A CMOS Magnetic Sensor Chip for Biomedical Assays
*** THIS PROJECT IS NO LONGER ACTIVE ***
Paul Liu and Bernhard Boser
Magnetic beads are widely deployed as labels for target analyte in biomedical applications. Nevertheless in the past to detect the micron-sized beads people have to use expensive magnetic sensors such as Superconducting Quantum Interference Device (SQUID) or Giant Magnetoresistance (GMR) as well as external coils and peripheral electronics. In this project we aims to develop a low-cost magnetic bead detection solution on a single CMOS chip. We face several challenges. First, the beads are paramagnetic which means they are magnetic only in the presence of externally applied field. This polarization field needs to be generated on chip. Second, the beads have small sizes of 1 to 5μm and are several microns away from the underlying magnetic sensor. As a result, the induced magnetic field from the beads is as small as 10μ Tesla when 1m Tesla polarization field is applied. The induced field is even smaller than earth magnetic field (~50μ Tesla). So rejection of polarization field and stray field is a must. Finally, the intrinsic error in CMOS sensors such as offset and 1/f noise is larger than the signal (~μV) and therefore makes the detection more difficult. We implement the magnetic sensor chip in standard 0.18um CMOS technology. The chip consists of an array of 128x128 CMOS Hall sensors. Each Hall sensor has a self-contained coil to generate polarization field. The size of Hall sensor is designed to be in the same range as the magnetic bead. We have demonstrated two bead detection methods. One is polarization field rejection relying on Hall sensor matching; the other is magnetic relaxation, which is based on the Neel relaxation nature of the superparamagnetic nanoparticles in the bead. A total of 8 readout channels are implemented to read the Hall voltage signals from the array in parallel. Each readout channel contains a signal conditioning circuit and an 8-b ADC that amplifies and digitizes the signals. We apply techniques such as Chopper Stabilization and Correlated Double Sampling to reject offset and flicker noise. A 14-bit counter counts the total number of beads and generates a digital output.