Photonic charge-pump device measures distance

A variety of imaging applications could benefit from the ability to discern how far an object is from the sensor. Although CMOS image sensors have gained popularity due to their low cost of fabrication and ease of integration with other electronics, 3D optical measurement remains a challenge for these devices. We are developing a compact CMOS-compatible device that uses the phenomenon of charge pumping to extract distance information from sensors.

Two methods are available for extracting depth information from a scene: one can use stereo vision, which requires processing a large amount of data quickly; or one can use a modulated light source synchronized with the sensor (see Figure 1 ). In the latter case, a transmitter generates a burst of light pulses, illuminating the scene. The back-scattered light is focused onto an imager, which measures the received phase delay with respect to the source, caused by the light travel path. The distance is proportional to the phase shift divided by the modulation frequency.

In existing devices, demodulation is performed by an electro-optical charge mixer, known as photonic mixer device (PMD).¹ Based on CCD-like charge-transfer mechanisms, this device performs mixing and sensing simultaneously. The main drawback of a PMD is that it cannot be made using standard CMOS techniques.

Our experimental device, the photonic charge-pumping (CP) mixer, is a CMOS-compatible alternative to the PMD. It consists of two positive-channel metal-oxide-semiconductor (PMOS) transistors working in counter-phase clocking, and is based on the CP phenomenon: this was discovered in 1969 and extensively adopted as a powerful analysis method for evaluating and quantifying the degradation of MOS transistors.² A CP device is a three-terminal MOS transistor with a clocked gate and the source and drain shorted together. The bias conditions force the transistor to switch between strong inversion and accumulation conditions at each gate change.

CP occurs during the transition from inversion to accumulation. In this case, some of the minority charge carriers from the channel get caught in the fast interface traps placed under the gate oxide. These minority charge carriers recombine with majority ones that are attracted from the substrate. This turns into a net charge packet, which flows to the source terminal at each clock cycle. Note that because CP is a purely electro-physical phenomenon, the generated charges are proportional to both the density of traps and the device's operating voltage.³

The basic idea of our photonic CP device is to force the transistor's bias conditions to depend on the light intensity. The PMOS transistor provides an elegant way to achieve this: we use the n-well/p-sub junction in the transistor as a photodiode embedded into the device itself. ⁴ By setting the well floating—after a precharging phase—a linear discharging process occurs when light impinges on this part of the device, which causes the desired change in the bias conditions. At the end of the optical integration time, the PMOS switches from inversion to accumulation. The PMOS also creates a pump charge packet Q_p(V_well) with characterics that depend on the light intensity. This charge is then integrated or accumulated onto an integrator placed behind the device.

The CP mixer device shown in Figure 2 consists of two integrators connected to two PMOS (a left and right channel), which share an n-well and a reset transistor. The two devices work independently, accumulating charge packets into the integrators at different time slots. When synchronized with the modulated light source, they integrate only a fraction of the received signal because of the delay introduced by the distance. Consequently, the difference of the two outputs (V_oR-V_oL) is a measure of the light phase shifting, and thus of the distance off the backscattering object. Figure 3 shows results of a preliminary experiment testing this device.

CP mixer devices represent an attractive alternative to conventional PMDs, in particular because they are compact and fully compatible with CMOS fabrication techniques. Our experimental results demonstrating the demodulation capability of the device are encouraging. We are continuing our work to prepare a setup for real distance measurements as well as designing a new circuit to compensate for background and offset.