^{1}

^{*}

^{2}

^{1}

^{1}

This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/).

The partial charge transfer technique can expand the dynamic range of a CMOS image sensor by synthesizing two types of signal, namely the long and short accumulation time signals. However the short accumulation time signal obtained from partial transfer operation suffers of non-linearity with respect to the incident light. In this paper, an analysis of the non-linearity in partial charge transfer technique has been carried, and the relationship between dynamic range and the non-linearity is studied. The results show that the non-linearity is caused by two factors, namely the current diffusion, which has an exponential relation with the potential barrier, and the initial condition of photodiodes in which it shows that the error in the high illumination region increases as the ratio of the long to the short accumulation time raises. Moreover, the increment of the saturation level of photodiodes also increases the error in the high illumination region.

Due to their ability to automatically produce clear images of an object plane that has extremely varying illumination levels, wide dynamic range image sensors are required for many applications such as cameras for security systems, automobiles and industry. There have been various approaches to enhance the dynamic range of CMOS image sensors [

The simplified layout of the pixel and its cross-section at line aa' are shown in ^{+} layer and still contributes to the dark current. Therefore, by using a small

The use of octagonal shape for PD can increase the speed of charge flow during charge transfer operation, preventing image lag [

The principle of the dynamic range expansion by the partial charge transfer technique is discussed in this section. In this sensor, one frame of accumulation time is divided into two sub-frames.

The next steps explain the operation of the wide dynamic range image sensor with partial transfer:

When a strong light is irradiated on the pixel, the accumulated charge in the PD reaches the saturation level (_{max}

In the final sub-frame, the accumulated charge is partially drained and charge accumulation is repeated.

Finally, a whole charge transfer operation is done and the transferred signal to the floating diffusion (FD) is read out.

From the operation, two set of output signals is obtained from a single photodiode, the long and short accumulation time signals. A Wide dynamic range image can be synthesized from the two setS of acquired signals because the difference of charge accumulation time can sufficiently expand the dynamic range of the sensor.

The signal from wholly transferred charge in the final sub-frame determines which signals have to be used, whether the long accumulation or the partially transferred short accumulation time signals. A method to judge which signals to be used is proposed. If the quantity of accumulated charge reaches a threshold value, _{T}_{T}_{T}

In the case of a weak light irradiated on the pixel, the same operation (1)∼(3) is performed. However, the read data at the end of first sub-frame is 0 because the accumulated charge in photodiode does not exceed threshold level, _{T}_{T}_{T}

The charge transfer mechanism plays an important role in this sensor. Hence, two type of charge transfer namely the whole charge transfer and partial charge transfer mechanism are described in this section.

The whole charge transfer is the same as a normal charge transfer in conventional 4T APS CMOS image sensors [

The saturated accumulated signals charge can cause smear and blooming [

The partial charge transfer mechanism is described in

In

To assure partial charge transfer works properly, an appropriate transfer gate voltage must be applied to increase the potential barrier under the transfer gate to be higher than the potential of the photodiode until part of the accumulated charge is transferred. After a short time, the charge transfers stop because the increment in photodiode potential as the accumulated charge is transferred to the FD. As a result, only a part of accumulated charge is transferred. This mechanism is the same in the case of partial charge transfer for draining or signal reading out purposes. The frequency of the partial charge transfer operation for read out purpose as illustrated in

A simulation to check the characteristics of partially transferred electrons, _{T}_{TX}_{T}_{T}

In the equation, _{n}, n_{p}, x, A_{n}_{p}/δ_{x} in

Calculating

Note that _{S}

By substituting the

In the equations:

Next, the equivalent circuit shown in

Differentiation of

By substituting the

Calculate the

From _{bi}-_{B} = 0),

Using the parameters in

Within the readout time, the transferred charge is given by:

If 1 ≫ t_{R}/τ:

If t_{R} = 0.5 [μs], the number of transferred electrons are:

If one electron is transferred, then, from

At this time, J[0] is renamed as _{1}:

From

Substituting

Therefore, from

From the above calculations, it is clear that the charge is transferred continuously until the _{B}_{B}

The initial condition of the photodiode can influence the partially transferred charge for the short accumulation time signal. The Initial condition is indicated by the number of initially accumulated electrons in PD, _{I1}_{R1}_{T}, N_{RES}_{a}_{R2}_{a}_{R1}, N_{a}_{R2}

The relationship between _{R1}, N_{a}_{R2}

An analysis of the non-linearity due to the partial charge transfer has been conceded. A simulation is carried to check the relations of re-accumulated electrons and partially transferred electrons with respect to the initial conditions of the photodiode. In the simulation, the transfer gate drive voltage and charge transfer time is set to 0.5 V and 1.0 μs, respectively. The simulation results are shown in _{R2}_{a}_{I1}_{R2}_{a}

The simulated relationship between _{R2}_{a}_{0}, I_{T}, I_{LM}, I_{M}_{M}_{L}_{S}_{a}_{max}_{M}

From _{a}

For example, if _{max}_{L}_{S}_{a}_{R2}_{a}_{L}_{S}_{max}

_{L}_{S}_{max}_{L}_{S}_{L}_{S}_{a}_{a}_{a}

_{max}_{L}_{S}_{max}_{max}_{a}_{a}_{max}_{max}_{DE}

Therefore, the dynamic range in this technique can be expanded either by using a photodiode with higher _{max}_{L}_{S}_{max}_{L}_{S}_{max}_{L}_{S}

The partial charge transfer technique is a countermeasure to improve dynamic range of CMOS image sensors and at the same time maintains a high fill factor because only one photodiode is integrated in each pixel. The dynamic range expansion in this sensor is controlled by partial charge transfer and if a very wide dynamic range is required, it can be achieved by taking a large accumulation ratio of the long to the short accumulation time signals. However, the technique suffers from non-linearity in the output of the synthesized wide dynamic range signals especially if a large accumulation ratio is taken. An analysis of the non-linearity utilizing this technique has been done and discussed. The calculation and simulation results show that non-linearity can be caused by two factors that are current diffusion from the potential well and initial conditions of photodiode. From the calculations, it is shown that the diffusion current has an exponential relationship with the potential barrier suggesting a non-linear relationship between the transferred charge and potential barrier under the transfer gate. The simulation results show that the error in the high illumination region is increases as the ratio of the long to the short accumulation time is increases. Furthermore, increasing the saturation level of photodiodes also increases the error in the high illumination region.

The authors would like to thank the members of imaging device laboratory, Shizuoka University, for their effort in the calculation, simulation and design progression.

The principle of Whole Charge Transfer,

The principle of Partial Charge Transfer.

Partially transferred electrons vs accumulated electrons in PD.

The pixels cross section and potential profile of a photodiode.

The equivalent circuit.

Initial condition influents the read out signals in partial transfer technique.

Partial transferred electrons for read out, _{R2}_{a}

Charge accumulation in one frame.

Photo-electric conversion characteristics of the synthesized wide dynamic range signals.

Non-linearity in high illumination region (error in %).

Photo-electric conversion characteristics of the synthesized wide dynamic range signals.

Non-linearity in high illumination region (error in %).

Pixel characteristics.

Technology | 0.18 μm CIS 1P4M |

Pixel size | 7.5 μm × 7.5 μm |

No. of photodiode | 1 |

Photodiode shape | Octagonal |

Fill factor | 14% |

Device parameters.

1.5 [μm] | |

0.7 [μm] | |

_{n} |
700 [cm^{2}/V·S] |

0.1 [μm] | |

_{D} |
2 × 10^{17} [cm^{−3}] |

_{A} |
10^{18} [cm^{-3}] |

_{i} |
1.45 × 10^{10} [cm^{−3}] |