Advertiser and

Buyer's Guide
Buyers Guide

Chip Shots blog

Greatest Hits of 2005
Greatest Hits of 2005

Featured Series
Featured Series

Web Sightings

Media Kit

Comments? Suggestions? Send us your feedback.

Using a real-time, point-of-use sensor to control liquid-chemical concentration

Ronald Chiarello, C. Eric Boyd, and Christopher Wacinski, Jetalon Solutions; and Thomas Kiez, Jerry Elkind, Bryan Presley, Roger McDermott, and George Harakas, Texas Instruments

Controlling the concentration of liquid chemicals at the process tool has become increasingly important in semiconductor device manufacturing as a result of several developments. First, the increased use of technology-enabling processes such as chemical-mechanical polishing (CMP) and electrochemical deposition (ECD) has led to significant increases in the consumption and cost of liquid chemicals. Second, as liquid-chemical process windows narrow (especially for advanced technology nodes), small variations in chemical concentration can affect wafer-surface defectivity levels and, therefore, yields. Etch, copper CMP, and ECD processes are examples of the types of processes that depend on liquid chemicals and have narrowing process windows. Third, the environmental impact of liquid chemicals and regulatory measures affect how IC manufacturers must treat and dispose of effluent streams bearing liquid-chemical waste.1–6 These developments have increased the overall cost of ownership (CoO), especially for fabs that manufacture devices at or below the 90-nm technology node.

The case studies discussed in this article investigate a point-of-use (POU) liquid-chemical-concentration analysis system from Jetalon Solutions (Campbell, CA) and were performed at two high-volume Texas Instruments fabs. One case focuses on etch processing in the surface-preparation area, while the other focuses on back-end processing.

Point-of-Use Sensors

Point of use is defined as any location in the fab where chemicals are blended or delivered, including inside process tools and chemical blending and delivery systems. Continuous, real-time liquid-chemical-concentration measurements at the point of use provide many benefits to process and equipment engineers. For example, they can be correlated to metrology data on wafer-surface defects and yield, leading to early identification of yield failures. In addition, they promote the development and automatic maintenance of best-known methods for wafer processing. They enable tools to become more process-flexible as chemical concentrations are adjusted, controlled, and monitored at the point of use. Tool failure modes can be identified and addressed, resulting in increased tool uptime and productivity.

In summary, POU sensors incorporated into monitor and control systems will help process and equipment engineers to accelerate process development, identify and eliminate tool failure modes (thereby increasing tool productivity), reduce chemical consumption and wafer scrap (leading to significant annual cost savings), and minimize the environmental impact of semiconductor manufacturing. Taken together, these benefits will improve the overall cost of ownership.

In general, process tools lack POU liquid-concentration sensors. Consequently, engineers rely on grab samples and laboratory analyses for information on liquid-chemical concentration, a method that is neither done in real time nor continuously. Although it is accurate, it does not provide the benefits of POU sensors.

To be of most benefit, liquid-chemical-concentration sensors must be designed specifically for semiconductor applications. They must provide real-time continuous data output with ≤1-second response time, they must have a small footprint, they should have no moving parts, they should be process transparent, and they should offer a reasonable return on investment. In addition, liquid-chemical-concentration sensors must be reliable, exhibiting high resolution and appropriate dynamic range. Finally, they must be compatible with acids, bases, solvents, oxidizers, and slurries.

Although the sensor in this study was used to gather data from surface-preparation processes, it has been developed for all liquid-chemical process areas. It can be used for blending CMP slurry at the point of use; monitoring ECD plating baths, solvents in ultrapure water, chemical drain lines, or process chemical baths; and performing a range of chemical spiking applications in batch-wafer processes.

Liquid-Chemical-Concentration Analysis System

To demonstrate the benefits of a POU sensor system, fab tests were performed using a CR-288 liquid-chemical-concentration analysis system (CAS) from Jetalon. The CAS consists of a fully integrated, miniaturized optical reflectivity device, an optical fluidic cell, and a DSP-based electronic circuit. It uses optical reflectivity to measure refractive index and, therefore, concentrations of liquids. The optical reflectivity device includes an LED light source, a mirror, a sapphire window, and a photodiode-array detector, which measures scattered-light intensity simultaneously over all relevant angles.

The device is packaged in a compact optical fluidic cell that measures 5 X 5 X 7.5 cm. Only the sapphire window and Teflon components make contact with the liquid chemical under analysis. The sensor measures temperature using a thermocouple placed inside the optical fluidic cell. Capable of achieving data output rates of 100 points per second, it has a temperature range of 10°–70°C, sensitivity of 0.01% of concentration, a response time of 10 milliseconds (1 second in practice), and a pressure range of 0–50 psi.

In the reflectivity geometry, light is incident on an optically transparent sapphire window (index of refraction n1), which is in contact with a liquid under analysis (index of refraction n2). For reflectance angles (θs) smaller than the critical angle (θc), light is predominantly reflected off the window (total external reflection). For reflectance angles greater than θc, light is both reflected off the window and transmitted through it. By measuring the change in scattered-light intensity as a function of θs, θc can be determined. From Snell's Law, θc is related to a liquid's index of refraction. It follows that by accurately determining θc, liquid concentration can be determined accurately.

Figure 1: Index-of-refraction measurements from the liquid-chemical-concentration sensor plotted as a function of measurement time for different H2O2 mixtures.

CAS performance results, obtained under controlled laboratory conditions, are shown in Figures 1 and 2. To conduct the tests, the optical fluidic cell was inserted into a chemical delivery system that can be operated in either closed-loop or open-loop control flow modes. The delivery system included a piston pump, temperature control bath, and an autotitrator. Initial laboratory measurements for testing and calibration using hydrogen peroxide (H2O2) are presented in Figure 1, which shows index of refraction plotted as a function of time. Measurements were taken for H2O2 concentrations of 1.0, 1.1, 1.13, and 1.141%. H2O2 concentrations were spiked into the delivery system, which operated in closed-loop mode in doses of 0.1, 0.03, and 0.011%, and were verified using an autotitrator. For convenience, data were acquired every 5 seconds over a 9-hour period. Figure 2 shows similar results obtained from a comparable laboratory setup for dilutions of SC-1 from 1:1:20 to 1:1:50 and ultrapure water (UPW).

Figure 2: Index-of-refraction measurements from the sensor plotted as a function of measurement time for different SC-1 dilutions and UPW.

Results and Discussion

The case studies presented here focus on a concentration analysis of SC-1 in immersion wet benches. In both studies, the optical fluidic cell was connected directly with the SC-1 bath recirculation system using a 1/4-in. connection. Figure 3 shows the sensor head installed at the point of use (the recirculation line of the SC-1 bath) inside a 200-mm wet tool from Dainippon Screen (Kyoto, Japan). The tool operated in full production mode during the case studies. In the first case study, which involved an etch process and open-loop H2O2 spiking, the initial SC-1 bath concentrations were 1:1:5. In the second case study, which involved a back-end process with no chemical spiking, 0.75% H2O2, 0.75% ammonia hydroxide, and 98.5% water were used. In both studies, concentration data were collected continuously for more than a month. The data presented here represent a small fraction of the total data that were collected and analyzed.

Figure 3: The sensor head (circled) installed at the point of use (in the air-bleed line of the filter) inside the SC-1 bath recirculation line of a 200-mm wet tool.

Case Study 1: Etch Process. Figures 4–7 present concentration measurements for the first case study. Figure 4 shows SC-1 concentration and process bath temperature plotted as a function of time during a 24-hour wafer production run. It also shows three bath life cycles. Figure 5 shows the same data as Figure 4 over a 5-hour period, illustrating concentration behavior between two SC-1 bath dump/fill cycles. In both tests, concentration and temperature data were collected simultaneously every second. Within each bath life cycle, 40 ml of H2O2 were spiked into the 27-L bath each time a batch of wafers was introduced into the tank.

Figure 4: SC-1 concentration and bath temperature plotted as a function of time during a 24-hour wafer production period.

The concentration data reveal several key bath characteristics, including the effectiveness of H2O2 spikes to maintain chemical concentration; SC-1 concentration changes as a function of time and wafers processed; bath change-out properties and conditions; variations in SC-1 concentration from bath to bath; and the effectiveness of the bath's temperature-control system and various tool failure modes related to pumps, valves, etc. As illustrated in Figure 5, the concentration data between 5 and 10 hours revealed several things about the etch process. First, immediately following each H2O2 spike, the SC-1 concentration increased beyond the initial bath concentration. Second, the overall concentration of the bath increased from 9.09% to 9.24%. Third, the sensor detected the H2O2 spikes.

Figure 5: The same data as in Figure 4 plotted for one bath lifetime, which was approximately 4 hours.

Figure 6 highlights an H2O2 spike. Initially, SC-1 concentration increased to approximately 9.38% during injection and then receded to 9.16% after mixing was completed. Overall, the difference in SC-1 concentration before and after the spike was 0.03%. As shown in Figures 4 and 5, the cumulative effect of the H2O2 spikes was to increase the SC-1 bath concentration, indicating that the spike concentrations were too high and that spiking frequency was too frequent.

Figure 6: Close-up of a 40-ml H2O2 spike in the 27-L SC-1 process bath.

Figure 7 presents a close-up view of SC-1 process bath concentration and temperature during a bath change plotted as a function of time. The steep drop in concentration occurred when the sensor was exposed to air during the SC-1 bath drain (dump). The data also show a significant variation in SC-1 concentration before and after the bath change. While the concentration before the bath change was 9.15%, it was 9.06% after the bath change, a 0.09% drop.

Figure 7: SC-1 concentration plotted as a function of time for an SC-1 bath change.

Case Study 2: Back-End Process. Figures 8 and 9 illustrate SC-1 concentration measurements for the second case study. Figure 8a shows SC-1 concentration plotted as a function of time, focusing on a bath change during which the diaphragm pump used to circulate the SC-1 bath unexpectedly turned on and off several times. The pump shutoff contributed to the accumulation of static chemical residues in the recirculation line and extended refill times. After the pump and refill-delay issues were corrected, the process returned to normal, as indicated in Figure 8b.

Figure 9: SC-1 concentration and bath temperature plotted as a function of time for the second case study. The data were collected continuously for 45 hours.

Figure 9 shows concentration data plotted as a function of time during 45 hours of continuous operation. The data reveal that there was a smooth transition between bath changes and initial SC-1 concentration values, which varied by less than 0.05% from bath change to bath change. However, the data also show a decrease of 0.7% in SC-1 bath concentration during wafer processing. This relatively large and unexpected change in concentration led to an investigation of the SC-1 bath, which determined that the bath was automatically being filled with UPW during wafer processing because of a leak.

In ongoing work, the sensor has been incorporated into a closed-loop-control spiking system to maintain SC-1 concentration and extend SC-1 bath lifetime. This change will lead to reduced chemical consumption, greater cost savings, and increased tool productivity.


This article has presented two case studies demonstrating the performance of a compact real-time liquid-chemical-concentration analysis system. In both studies, SC-1 used in immersion baths was analyzed at the point of use and in real time during wafer production. Concentration data were used to diagnose the efficiency of H2O2 spiking, determine appropriate bath lifetimes, and identify and eliminate tool failure modes.

Future studies will investigate the use of the sensor in chemical spiking, delivery, control, and monitoring systems, which are expected to have a significant impact on cost of ownership because they will reduce chemical consumption; identify and eliminate tool failure modes; maintain best-known methods for chemical delivery, spiking, and blending; and increase tool uptime and productivity.

Real-time POU chemical-concentration data can be correlated to wafer-surface metrology data and catastrophic wafer-failure events, providing opportunities for determining the effects of chemical concentration on wafer-surface defects and yield. That knowledge is especially important in applications with narrow process windows, such as hydrofluoric acid etch, CMP, photolithography, and ECD.


This article is an edited, revised version of a presentation from the Semiconductor Pure Water and Chemicals Conference (SPWCC), held February 14–16, 2005, in Santa Clara, CA. Copyrights reserved by SPWCC. Reprinted with permission of SPWCC.


1. R Chiarello, R Parker, and M Tritapoe, "Optimizing Wafer Rinsing Processes to Conserve DI Water," MICRO 18, no. 6 (2000): 111–119.

2. R Chiarello et al., "Multidisciplinary Approaches Target ES&H," Semiconductor International 22, no. 2 (1999): 62–66.

3. R Chiarello, "ESH Issues Make Progress," Semiconductor International 24, no. 3 (2001): 81–88.

4. The International Technology Roadmap for Semiconductors (San Jose: Semiconductor Industry Association, 2004); available from Internet:

5. AR Sethuraman, "CMP—Past, Present and Future," Future Fab International 5 (July 1999): 261–264.

6. B Morrison, S Joshi, and R Tolles, "Copper and STI CMP Technology: The Challenges and the Cost," Future Fab International 11 (June 2001): 269–273.

Ronald Chiarello, PhD, is president of Jetalon Solutions (Campbell, CA). With more than 15 years of experience in high-tech R&D at Stanford University and the University of Chicago, he has consulted for the world's largest high-tech manufacturers in the semiconductor, data-storage hardware, flat-panel display, and other industries. Chiarello has published more than 50 articles in the areas of semiconductor device processing, solid-state physics, biophysics, surface science, electrochemistry, and geochemistry. A NATO fellow, he has won the Depart- ment of Energy Award for Excellence in Research and the University of Chicago Pace Setter Award for Outstanding Contributions in Synchrotron X-ray Techniques. He received a BS in physics from the University of California at Santa Barbara and an MS and PhD in physics from Northeastern University in Boston. (Chiarello can be reached at 408/866-6318 or

C. Eric Boyd is solutions engineer at Jetalon. Before joining the company, he participated in other high-tech start-ups. He received a degree in applied science (engineering) from Queens University (Kingston, ON, Canada), graduating with first-class honors. His education included extensive mathematical training and numerous hands-on application courses in mechanical and electrical design. (Boyd can be reached at 408/866-6318 or

Christopher Wacinski is a lead engineer at Jetalon. His work as a control systems specialist in the semiconductor industry has focused primarily on the area of liquid flow and monitoring. As a certified LabVIEW developer, he has consulted on projects ranging from real-time wet-chemical dispensing to gas flow control. Wacinski spent three years as a lead engineer in the biomedical field working for Navigant Biotechnologies, where he developed spectrometer-based irradiance measurement and calibration instruments. He received a BS in electrical engineering technology from the Metropolitan State University of Denver, CO. (Wacinski can be reached at 408/866-6319 or

Thomas Kiez is an equipment engineering technician in the projects group at Texas Instruments' DMOS5 wafer fab in Dallas, where he has worked since 1995. The group supports all equipment engineering groups, including photolithography, plasma, thin films, diffusion, wet etch, implant, and CMP. Previously, Kiez worked for Pratt and Whitney Canada as a test cell technician and served in the Canadian Forces for 10 years as an aircraft technician. He has been a member of the Alberta Society of Engineering Technologists for 10 years. He received an associates degree in semiconductor manufacturing technology from Texas State Technical College in Waco and an associates degree in IC design layout from Eastfield College in Mesquite, TX. (Kiez can be reached at 972/927-7479 or

Jerry Elkind, PhD, is a senior member of the technical staff at Texas Instruments in Dallas, where he manages the company's analytical sensors branch. With almost 17 years of experience at TI, he has worked in both the central research laboratory and the Houston wafer fab. He received BA and MA degrees in chemistry from Brandeis University in Waltham, MA, and a PhD in physical chemistry from the University of California at Berkeley. (Elkind can be reached at 972/995-1214 or

Bryan Presley is an equipment engineer in the assembly group at Texas Instruments' DMAT fab in Dallas, where he has worked since 1999. The group supports back-end equipment engineering projects and process support, including plasma ash, wet cleans, passivation, die attach, bonding, and welding. With 14 years of experience in the semiconductor industry, Presley worked for Ball Semiconductor as a project leader in the development of single-crystal silicon spheres and at a semiconductor equipment manufacturer as an equipment design engineer and process engineer. He received a BS in industrial engineering from the University of Texas in Arlington and an associates degree in science at Richland College in Dallas. (Presley can be reached at 972/927-3060 or

Roger McDermott has been a senior equipment specialist at Texas Instruments for 19 years. For 14 of those years, he has been involved in equipment engineering at the company's DMOS4 and DMOS5 fabs. His areas of specialty include diffusion and wet processes. He received an AAS in electrical engineering technology from ITT Technology Institute. (McDermott can be reached at 972/995-1518 or rmcdermott

George Harakas, PhD, is a member of the diffusion/wet team at Texas Instruments' DMOS5 fab. His primary responsibilities include metal silicide and metal nitride wet etch processes. Before joining the group, he spent three years in the photolithography group at TI. He received a PhD in inorganic chemistry from Texas Tech University in Lubbock. (Harakas can be reached at 972/995-1241

MicroHome | Search | Current Issue | MicroArchives
Buyers Guide | Media Kit

Questions/comments about MICRO Magazine? E-mail us at

© 2007 Tom Cheyney
All rights reserved.