discharge much like the ignition in a spark plug. The position of the rotating disk chopper will trigger the discharge much in similar way as a car's ignition system is triggered by the proper position of the pistons. The chopper disc and surrounding chamber can be made of metal to act as a Faraday cage blocking potentially perturbing electromagnetic radiation when igniting the discharge.

Application involving long lived excited state species So far, discussed afterglow emission resulted mainly from energy transfer from non emitting metastable species to contamination molecules. The invention can also utilize emission from long lived species created in the plasma. An application according to the invention is the detection of traces of nitrogen in a matrix of pure oxygen or a matrix of a mixture of oxygen and, e.g., argon. Nitrogen and oxygen present in a plasma discharge will lead to the formation of species like excited state nitrogen dioxide. Monitoring characteristic nitrogen dioxide emission in the afterglow will be a measure for the presence of frace nitrogen.

Similarly, traces of oxygen in a nitrogen matrix can be detected. The above described high pressure intermittent afterglow according to the invention is well suited for applications that can be found in air distillation plants described, e.g., in patent US 4784677 and US 544889.

Conclusion, Ramifications, and Scope of Invention

Thus the reader will see that the afterglow emission spectroscopy in situ monitor can rapidly determine the status of an ongoing pump/purge cycle in vacuum equipment. It can determine the level of cleanup of priory entered reactive gas, and indicate the wetness of such gas when it was entered. It can not only report unusual high levels of contaminants in the purge gas but is sensitive enough to monitor the presence of contaminants such as oxygen and moisture at trace levels. This last capability provides the data to a tool operator that the pump/purge cycle is actually removing contaminants at levels that matter, or is taking painfully long to lower the last few insignificant ppb. Distinguishing the last could make purge time safety margins available as process time for the tool thus increasing the overall efficiency. The afterglow emission spectroscopy in situ monitor is simple and compatible with the materials used in high vacuum and plasma environment while it is easy to calibrate and maintain. While my above description contains many specificities, these should not be construed as limitations on the scope of the invention, but rather as an exemplification of one preferred embodiment thereof. Many other variations in, e.g., operating conditions and applications are possible. For example, an

Claims

alternative to a spark generated plasma in the atmospheric pressure intermittent afterglow can be a plasma generated by a short high energetic laser pulse concenfrated in the focal point of a lens. When the laser beam and the afterglow observation axis are perpendicular to each other there will be no observation of eventual fluorescence induced in the laser lens.

Accordingly, the scope of the invention should be determined not by the embodiment(s) illustrated, but by the appended claims and their legal equivalents.

Claims

Claim 1 A MONITOR device comprising PLASMA GENERATING MEANS where the

MONITOR device uses AFTERGLOW EMISSION to deteπnine IMPURITIES in PURGE GAS used to purge a TOOL CHAMBER where the AFTERGLOW EMISSION is observed by OBSERVATION MEANS shielded from DIRECT PLASMA EMISSION by SHIELDING MEANS.

Claim 2 A device as in claim 1 where the TOOL CHAMBER is a vacuum plasma process chamber or a load lock such as used in semiconductor manufacturing OR where the PURGE GAS enters the TOOL CHAMBER as an ultra high purity inert gas such as argon or nitrogen OR where the IMPURITIES are molecular species that are potentially damaging for the process performed in the TOOL CHAMBER like moisture, oxygen or hydrocarbons that are not intentional PURGE GAS constituents or where the IMPURITIES are remaining reactive species introduced prior to the PURGE GAS's introduction OR where the AFTERGLOW EMISSION consists of light emitted by molecules such as OH, NO, CH, CN and N2 from a high energetic excited state which formation involves long lived metastable species formed in a plasma OR where the device uses PLASMA GENERATING MEANS from the TOOL CHAMBER'S process OR where the device uses PLASMA GENERATING MEANS creating a plasma by, e.g., DC, RF or microwave discharge, independent from the TOOL CHAMBER'S process OR where the OBSERVATION MEANS consist of an hermetically sealed optical window, means to select desired optical emission wavelengths such as a monochromator, interference filters, Fourier transformed spectrometer or an optoacoustic filter and means to detect the selected wavelengths such as a photo multiplier, optical sensors such as a photo diode, transistor or other light sensitive device OR where DIRECT PLASMA EMISSION is light emitted by atom or molecular species mainly excited by high energetic electrons in the plasma..

Claim 3 A device as in claim 1 or 2 where the PLASMA GENERATING MEANS generate a continuous plasma during most or part of the purge procedure of the TOOL CHAMBER while the SHIELDING MEANS are based on the separation in space of the afterglow observation zone and the actual plasma OR where the PLASMA GENERATING MEANS are intermittently generating a plasma during most or part of the purge procedure of the TOOL CHAMBER where the afterglow observation zone and the actual plasma are identical while the SHIELDING MEANS are present in between the plasma and the AFTERGLOW OBSERVATION MEANS when the plasma is ignited and the SHIELDING MEANS are removed when the plasma is extinguished.

Claim 4 A device as in claim 1 , 2 or 3 where data regarding a zero or reference afterglow emission spectrum can be obtained exposing the MONITOR device to PURGE GAS that has not been into contact with the TOOL CHAMBER while such zero or reference afterglow emission spectrum can be used to compensate regular measurements for changes in the OBSERVATION MEANS's response function OR where additional properties, such as the sticking coefficient in wall collisions, of the IMPURITIES are taken into account to distinguish different species of IMPURITIES based on differences in temporal behavior of the observed afterglow emission OR where monitoring during a purge cycle provides information of the wetness of reactive or corrosive process gases entered during the process cycle prior to the monitored purge cycle using the additional property of the moisture molecule to stick to walls and to be released relatively slow OR where purge gas exiting the tool is diluted with purge gas that bypassed the tool prior to entering the afterglow monitor enabling to reduce the concentrations of impurities to levels where the afterglow monitor shows a linear response OR where a single MONITOR device can tee off gas from the exhaust streams of multiple tools such as tools arranged in a clustertool.

Claim 5 A device as claim 1, 2, 3 or 4 where the energy to create the AFTERGLOW

EMISION is transferred by METASTABLE SPECIES and where the intensity of the AFTERGLOW EMISION can be normalized for changes in the concenfration of the METASTABLE SPECIES by a separate measurement of a part of the AFTERGLOW EMISSION that is related to the concentration of the METASTABLE SPECIES.

Claim 6 A device as claim 1, 2, 3, 4 or 5 where the quenching of the part of the

AFTERGLOW EMISSION that is related to the concentration of the METASTABLE SPECIES is used as a measure of the presence of IMPURITIES or remaining reactive species introduced prior to the PURGE GAS's introduction AND where optionally such quenching measurement is made relative to a quenching measurement in PURGE GAS that has not been into contact with the TOOL CHAMBER.

Claim 7 A device as claim 1, 2, 3, 4, 5 or 6 where the PURGE GAS used to purge a TOOL

CHAMBER is straight ANALYTE GAS not used to purge a TOOL CHAMBER and where the device is used to determine IMPURITIES present in such ANALYTE GAS.

Claim 8 A device as claim 7 where the ANALYTE GAS is the outlet gas of a purifier and the device is used as an end of life detector for the purifier.

Claim 9 A device as claim 7 where the AFTERGLOW EMISSION comes from excited state nitrogen dioxide created in the plasma and EITHER the ANALYTE GAS is a mixture containing oxygen in which case the IMPURITIES are nitrogen OR the ANALYTE GAS is a mixture containing nitrogen in which case the IMPURITIES are oxygen.

Claim 10 A MONITOR method using PLASMA GENERATING MEANS where the

MONITOR method uses AFTERGLOW EMISSION to determine IMPURITIES in PURGE GAS used to purge a TOOL CHAMBER where the AFTERGLOW EMISSION is observed by OBSERVATION MEANS shielded from DIRECT PLASMA EMISSION by SHIELDING MEANS.

Claim 11 A method as in claim 10 where the TOOL CHAMBER is a vacuum plasma process chamber or a load lock such as used in semiconductor manufacturing OR where the PURGE GAS enters the TOOL CHAMBER as an ultra high purity inert gas such as argon or nitrogen OR where the IMPURITIES are molecular species that are potentially damaging for the process performed in the TOOL CHAMBER like moisture, oxygen or hydrocarbons that are not intentional PURGE GAS constituents OR where the AFTERGLOW EMISSION consists of light emitted by molecules such as OH, NO, CH, CN and N2 from a high energetic excited state which formation involves long lived metastable species formed in a plasma OR where the method uses PLASMA GENERATING MEANS from the TOOL CHAMBER'S process OR where the method uses PLASMA GENERATING MEANS creating a plasma by, e.g., DC, RF or microwave discharge, independent from the TOOL CHAMBER'S process OR where the OBSERVATION MEANS consist of an hermetically sealed optical window, means to select desired optical emission wavelengths such as a monochromator, interference filters, Fourier transformed spectrometer or an optoacoustic filter and means to detect the selected wavelengths such as a photo multiplier, optical sensors such as a photo diode, fransistor or other light sensitive device OR where DIRECT PLASMA EMISSION is light emitted by atom or molecular species mainly excited by high energetic electrons in the plasma.

Claim 12 A method as in claim 10 or 11 where the PLASMA GENERATING

MEANS generate a continuous plasma during most or part of the purge procedure of the TOOL CHAMBER while the SHIELDING MEANS are based on the separation in space of the afterglow observation zone and the actual plasma OR where the PLASMA GENERATING MEANS are intermittently generating a plasma during most or part of the purge procedure of the TOOL CHAMBER where the afterglow observation zone and the actual plasma are identical while the SHIELDING MEANS are present in between the plasma and the AFTERGLOW OBSERVATION MEANS when the plasma is ignited and the SHIELDING MEANS are removed when the plasma is extinguished.

Claim 13 A method as in claim 10, 11 or 12 where data regarding a zero or reference afterglow emission spectrum can be obtained exposing the MONITOR method to PURGE GAS that has not been into contact with the TOOL CHAMBER while such zero or reference afterglow emission spectrum can be used to compensate regular measurements for changes in the OBSERVATION MEANS's response function OR where additional properties, such as the sticking coefficient in wall collisions, of the IMPURITIES are taken into account to distinguish different species of IMPURITIES based on differences in temporal behavior of the observed afterglow emission OR where monitoring during a purge cycle provides information of the wetness of reactive or corrosive process gases entered during the process cycle prior to the monitored purge cycle using the additional property of the moisture molecule to stick to walls and to be released relatively slow OR where purge gas exiting the tool is diluted with purge gas that bypassed the tool prior to entering the afterglow monitor enabling to reduce the concentrations of impurities to levels where the afterglow monitor shows a linear response OR where a single MONITOR method can tee off gas from the exhaust streams of multiple tools such as tools arranged in a clustertool.

Claim 14 A method as claim 10, 11, 12 or 13 where the energy to create the

AFTERGLOW EMISION is transferred by METASTABLE SPECIES and where the intensity of the AFTERGLOW EMISION can be normalized for changes in the concentration of the METASTABLE SPECIES by a separate measurement of a part of the AFTERGLOW EMISSION that is related to the concentration of the METASTABLE SPECIES.

Claim 15 A method as claim 10, 11, 12, 13 or 14 where the quenching of the part of the AFTERGLOW EMISSION that is related to the concentration of the METASTABLE SPECIES is used as a measure of the presence of IMPURITIES or remaining reactive species introduced prior to the PURGE GAS's introduction AND where optionally such quenching measurement is made relative to a quenching measurement in PURGE GAS that has not been into contact with the TOOL CHAMBER.

Claim 16 A method as claim 10, 1 1, 12, 13, 14 or 15 where the PURGE GAS used to purge a TOOL CHAMBER is straight ANALYTE GAS not used to purge a TOOL CHAMBER and where the method is used to determine IMPURITIES present in such ANALYTE GAS.

Claim 17 A method as claim 16 where the ANALYTE GAS is the outlet gas of a purifier and the method is used as an end of life detector for the purifier.

Claim 18 A device as claim 16 where the AFTERGLOW EMISSION comes from excited state nitrogen dioxide created in the plasma and EITHER the ANALYTE GAS is a mixture containing oxygen in which case the IMPURITIES are nitrogen OR the ANALYTE GAS is a mixture containing nitrogen in which case the IMPURITIES are oxygen.