Apparatus and method for in-situ cleaning of borosilicate (BSG) and borophosphosilicate (BPSG) films from CVD chambers

Abstract

A method for cleaning borosilicate (BSG) and borophosphosilicate (BPSG) films from CDV chambers including controlling the pressure within the chamber, introducing Ar into the chamber, introducing NF3 into the chamber, adjustably spacing a heater relative to the chamber, and adjusting the temperature within the chamber.

Description

FIELD OF THE INVENTION

[0001] This invention relates to the in-situ cleaning of borosilicate (BSG) and borophosphosilicate (BPSG) films from the interior surfaces of chemical vapor deposition (CVD) chambers.

BACKGROUND OF THE INVENTION

[0002] A problem with depositing BSG and BPSG films upon semiconductors is that the inside of the CVD chamber are also coated with the film being deposited. This coating should be cleaned off periodically to avoid contamination and defects which can arise due to the deposited film “flaking off” from the inner walls of the reactor.

[0003] The semiconductor industry uses large quantities of expensive, fluorine-containing gases, for example, CnF2n+2 and nitrogen triflouride (NF3) to clean CVD chambers. By-products from the cleaning reactions include F2, CnF2n+2, HF, and other species that are poisonous and difficult to scrub from process waste streams. For example, a typical cleaning procedure using NF3 would remove a BSG or BPSG glass film from the inside of a CVD chamber according to the generic formula:

(2+w)F+SiOxByPz=SiFw+HF+F+volatiles

[0004] where F is free fluorine that condenses to molecular F2 upon exhaust.

[0005] There is an ongoing effort in the semiconductor manufacturing industry to reduce consumption of gases with high global warming potentials (GWP). One approach has been to move towards a remote plasma clean system using nitrogen trifluoride (NF3), which is more environmentally favorable. The high cost of NF3 as well as the recurring threats of industry-wide shortages, however, have resulted in numerous investigations to optimize cleans, many of which are described in D. Wuebbles, Perfluorocompound Emission Control, Proceedings of the Global Semiconductor Industry Conference, Monterey, Calif., Apr. 7-8, (1998).

[0006] One method of cleaning the deposition reactor is with the help of a remote plasma source (RPS) unit, such as a pressurized torroidal plasma generator or a microwave generator, which is used to generate free fluorine for the cleaning process. The RPS unit is connected to the chamber by a valved conduit which is commonly available from a number of vendors. CVD reactor chamber walls will become contaminated with a film of BSG or BPSG every time a semiconductor wafer is coated with such a film. It is necessary to clean the reaction chamber walls periodically, depending on the film thicknesses being deposited. A ballpark estimate of the maximum thickness before a clean is essential is about 20,000 Å (2,000 nm or 2 &mgr;m) of film deposition so as to prevent these deposits from flaking off the chamber surfaces and contaminating wafers in the form of particulates.

[0007] The remote plasma chamber will usually be pressurized with an inert gas such as argon (Ar), which is used to “ignite” the plasma and a fluorine-containing compound, such as NF3. The NF3 is dissociated into a low-energy plasma, resulting in the generation of free fluorine which is fed into the CVD chamber through the conduit. The pressure in the reactor is controlled by a so-called “throttle valve.”

[0008] A typical remote plasma source (RPS) cleaning processes for borosilicate glass (BSG) and borophosphosilicate glass (BPSG) films involves a main clean of the reaction chamber at 2.2 torr with NF3 and Ar set at 950 sccm and 1400 sccm, respectively, followed by a second cleaning step with identical flows but with the throttle valve set to 1600 steps (to clean the back of the throttle valve). The throttle valve at 800 steps is fully open, but only one surface of the throttle valve is thereby cleaned. To clean the back surface that has not come into contact with the fluorine, the throttle valve must flip to the 1600 steps position, at which setting the backside of the throttle valve is cleaned. If the throttle valve is not cleaned adequately, pressure faults occur due to the inadequate seal made by the throttle valve. The cleaning durations are not optimized because of the lack of a suitable endpoint detection system to determine when cleaning is complete.

[0009] What is needed is an optimized method of cleaning BSG and BPSG films so as to minimize cleaning time, costs, and fluorine effluents.

SUMMARY OF THE INVENTION

[0010] The above disadvantages of the prior art are addressed by a method for a remote plasma source (RPS) cleaning process for BSG and BPSG films deposited in a CVD chamber. The presently disclosed method includes controlling the pressure within the chamber, introducing Ar into the chamber, introducing NF3 into the chamber, adjustably spacing a heater relative to the chamber, and adjusting the temperature within the chamber. The actual temperature used will depend on the temperature used for the deposition process. Additionally, a methodology for determining the endpoint of the clean using residual gas analysis and the evolution of foreline pressure is described.

[0011] These and other aspects, features and advantages of the present disclosure will become apparent from the following description of exemplary embodiments, which is to be read in connection with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

[0012] FIG. 1a and 1b graph the effects of process parameters upon etch rate.

[0013] FIG. 2 graph the effects of process parameters upon etch rate using a statistical software package sold by SAS institute, Inc. under the tradename JMP.

[0014] FIG. 3 and 4 show graphs of experimental results and statistical analysis.

[0015] FIG. 5 shows some typical values for using the method of the invention.

[0016] FIG. 6 shows an embodiment of the invention.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

[0017] Disclosed herein is a method for an optimized remote plasma source (RPS)—based cleaning for BSG and BPSG films deposited in an Applied Materials CxZ chamber. It is found that the dominant parameters for such an optimization are chamber pressure, NF3 flow, and argon flow. An optimized set of parameters is disclosed and verified with stability tests using blanket wafers. These are wafers with no patterning on them, basically bare silicon wafers. It is disclosed that overall cost reductions of about 20% per wafer and process time reductions of about 30% are attainable over the prior art using the methods of the invention.

[0018] The etch rate measured on oxide wafers is not an adequate measure of the clean efficiency. The “effective” etch rate (i.e. total film thickness deposited on the wafer divided by the total clean time ) was compared for different films (BSG and BPSG) and different thicknesses. The mean “etch rate” for BSG increases from 8406 Å/min (Prior Art) to 12304 Å/min (New Process) and for BPSG from 15062 Å/min (Prior Art) to 21028 Å/min (New Process). This represents a 46% increase for BSG and a 39.6% increase in etch rate for BPSG.

[0019] Referring to FIGS. 1a and 1b, to determine the effects of the critical process parameters, such as chamber pressure, NF3, Ar, spacing (the distance between the heater on which the wafer sits and the shower head from which the process gases are delivered), and the like on the etch rate, experiments were conducted with thermal oxide wafers.

[0020] As can be seen in FIG. 1a, baseline tests show a strong effect of wait time on the etch rate—thereby indicating that wafer temperature affects the etch rate. “Wait time” refers to the duration the wafer sits on the heater (inside the deposition chamber) before the process gases are introduced. The longer the wafer sits in the chamber, the closer the temperature of the wafer approaches that of the heater (i.e., 480C.). “9PT” and “49PT” refer to the number of sites where thickness is measured, namely 9 locations or 49 locations.

[0021] Referring to FIG. 1b, the test also demonstrates that the etch rate exhibits a transient behavior during the first few seconds of the etch. “Power” refers to the modeling equation, specifically how the relations between etch rate and time are fitted, and actually has nothing to do with the clean itself, but rather is generated by the statistical software used.

[0022] Additional tests indicated that the chamber condition prior to the experiment affected the etch rate (i.e., whether a BSG/BPSG film has been deposited and whether the chamber has been cleaned prior to the experiment). The results of these tests are not shown in the figures. These were separate tests where the goal was to determine what factors influenced the etch rate vs. time relationship. Experimental conditions were identified to minimize the variability introduced by these controllable factors.

[0023] Referring to FIGS. 2a and 2b, full factorial experiments were run to identify the effects of the critical variables. The results are summarized in FIGS. 2a and 2b through interaction plots provided by a commercially available statistical software package sold by SAS institute, Inc. under the tradename JMP. Chamber pressure and NF3 flow rate have the strongest effects on etch rate. Ar was fixed at 1600 Sccm (standard centimeter cubed per minute) because it has a positive influence on uniformity and would help to provide a more stable plasma at higher NF3 flow rates. However, the oxide etch rate tests essentially provided information about the main effects on the localized etch rate above the heater and did not address the etch rate at other locations in the chamber. One useful tool that was utilized is a quartz window near the rear of the CVD chamber, which allows a visual observation of the film deposition and removal (during the clean). The time to clean the window is found to follow a linear trend with increasing film thickness.

[0024] FIG. 2a shows the “interaction plots”—this is a useful graphical method of analyzing the effect of multiple variables on a response. For example, you can see simultaneously the effects of pressure, spacing, and NF3 flow on Etch rate: the X-axis consists of three segments—pressure 1.78 to 2.64 torr; spacing—440 to 800 mils; and NF flow—880 to 1050 sccm. The Y-axis has three segments to show the effect of two of the three variables. For example, with pressure fixed at 1.78 torr, increasing NF3 would increase etch rate and increasing spacing would also increase etch rate, but increasing spacing has a much greater impact. These plots also show if there are “interactions” between variables. Interactions appear where two lines are not parallel to one another.

[0025] FIG. 2b is a simpler graph called a “prediction profile”—this is a tool provided by the JMP statistical software to find an “optimum” setting for multiple variables and responses. In other words, you can move around the set points for each variable and see how it impacts the response.

[0026] Referring to FIG. 3, a response surface design was used to determine the optimum settings for NF3 and chamber pressure during the clean. In the graph called Actual vs. Predicted, the different lines show the confidence limits. This is a plot which shows the values (for the window clean time) predicted by the model (X-axis) and the actual values obtained during experiments (Y-axis). The other two graphs show the effect of individual variables, like pressure and NF3 flow, on the response. The response in this case is the time to “clean the window”. The “window” is a translucent quartz window located on the chamber lid near the foreline. During film deposition, the window builds up a white film. The cleaning of this window is a secondary tool to determine that the clean is working. The time to clean the window could be correlated to one of the peaks observed in the pressure versus time trace for the “backing pressure” (i.e., pressure measured in the foreline). The etch rate referred to in FIGS. 1a and 1b is the rate at which the oxide film on a flat wafer is removed by the clean. The etch rate measured using a thermal oxide wafer was not an appropriate method to determine the clean efficacy (because it only addresses the etch rate above the heater and does not take into account the cleaning of the entire chamber). The results are summarized in FIGS. 4a and 4b with a prediction profiler.

[0027] Referring to FIGS. 4a and 4b, the prediction profiler, which graphically displays the effects of variables on responses, is shown. A minima in the time to clean the window at a pressure setting of about 3 torr of NF3 in the reaction chamber can be observed. The model fit indicates that 97% of the variation could be explained. On the other hand, increasing NF3 flow shows a continuous decrease in the clean time (i.e., increasing etch rate).

[0028] The results therefore identify two critical parameters for the cleaning method of the invention. The issues to be resolved are the NF3 flow rate and the clean duration. Because reduction in NF3 consumption is one of the desirable goals of the invention, increasing the NF3 appears counterproductive.

[0029] To verify the findings from the design of experiments (DOE), tests were conducted with varying NF3 flows and clean durations. The goal was to identify conditions under which a particle failure could be created over the course of a five-wafer deposition run. “Particle failure” means a situation wherein a wafer shows a high number of particles that exceeds the control limits, meaning the limit above which the parameter is considered to have “failed” and requires further investigation and passing results before production can be resumed. In addition, if there was a problem with incomplete cleans, a significant separation in the foreline pressure traces (a plot of the pressure in the foreline as a function of time) for consecutive wafers could be discerned. NF3 flows ranging between 950 sccm to 1,150 sccm were tested. The results confirm that at the 1,150 setting a particle failure could not be forced even with a 50% reduction in clean time, unlike the prior art.

[0030] The next step was to track the evolution of chamber parameters during the clean, such as foreline pressure, throttle valve position, chamber pressure, and the like, and correlate the variation of the parameters with the window clean time and residual gas analysis (RGA). The foreline is the conduit connecting a proportional, integral and derivative (“PID”) controlled servo-motor to the chamber. The proportional, integral and derivative (“PID”) controlled servo-motor is used to reduce the pressure in the chamber from atmospheric pressure (760 torr, or 1 atmosphere) to the operating pressure (3 torr, or 200 torr)—depending on whether it is the clean operation or the deposition process.

[0031] The residual gas analysis (RGA) allows identification of the gaseous species present in the chamber. The RGA was hooked up to the foreline just below the chamber, and 200 AMU (atomic mass units) runs were conducted to determine which species showed significant changes in partial pressure during the clean. The critical species were identified as SiFx, HF, Fx and BFx. While this list not all-inclusive, these species provide a very good estimate of the clean endpoint as will be demonstrated below.

[0032] Referring to FIGS. 4a and 4b,it can be seen that the foreline pressure during the clean provides a reproducible means of tracking the evolution of the clean. JMP was used to identify an excellent correlation between the window clean time and the 2nd peak in the foreline pressure. Both parameters exhibit a linear variation with film thickness. The foreline pressure exhibits a change in slope, which coincides with stabilization of F2, HF, SiF4, and BF3, as can be seen in FIG. 4a.

[0033] Referring to FIG. 4b, to verify the completion of a clean, an alternative approach is to fix the setting of the throttle valve position to the value required to achieve the desired pressure during the main clean, and to observe the variation of the chamber pressure. As can be seen in FIG. 4b,the chamber pressure can be seen to stabilize at the same time as seen in the experiments with the RGA.

[0034] Having optimized the clean recipe in terms of etch rate and clean duration, twenty five wafer verification runs were completed for the complete range of BSG and BPSG films at the optimized rate and duration as indicated in FIG. 5. Optimized rate and duration is dependent upon film thickness and the film chemistry, for example, BSG or BPSG. The films tested include 9,000 Å BSG, 7,000 Å BSG, 2,600 Å BSG, 5,500 Å BPSG, 11,000 Å BPSG and 16,500 Å BPSG. All the film properties were stable and particle performance was stellar. The etch rate of the method of the invention will typically be about 20% higher than that of the prior art.

[0035] FIG. 5 shows the preferred parameters for the methods of the invention. It provides details of what each of the parameters on an AMAT Centura 5200 should be set up for the “optimized” clean. Novel preferred values for various parameters are shown in bold. To generate the chart, the clean was done after every wafer for the BSG films and every two wafers for BPSG. In general, the clean will be implemented perhaps every three wafers for BPSG.

[0036] Referring to FIG. 5, steps 1 and 2, note that a NF3 Pump and Ar Strike, are identical to the prior art. Step 3, the NF3 Ramp is altered according to the invention by raising the flow rate of Ar above the 1400 sccm level typical of the prior art. It can be seen from the figure that this will be true for each of steps 3 through 6. In general, a first mass-flow controller will regulate the Ar rate which will be greater than 1,400 sccm, preferably greater than 1,500, more preferably from about 1,500 to about 1,800, or from about 1,600 to about 1,750 sccm.

[0037] The rate of NF3 flow is also increased over the prior art, but only for steps 3 through 5. The standard prior art flow rate for NF3 for these steps in 950 sccm. For the method of the invention, a second mass-flow controller will regulate the NF3 rate which will generally be greater than 950 sccm, preferably at least 1,000 sccm, still more preferably from about 1,100 to about 1,200 sccm, or generally about 1,150 sccm.

[0038] Note in Step 4, the Clean, the chamber pressure is elevated above the 2.2 torr typical of the prior art. In general, the chamber pressure will be above 2.2 torr, preferably between 2.5 and 4.5 torr, or about 3 torr.

[0039] Referring again to FIG. 5, note that in steps 5 through 9, the heater space controller is set so that heater spacing is elevated over the 600 mils typical of the prior art. This is optional, but will enhance performance further. The preferred range for the heater space controller for the invention is from greater than 600 to about 1,500 mils.

[0040] Note that in Steps 4 and 5, the maximum step time is labeled TBD, meaning To Be Determined. This is because the duration of these steps is a function of the type of film (BSG takes longer than BPSG) and the film thickness. This is true for the prior art as well, but the method of this invention will generally show step times about 20% to 30% shorter than the prior art.

[0041] The temperature used was 480° C., but of course any temperature effective in removing the film is within the range of the invention. The actual temperature used will depend on the temperature used for deposition process. Ideally, a temperature control sets the temperature for the clean processes to be about identical to the temperature set for the deposition processes so that there is no impact on the repeatability of the film properties from one wafer to the next.

[0042] It is to be understood that all physical quantities disclosed herein, unless explicitly indicated otherwise, are not to be construed as exactly equal to the quantity disclosed, but rather as about equal to the quantity disclosed. Further, the mere absence of a qualifier such as “about” or the like, is not to be construed as an explicit indication that any such disclosed physical quantity is an exact quantity, irrespective of whether such qualifiers are used with respect to any other physical quantities disclosed herein.

[0043] Referring to an embodiment in FIG. 6 for the clean process, the inside of a chamber 602 is pressured controlled, preferably between 2.5 torr and 4.5 torr to about 3.0 torr, by a PID servo device 604. Connected to the chamber 602 is a first mass-flow device 608 for introducing Ar into the chamber 602. The Ar provides a more stable plasma at higher NF3 flow rates. Further, to strengthen the etch rate by controlling NF3 flow rate, connected to the chamber 602 is a second mass-flow device 606 for introducing NF3 into the chamber 602. Increasing heater spacing also can increase etch rate. Therefore, within the chamber 602 is a heater spacing device 612 that can adjustably space a heater 614 relative to the chamber 602. A temperature control 610 can adjust the temperature within the chamber 602 so that the clean processes' temperature is substantially identical to the temperature set for the deposition processes. The result is that there is no impact on the repeatability of the film properties from one wafer to the next.

[0044] While preferred embodiments have been shown and described, various modifications and substitutions may be made thereto without departing from the spirit and scope of the invention. Accordingly, it is to be understood that the present invention has been described by way of illustration only, and such illustrations and embodiments as have been disclosed herein are not to be construed as limiting to the claims.

[0045] The teachings of the present disclosure are preferably implemented as a combination of hardware and software. Moreover, the software is preferably implemented as an application program tangibly embodied on a program storage unit. The application program may be uploaded to, and executed by, a machine comprising any suitable architecture. Preferably, the machine is implemented on a computer platform having hardware such as one or more Central Processing Units (“CPUs”), a Random Access Memory (“RAM”), and Input/Output (“I/O”) interfaces. The computer platform may also include an operating system and microinstruction code. The various processes and functions described herein may be either part of the microinstruction code or part of the application program, or any combination thereof, which may be executed by a CPU. In addition, various other peripheral units may be connected to the computer platform such as an additional data storage unit and an output unit.

[0046] It is to be further understood that, because some of the constituent system components and steps depicted in the accompanying drawings may be implemented in software, the actual connections between the system components or the process function blocks may differ depending upon the manner in which the present disclosure is programmed. Given the teachings herein, one of ordinary skill in the pertinent art will be able to contemplate these and similar implementations or configurations of the present disclosure.

[0047] Although illustrative embodiments have been described herein with reference to the accompanying drawings, it is to be understood that the present disclosure is not limited to those precise embodiments, and that various changes and modifications may be effected therein by one of ordinary skill in the pertinent art without departing from the scope or spirit of the present disclosure. All such changes and modifications are intended to be included within the scope of the present disclosure as set forth in the appended claims.

Claims

1. A method for cleaning a CVD reaction chamber, comprising the steps of:

controlling the pressure within the chamber;

introducing Ar into the chamber;

introducing NF3 into the chamber;

adjustably spacing a heater relative to the chamber; and

adjusting the temperature within the chamber.

2. The method of claim 1, wherein the pressure is controlled between 2.5 Torr and 4.5 Torr.

3. The method of claim 1, wherein the pressure is controlled at about 3 Torr.

4. The method of claim 1, wherein the Ar is introduced at a rate of about 1,500 sccm to about 1,800 sccm.

5. The method of claim 1, wherein the Ar is introduced at a rate of about 1,600 sccm to about 1,750 sccm.

6. The method of claim 1, wherein the NF3 is introduced at a rate of about 1,100 sccm to about 1,200 sccm.

7. The method of claim 1, wherein the NF3 is introduced at a rate of about 1,150 sccm.

8. The method of claim 1, wherein spacing the heater relative to the chamber comprises spacing the heater from greater than 600 mils to about 1,500 mils relative to a showerhead.

9. The method of claim 1, wherein adjusting the temperature within the chamber comprises adjusting the temperature to be substantially identical to the temperature used in the deposition process.

10. The method of claim 1, including the step of cleaning a backside of a throttle valve comprising setting a pressure set point to throttle to 1600 steps, introducing Ar at a rate of about 1,600 sccm into the chamber and introducing NF3 at a rate of 1,150 sccm into the chamber.

11. The method of claim 1, including the step of optimizing the clean duration for a specific film comprising correlating a clean endpoint with a change in a slope of a foreline pressure versus time plot graph.

12. The method of claim 11, wherein said specific film includes one of Borosilicate (BSG) film and Borophosphosilicate (BPSG) film.

13. The method of claim 1, including the step of optimizing the clean duration for a specific film comprising correlating a clean endpoint with a stabilization of a species selected from the group consisting of HF, SiFx, BF3, and Fx.

14. The method of claim 13, wherein said specific film includes one of Borosilicate (BSG) film and Borophosphosilicate (BPSG) film.

15. The method of claim 1, including the step of optimizing the clean duration for a specific film comprising correlating the time to clean a window within a CVD reaction chamber with a second peak in a foreline pressure versus time plot graph.

16. The method of claim 15, wherein said specific film includes one of Borosilicate (BSG) film and Borophosphosilicate (BPSG) film.

17. An apparatus for cleaning a CVD reaction chamber, the apparatus comprising:

servo device for controlling the pressure within the chamber using PID methodology;

first mass-flow device for introducing Ar into the chamber;

second mass-flow device for introducing NF3 into the chamber;

heater spacing device for adjustably spacing a heater relative to the chamber; and

temperature control device for adjusting the temperature within the chamber.

18. A program storage device readable by machine, tangibly embodying a program of instructions executable by the machine to perform method steps for cleaning a CVD reaction chamber, the method steps comprising:

controlling the pressure within the chamber;

introducing Ar into the chamber;

introducing NF3 into the chamber;

adjustably spacing a heater relative to the chamber; and

adjusting the temperature within the chamber.