Micromachines acoustic sensor for monitoring electrochemical deposition

[0001] The U.S. Government may have a paid-up license in this invention and the right in limited circumstances to require the patent owner to license others on reasonable terms as provided for by the terms of Grant No. EEC-9402723 awarded by the National Science Foundation of the U.S.

[0002] The present invention relates generally to sensors and fabrication methods, and more particularly, to micromachined acoustic sensors for monitoring electrochemical deposition, methods for fabricating micromachined acoustic sensors, and methods for in-situ monitoring of electrochemical deposition processes using such sensors.

[0003] As microsystem feature sizes continue to shrink, the level of integration and complexity of microelectronic devices continues to increase. Microsystems packaging must also evolve to accommodate these trends in integrated circuit (IC) technology. Packaging requirements, such as cost, power, input/output (I/O) count, and operating frequency, for example, must be consistent with IC roadmap projections. Moreover, manufacturing challenges also increase as feature sizes decrease. Improvements in fabrication methods and processes are critical if inexpensive and reliable devices with such small dimensions are to be produced. Semiconductor manufacturing processes require hundreds of successive processing steps, and each of these can potentially contribute to yield loss and increased cost. Therefore, manufacturing technology development requires not only advancement in fabrication processes, but also advances in process monitoring and control.

[0004] Effective process control can be provided by in-situ monitoring, process modeling, and real-time, closed-looped operation. In-situ monitoring employs sensors to measure and monitor critical manufacturing process parameters in real time, thus permitting early detection of defective components and process shifts. There is a shortage of sensors that have been developed as in-situ process monitors, however. Many current methods utilize optical techniques to monitor various process parameters. These methods rely on “looking” at a process to monitor its status. While these methods are relatively common for in-situ process monitoring, the equipment they employ can be expensive and can have equipment mounting and calibration challenges.

[0005] It would be desirable to have inexpensive and reliable sensors for monitoring electrochemical deposition, methods of fabricating such sensors and methods for in-situ monitoring of electrochemical deposition processes using such sensors.

[0010] Disclosed below are alternative or complementary techniques to those discussed in the Background section that are based on acoustic methods of in-situ monitoring. Essentially, acoustic methods involve “listening” to the manufacturing process to enhance the amount and sensitivity of data collection to facilitate process diagnostics and control.

[0011] Described herein are acoustic devices (microsensors) that function as microphones. The sensing elements of these microsensors are usually bendable beams or membranes of silicon dioxide, silicon nitride, or silicon layers. Because of the application of IC and micromachining fabrication techniques, these elements can be created with a thickness as small as 1-30 μm. See, for example, J. Franz and D. Hohm. “Silicon Acoustic Sensors.” Eurocon '86: Part 2 (1986): 156-161. Thus, vibrations from undesired sources are less likely to interfere with the input signal to the microsensors. Also, by using microfabrication techniques, various electrode patterns can be obtained, improving the sensitivity and frequency response of these microsensors.

[0012] Disclosed herein is a micromachined piezoelectric microphone sensor 10, or acoustic sensor 10, (FIG. 1e) which may be used for in-situ monitoring of electrochemical deposition processes, such as nickel-iron (Ni—Fe) electrochemical deposition, for example. During operation, changes occur in a plating bath 31 (FIG. 2) used in the electrochemical deposition process as the nickel-iron alloy is deposited. Changes in the process medium (i.e., the solution 32) affect the acoustic response of the acoustic sensor 10. An exemplary acoustic sensor 10 was used in an electroplating process and its response was monitored during depositions. By mapping (calibrating) the response of the acoustic sensor 10 to the corresponding film thickness of the plated alloy measured at certain times, a predictive model of the thickness of the plated alloy was derived as a function of sensor output and plating time. Such a model provides for real-time monitoring of the thickness of the plated nickel-iron alloy, for example. This is discussed in detail below.

[0013] Sensor Design and Fabrication

[0014] The microphone sensor 10 described herein exploits the piezoelectric effect, in which a mechanical stress applied to a polarized crystal produces a mechanical deformation in the crystal and results in an electric charge.

[0015] An exemplary reduced-to-practice acoustic sensor 10 is shown in FIG. 1e, and comprises a membrane 12, an encapsulated piezoelectric layer 14, and segmented surface electrodes 19. This concept is generally discussed in J. Franz and D. Hohm. “Silicon Acoustic Sensors.” Eurocon '86: Part 2 (1986): 156-161. A highly doped silicon layer comprises the membrane 12 because silicon does not possess the tensile stress associated with silicon nitride or silicon dioxide layers. An exemplary silicon membrane 12 is approximately 3.5 μm thick. Zinc oxide (ZnO) serves as the piezoelectric layer 14 and is approximately 0.8 μm thick. The ZnO piezoelectric layer 14 is encapsulated by a silicon nitride layer 15, which helps to reduce surface charge leaks. An exemplary deflectable diaphragm structure 20, or diaphragm 20, comprising the membrane 12, piezoelectric layer 14, and silicon nitride layer 15, and which is the sensing element of the microphone sensor 10, has dimensions of 1270 μm×1270 μm.

[0016] To optimize performance, the microphone sensor 10 was modeled using finite element analysis software. The results of the analysis indicated that movement of the sensor's membrane 12 due to sound pressure experiences the greatest bending stress in the middle and outer region of the defined area of the diaphragm 20. Thus, electrodes 19 (FIG. 1e) were placed in these regions on a surface of the sensor 10.

[0017] Referring to FIGS. 1a-1e, they illustrate an exemplary fabrication process 20 for making a micromachined acoustic sensor 10. The sensor 10 was fabricated using IC and micromachining techniques, enabling the batch fabrication of small, precise devices that may be reproduced. A two-inch n-type (100) silicon wafer 13 (substrate 13 or layer 13) was doped with a high concentration of boron atoms to form a doped silicon layer 12 at the top surface (FIG. 1a). The doped silicon layer 12 comprises the membrane 12. The doped silicon layer 12 acted as an etch stop in a later processing step. A thin piezoelectric layer 14 comprising ZnO was then deposited by RF sputtering (FIG. 1b). Silicon nitride layers 15, 16 were deposited over the ZnO piezoelectric layer 14 as well as on the back of the wafer 11 using plasma-enhanced chemical vapor deposition (FIG. 1c). The silicon nitride layer 15 forms an encapsulating layer over the piezoelectric layer 14. The silicon nitride layer 16 on the back side of the silicon wafer 13 comprises a masking layer 16.

[0018] A post-deposition anneal in an inert gas was performed to relax the stress in the ZnO piezoelectric layer 14. The silicon nitride layer 16 on the back side of the wafer 11 was patterned to define an opening 17 in the silicon nitride layer 16 (FIG. 1c). The pattern, which defined the surface of the diaphragm 20 was etched into the silicon nitride layer 16 by reactive ion etching (FIG. 1c). The wafer 13 was then back-etched in an aqueous solution of potassium hydroxide (KOH) to produce an opening 18 on the back side of the silicon wafer 11. The boron-doped silicon layer 12 of the silicon wafer 13, as well as the slower etching (111) planes of the silicon wafer 13, were used as an etch stop to produce the opening 18. The (111) planes the silicon wafer 11 delineated a relatively square shape for the diaphragm 20 (FIG. 1d). Aluminum contacts 19 or electrodes 19 were formed on top of the diaphragm 20 using an electron beam evaporator (FIG. 1e). The aluminum contacts 19 or electrodes 19 were then patterned using lift-off.

[0019] Nickel-iron (Ni—Fe) Deposition Monitoring

[0020] As is shown in FIG. 2, an electroplating bath 31 was set up using a nickel-iron electroplating solution 32. A nickel anode 33 (positive anode) and a sample cathode 34 (negative cathode) to be plated were placed in the electroplating solution 32. A current supply 35, was interconnected between the nickel anode 33 and the sample cathode 34. A loudspeaker 36 was disposed adjacent the electroplating bath 31 and a prototype reduced to practice sensor 10 was disposed opposite the loudspeaker 36 adjacent the electroplating bath 31.

[0021] The variation of acoustic velocity with molarity was first verified experimentally using a commercial piezoelectric sensor. A reduced-to-practice prototype acoustic sensor 10 was then incorporated into a nickel-iron electroplating bath 31 to provide in-situ monitoring of the deposition process (FIG. 2). Theory found in available literature indicates that changes that occur in the solution 32 in the plating bath 31 alter the behavior of a sound wave propagating through the solution 32.

[0022] The loudspeaker 36 was used to launch a sound wave through the bath 31 during deposition runs. The sensor 10 was used to monitor pressure changes and indirectly, the changes in the electroplating bath 31. The output of the sensor 10 and the time of the plating deposition were used to predict the Ni—Fe thickness, which was measured after each deposition run.

[0023] A graph showing the output signal from the sensor 10, recorded as a function of total plated thickness, is shown in FIG. 3, which illustrates that the greater the thickness of the Ni—Fe electrodeposit, the smaller the magnitude of the recorded sensor response. Thus, the pressure in the bath 31 decreases with increasing plated thickness as a consequence of ions being plated out of the solution. FIG. 3 displays a greater positive slope in Region 1 than in Region 2 of the graph. Region 1 includes the time period starting with the onset of deposition and during conditioning of the bath. The explanation for this difference between regions is that initially ions deplete out of the system faster or non-uniformly until some stable condition is met resulting in the smaller slope in Region 2. Statistical analysis was used to map the output of the sensor 10 and the plating time to the Ni—Fe thickness. A regression model was generated that provided a good fit to the experimental data. This model derived an equation of metal thickness as a function of the output of the sensor 10 and plating time. The predictive model may be used to provide real-time monitoring of Ni—Fe electroplating, for example.

[0024] Thus, a micromachined acoustic sensor for monitoring electrochemical deposition and fabrication methods have been disclosed. It is to be understood that the above-described embodiments are merely illustrative of some of the many specific embodiments that represent applications of the principles of the present invention. Clearly, numerous and other arrangements can be readily devised by those skilled in the art without departing from the scope of the invention.