US 20060027566 A1
A method and system for generating control settings for a multi-parameter control system. The interdependencies of processing tools and the related effect on semiconductor wafers within a processing tool is factored into a mathematical model that considers desired and measured wafer quality parameters in the derivation of specific solutions of sets of possible quality parameter adjustments. A selection process determines a set of adjustments such as one that results in minimal changes to the process.
1. A method for generating temperature control settings for a multi-zone semiconductor furnace, comprising:
measuring a measured film thickness change from a semiconductor wafer from each zone of said multi-zone semiconductor furnace; and
calculating temperature control settings for each zone of said multi-zone semiconductor furnace from a preferred set of temperature adjustments derived from said measured film thickness change and a desired film thickness change for each zone.
2. The method, as recited in
3. The method, as recited in
4. The method, as recited in
5. The method, as recited in
6. The method, as recited in
7. The method, as recited in
8. The method, as recited in
9. The method, as recited in
10. The method, as recited in
This application is a continuation of application Ser. No. 10/378,757, filed Mar. 4, 2003, pending.
1. Field of the Invention
The present invention relates generally to processes for semiconductor wafers and, more particularly, to control and specification of temperature processes for a semiconductor processing furnace.
2. State of the Art
Semiconductor processing advancements continue to facilitate reduction in feature dimensions which in turn results in increased performance and circuitry integration. While feature size reductions have been considerable, continuous improvements are being pursued. With the reduction in feature size, the criticality of accurate control during manufacturing processes becomes more essential. Specifically, the temperature imposed upon a semiconductor wafer during a manufacturing process affects the diffusion of dopants, as well as the deposition of materials on the semiconductor wafer. Thus, it is important that processing systems achieve accurate control to obtain the desired effect on the semiconductor wafer undergoing processing.
In order to maintain an acceptable level of quality control, many semiconductor processes are performed in iterative incremental “runs” that allow for control feedback to be entered into the processing specification of each subsequent run. On a specific run, a process specification may include a set-point temperature, temperature process duration, a temperature ramp rate, etc., that define the overall thermal process to be encountered by the semiconductor wafer.
Process specifications may be further complicated by the introduction of a gas or other vapor which may also be affected by the temperature and thermal profile. Therefore, various temperature control problems must be attended to by a processing system in order to maintain an acceptable level of quality control, i.e., yield, of semiconductor wafers. In an effort to enhance the yield of operable circuits on a wafer, it would be desirable that each wafer in a group or “batch” be subjected to the same temperature conditions over an entire thermal processing cycle or run. If uncontrolled, the variations in the processing of the semiconductor wafers within a batch result in unacceptable deviations and such deviations are further exaggerated over subsequent runs which inject further deviations into the overall process.
A still further temperature control problem exists in thermal processing systems, such as furnaces, that utilize multiple heating elements. Multiple-zone furnaces can be used to better optimize the thermal profile in larger furnaces capable of handling a large number of wafers. Large capacity furnaces are designed with multiple heating elements to provide more uniform heating across the entire cavity of the furnace. However, due to the nature of a zone's heat flow, as well as the interactive nature of adjacent zones within an open cavity, thermal variations can and do exist within a single cavity of a multi-zone furnace. While it is possible to control a specific zone within a multi-zone configuration, the affect of the interaction between zones has largely gone unaddressed and unappreciated. While one or more wafers within a specific zone may be evaluated following the completion of a specific run, thermal compensation for such processing deviations have been limited to an adjustment of that specific zone for a subsequent thermal process during a subsequent run of that specific wafer or group of wafers, regardless of the influence or affect of such an adjustment on other wafers within other zones of the system.
Thus, it is desirable to implement a system and method that takes into account the affect of modifications of thermal adjustments implemented in one zone and the related effect of such an adjustment on the other zones within the system.
The present invention includes methods and systems for generating control settings for use in a multi-parameter semiconductor tool that processes semiconductor wafers on a run-to-run basis. As used herein, the term “semiconductor wafer” includes not only conventional silicon wafers, but also other bulk semiconductor substrates such as silicon-on-insulator (SOI), silicon-on-glass (SOG) and silicon-on-sapphire (SOS) substrates, as well as other semiconductor materials such as, by way of example only, gallium arsenide, and iridium phosphide. Generally, the present invention provides a polynomial system model of a processing tool which enables simultaneous calculation of parameter adjustments in pursuit of run-to-run control of a wafer process. The method takes into consideration the cross-influence of the changes in each controlled parameter and the effect such change has on all of the other parameters of interest. The present invention facilitates an improved yield due to the more precise characterization of the influence of a controlled parameter adjustment on the quality parameters of wafers in other chambers within the tool. Furthermore, the control settings are calculated from models as opposed to being adjusted using ad hoc approaches.
The present invention also includes specific methods and systems for generating control settings for use in a multi-zone semiconductor furnace that processes semiconductor wafers on a run-to-run basis. One specific embodiment of the present invention provides a polynomial system model of a diffusion furnace which enables simultaneous calculation of temperature adjustments in pursuit of run-to-run control of a wafer process. The method takes into consideration the cross-influence of the temperature changes in each zone and the effect such change has on all of the other zones. The present invention facilitates an improved yield due to the more precise characterization of the influence of a zone's temperature adjustment on the film growth of wafers in other zones within the furnace. Furthermore, the temperature control settings are calculated from models as opposed to being adjusted using ad hoc approaches.
According to one embodiment of the present invention, temperature control settings are generated for controlling a multi-zone semiconductor furnace used for the processing of semiconductor wafers. The zonal influence between the multiple zones within a multi-zone furnace is characterized and a transfer function is derived. Following a run of the wafers within the multi-zone furnace, a wafer from each zone is evaluated. The thickness of the deposited film is measured or, alternatively, a thickness of the deposited film is otherwise characterized for each of the zones for use as inputs into the polynomial model. Optionally, a filter is injected into the model following the measurement of the film thickness. Such a filter facilitates the removal or mitigation of random noise injected into the system by the measurement characterization instruments or random fluctuations of the process.
Temperature adjustments are calculated from the measured film thicknesses, as well as from the desired film thickness for each of the zones. The calculation of possible temperature adjustments results in a number of possible solutions corresponding to the order of the polynomial. A preferred solution or set of temperature adjustments is calculated according to a predefined criterion. One such criterion is the minimum overall adjustments to the system. Such a solution inserts the least perturbations into a heretofore refined, but otherwise comprehensively uncharacterized process.
According to another embodiment of the invention, a multi-zone semiconductor furnace controls a run-to-run semiconductor process on semiconductor wafers by applying the control settings respectively generated from the process above to control the respective zones within a multi-zone semiconductor furnace. The present invention finds application not only to diffusion processes but to all semiconductor processes that grow or deposit films in a chamber or environment that has separately controllable environment control elements, such as heating elements.
According to yet another embodiment of the present invention, a multi-parameter control system controls a run-to-run semiconductor process on semiconductor wafers by applying the control settings respectively generated similar to the process above to control respective quality wafer parameters within a multi-chambered or single chambered tool. The present invention finds application not only to semiconductor processes that grow or deposit films but to all semiconductor processes that generally have separately controllable control elements, such a gas flow controllers, chemical liquid flow controllers, plasma control elements, temperature and humidity control elements, and the like.
In the drawings, which illustrate what is currently considered to be the best mode for carrying out the invention:
One exemplary and illustrative process includes the deposition of nitride which is generally done in a Low Pressure Chemical Vapor Deposition (LPCVD) process that may be thermally driven by irradiative heat transfer which in turn drives or induces a chemical reaction on a substrate, such as a silicon wafer, between, for example, dichlorosilane and ammonia to deposit what is commonly referred to as silicon nitride or oftentimes is simplistically referred to as just “nitride.”
The equation below illustrates the chemical reaction of this specific illustrative example.
Those of ordinary skill in the art appreciate that a growth rate, for example, of nitride film is dependent upon process variables including: (1) heat and temperature, (2) pressure, (3) reactive gas flow, (4) reactant ratios, and (5) time. Due to the partitioned nature or zoning of wafer chamber 14, the plurality of semiconductor wafers 16 is subjected to independent control of the available heat for each zone 30-38. Such independent or regionalized control assists in generating a more uniform chemical reaction across the entire vertical array of semiconductor wafers 16, thereby resulting in a more uniform or desired film thickness on each of the plurality of wafers 16. In a process control environment, each zone may integrate a test wafer within the vertically stacked wafers 16 for use as a wafer from which to derive a measured film thickness between subsequent runs of wafer 16 within semiconductor furnace 10.
The film thickness on a semiconductor surface achieved through a process within semiconductor furnace 10 exhibits a strong correlation between both the temperature in a particular zone in the presence of a corresponding gas and the maintenance of temperature in a zone over a period of time. It is appreciated that certain temperature ramps are required for obtaining uniform depositions, i.e., film deposition rate generally exhibits linear dependence on the process temperature. After a specific period of time, a strong linear relationship between film thickness and elapsed time exists as process temperature and flow of gas reactants remains relatively constant.
Therefore, since the film deposition rate is dependent on a temperature gradient across an entire wafer chamber 14, an entire process at a stable linear stage within semiconductor furnace 10 may be described using a model unifying all zones. However, the present invention appreciates that any temperature change in any one zone will affect film thickness in all other zones. Therefore, an improved approach is to model all zones affecting each other in accordance with a thermal transfer function which directly affects resultant film thickness.
In the proposed model of the present embodiment, the process may be described using a system of polynomial equations approximating the transfer functions in which the resultant film thickness change in each zone depends on all of the changes of zone temperatures. For example, for the five-zone furnace exemplary illustration, the process may be mathematically depicted as follows:
It should be noted that ΔTn in equations (3) represent vectors of temperature changes in corresponding power, not the power of a vector of temperature changes. The system of equations (3) quantitatively describes processing in the illustrated five-zone diffusion furnace with modifications to any number of zones through the scaling of the matrices. It should be noted that equations (3) do not contain members describing combined influence of the temperature changes as they are considered negligible when compared with the prime members of equations (3). Additionally, it should be noted that equations (3) represent a system of first order differential equations describing a diffusion furnace reaction to the temperature changes in each zone. Those of ordinary skill in the art may determine a similar system of differential equations for any semiconductor processing tool describing the tool reaction to the process control parameters changes using a design of the experiment (DOE) approach.
Multi-parameter semiconductor processing system 39 is further comprised of a multi-parameter control system 47 which utilizes the modeling described herein for the derivation of the specific commands illustrated generally as specific tool process parameter settings 49. The calculation of tool process parameter settings 49 is further performed within multi-parameter control system 47 and is calculated from inputs illustrated as desired wafer quality parameters 51 (e.g., process target values, film thickness, particle count, dopant concentration, critical dimension, etc.) and from post-run measured wafer quality parameters 53, generally depicted as measured wafer quality parameters 53. Measured wafer quality parameters 53, derived from an evaluation tool such as a wafer quality parameters measurement device 55, provides feedback to multi-parameter control system 47 upon the evaluation of the quality parameters post-run on specific test or process wafers, such as wafers 57. As illustrated, once wafers 57 undergo an evaluation by wafer quality parameters measurement device 55, wafers 57 may be returned to semiconductor process tool 9 for subsequent run processing. While the present illustration depicts wafers 57 being removed from semiconductor process tool 9 for an ex situ measurement, the present invention further contemplates measurement device capability being present within semiconductor process tool 9 for an in situ measurement of a wafer quality parameter.
Multi-zone thermal processing system 40 is further comprised of a furnace multi-zone temperature control system 48 which utilizes the modeling described herein for the derivation of the specific zone temperature 42 illustrated generally as temperature control settings 50. The calculation of temperature control settings 50 is further performed within multi-zone temperature control system 48 and is calculated from inputs illustrated as desired film thickness 52 and from post-run measured film thicknesses for each zone, generally depicted as measured film thickness 54. Measured film thicknesses 54, derived from an evaluation tool such as film thickness measurement device 56, provides feedback to multi-zone temperature control system 48 upon the evaluation of the deposited films post-run on specific wafers, such as test wafers 58. As illustrated, once test wafers 58 undergo an evaluation at film thickness measurement device 56, they may be returned to multi-zone semiconductor furnace 10 for subsequent run processing. While the present illustration depicts test wafers 58 being removed from multi-zone semiconductor furnace 10 for an ex situ measurement, the present invention further contemplates film thickness measurement capability being present within semiconductor furnace 10 for an in situ measurement of a film thickness.
Table 1 below summarizes measured data obtained during an arrangement of the example described above with all data being shown as a reflection of the baseline data which depicts changes in film thickness as measured in angstroms in all zones with the corresponding temperature change in each specific zone. Additionally, the test results were rounded to zero if the absolute data value was below 3 angstroms in order to mitigate the effects of any measurement noise.
Nitride Process Example of Film Thickness Changes Resulting from Temperature Changes in Corresponding Zones.
In Table 1, the designation, “Zone x PLUS,” corresponds to a plus 1.5° C. temperature change in the corresponding zone while a “Zone x MINUS” designator corresponds to a minus 1.5° C. temperature change in the respective zone. In the present test case, three data points (PLUS, baseline, MINUS) for each zone enables the generation of quadratic equations that describe the process in a narrow temperature zone of approximately 3° C. around a specific process temperature. The specific quadratic equation becomes:
Given A2 and A1 matrices and film thickness changes for each zone, the system of quadratic equations (4) may be solved for ΔT and temperature changes for each zone may be calculated for a specific furnace. It should be noted that the equations (4) hold for temperature charges outside of the experimental temperature range, however, the accuracy can vary.
The mathematical model illustrates a relationship between film thickness changes and zone temperature changes. As such, it is possible to solve the equations to determine a relationship or function in which each zone temperature change can be treated as a function of the film thickness changes for all zones, namely:
Solving the system of equations (3) for ΔT arrives at the feedback control mechanism for performing run-to-run control within the multi-zone furnace as calculated from coefficients αki,j and post-process measured film thickness FTi at each zone after each run:
Although an analytical solution for the system or model of polynomials in equations (3) is not directly derivable, it is possible to solve equations using numerical methods. By deriving iteratively approximating roots of the system through numerical solutions, practical accuracy is attainable for a solution of ΔT. In a practical implementation of the deployment of the modeling of the present invention, it is desirable to minimize process fluctuations from run-to-run by calculating suggested zone temperature settings, Tis, using all measured film thicknesses and previous temperature settings for each zone as is inherently described in equation (6).
A thorough solution of equation (3) may result in a number of possible real root sets. Therefore, a selection of a preferable or preferred set of temperature adjustments from among the available solutions is desirable. One such selection criteria is to select a solution that results in a minimal amount of adjustments to the system, meaning the minimum overall change of a somewhat characterized process. Therefore, one such selection process for deriving a preferred set of temperature adjustments from a larger set of solutions of temperature adjustments is to select the set of temperature adjustments that results in the smallest overall sum of temperature adjustments. Such a minimization may be arrived at using a distance formula to select the set that results in minimum changes in the furnace control settings.
A form of the Euclidean distance function is an acceptable selection process and is defined as:
Additionally, noise associated with wafer quality parameters measurement device 55 (
Additionally, noise associated with film thickness measurements from the film thickness measurement device 56 may be removed through the location of a filter, illustrated as filter 62 which removes noise attributed to the film thickness measurement process. One such filtering scheme is illustrated as an exponentially weighted moving average (EWMA) which is utilized for estimating process deviations. Other techniques are also contemplated within the scope of the present invention including the utilization of a proportional integral derivative (PID) control to perform filtering.
The modeling technique of an embodiment of the present invention utilizes measured film thicknesses from all zones to simultaneously calculate the temperature change for each specific zone which applies the temperature control strategy more effectively than by treating each zone separately. Furthermore, selecting or voting the best solution from among the pool of possible solutions makes the process more stable as well as providing for a more uniform process.
Once the measured film thickness is derived, an optional act 70 filters noise as induced by the measuring process. It is appreciated that measurement tools inject noise into a system because of their quantization properties. Therefore, if the noise is appreciable and can be readily characterized, then filtering of the noise is desirable. Similarly, the process itself may be subject to random fluctuations that are not captured in the model. Therefore, optional act 70 helps to remove combined random noise of the manufacturing process and measuring process.
An act 72 receives both the measure film thickness 68 as well as the desired or target film thickness 74 for performing the calculation of at least one set of temperature adjustments for each of the zones based upon the mathematical modeling previously described. An act 76 selects a preferred set of temperature adjustments from among a plurality of possible sets when the polynomial solution provides for a plurality of possible solutions.
An act 78 calculates temperature control settings for each zone of the multi-zone semiconductor furnace from the selected set of temperature adjustments. As described above, the selection criteria may vary depending upon the process sensitivities. One such criteria includes selecting a set of possible temperature control settings that interjects the minimal adjustments to the system. Such a least-sum solution described above minimizes the adjustments from the norm thereby providing lesser disruptions to a somewhat characterized process.
An act 80 provides an optional filtering process for data filtering to accommodate stability or other process noise attributes derived from heuristic analysis. Exemplary filtering techniques include PID and EWMA control schemas appreciated by those of ordinary skill in the art. Temperature control settings process 66 then returns (
It is also contemplated that the aforementioned techniques, systems, and methods may be applied to other types of semiconductor processes that have one or more desired wafer quality parameters and one or more process control parameters. By way of example and not limitation, processes such as Chemical Vapor Deposition (CVD), dry etching, wet etching, photo lithography and other semiconductor manufacturing processes are also suitable processes for controlling using the approaches described herein.
Specific embodiments have been shown by way of example in the drawings and have been described in detail herein, however the invention may be susceptible to various, modifications and alternative forms. It should be understood that the invention is not intended to be limited to the particular forms disclosed. Rather, the invention includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention as defined by the following appended claims.