1. Introduction

Charge-coupled device (CCD) image sensors have replaced conventional photographic film and tube-type detectors and become the most popular image sensor. This popularity is attributed to the high sensitivity, long dynamic range, low volume, low weight, real-time display, and advantageous image transfer capability of such sensors. Buried channel CCDs (BCCDs) display higher transfer efficiency than surface channel CCD does; the transfer channels of the former are designed in the interiors of silicon (Si) crystals, whereas those of the latter are located in the Si crystal and silicon oxide interface (Si—SiO₂ interface). The signal charges in BCCD do not interact with the interface states. Moreover, charge transfer inefficiency (CTI) can be neglected given the mature BCCD fabrication technology. Thus, almost all CCD image sensors are BCCDs. At present, BCCD cameras are widely used in mutography, detection, measurement, and monitoring.

Nonetheless, CCD can easily be affected by intense light due to its high sensitivity. Many investigations have indicated that CCDs irradiated by lasers generate various abnormal responses [1–8] that impair CCD image quality. These responses include optical saturation [1,2] that obscures the distribution of light intensity, charge blooming crosstalk [3,4] that distorts spot outlines, and excessive saturation [5] that induces black screens. The laser dazzling phenomenon does not only distort laser signals themselves but also significantly disturbs other important image signals [9–11]. A thorough understanding of such dazzling phenomena can prevent errors in CCD applications and provide new suggestions to improve CCD design.

Another abnormal BCCD response, namely, an entirely saturated unilateral tail, is observed in the current study. This response contradicts the notion that CTI can be neglected in BCCD. On the basis of the analysis of spot tail mechanism, the current study proposes the novel concept that the CTI in BCCD is a jump function of charge quantity. This idea is then validated preliminarily.

2. Experimental set-up and phenomenon

2.1 Experiment set-up

The experimental set-up is shown in Fig. 1. The intense light source used is a 532-nm CW laser. Attenuators and polarizers are employed for coarse and fine laser power adjustments, respectively. Beam splitters and power meters are utilized for the real-time monitoring of laser power. Moreover, splitting ratio is measured beforehand. To eliminate the influence of other light sources, the illuminating lamp in the room is turned off.

 

Fig. 1 Image of the experimental set-up of the 532-nm CW laser irradiating BCCD.

Download Full Size | PDF

The image sensor on the camera is a TCD 1200D linear BCCD, which is a product of TOSHIBA Corporation. The linear CCD camera is placed on a rotating table to scan 2-D scenery. To ensure that the center of the lens is always irradiated by a laser beam, it is positioned above the center of the table. To obtain an image with the correct aspect ratio, the rotational speed $W$ of the table must match the line output frequency $F$ of CCD. The relation between these two factors is given by

2.2 Spot tail phenomenon

The phenomenon of interest in the experiment is depicted in Fig. 2. As laser intensity increases, the tail of the laser spot is increasingly clear. This tail is detected on only one side of the spot and in a linear array direction. Almost all areas of the tail image are saturated. Hence, the phenomenon in which laser spots are distorted is temporarily labelled as the entirely saturated unilateral laser spot tail.

 

Fig. 2 Laser spot tail phenomenon that appears and becomes increasingly obvious with increasing laser intensity.

Download Full Size | PDF

The distance between the center of the original spot and the terminal of the spot tail is considered the tail length. This length is applied to weigh spot distortion degree. Laser intensity is controlled, and variations in CCD integration time are minimized. The tail lengths corresponding to the four-step driving frequencies of CCD are listed in Table 1. Based on these results, tail length remains almost unchanged, as does the doubling variety of drive frequency.

Table 1. Relationship between spot tail length and drive frequency in BCCD

View Table

A spot with a tail was captured with the beam splitter after turning on the illuminating lamp, as shown in Figs. 3(a) and 4(a). A CCD camera was focused on this beam splitter. The profiles of the original spot and of the spot with a tail are displayed in Figs. 3(b) and 3(c), respectively. The tail expands only on one side of the spot. With the exception of the terminal pixel, all pixels of the tail are saturated.

 

Fig. 3 Profiles of the laser spot and of its tail in BCCD. (a) Laser spot with tail. (b) Profile of the original spot. (c) Profile of the spot with a tail.

Download Full Size | PDF

 

Fig. 4 Comparison of spot tails observed by the CCD image sensors in different stances. (a) CCD image sensor in a normal stance. (b) CCD image sensor in an inverse stance.

Download Full Size | PDF

Then, another spot with a tail is captured on the beam splitter when the lens and beam path are maintained and the CCD image sensor is inverted, as indicated in Fig. 4(b). Based on the comparison of the two images in Fig. 4, the spot tail is not inverted when the beam splitter is positioned in line with the inversion of the CCD image sensor. Hence, the spot tail does not distort light distribution on image sensor; rather, this tail distorts the charge distribution in the image sensor.

3. Mechanism analysis and simulations

3.1 Mechanism analysis

The distortion of charge distribution is presumably caused by BCCD charge transfer loss. Based on the series of aforementioned spot tail characteristics, a jump function model of CTI that depicts this loss is given by

 

Fig. 5 Process of laser spot tail formation.

Download Full Size | PDF

Let $Q_{mn}$ be the charge quantity in the mth pixel after n transfers. $Q_{mn}$ can be derived through

CTI can be heightened in BCCD through two potential mechanisms: excessively high driving frequency and the interaction of charges with defective states. In the current study, the former is excluded because tail length does not change with the doubling variety of the driving frequency, as indicated in Table 1. Moreover, the defects in the Si crystal interior are too minimal to be considered due to advanced Si semiconductor technology. Thus, the CTI in BCCD can be reasonably neglected. In fact, a laser spot of minimal intensity is not distorted, as shown in the left-hand side of Fig. 2. The defect states that aggravate CTI in BCCD must then be the Si—SiO₂ interface states. Although the channels for signal charge transfer are designed in the interior of the Si crystal, some of the charges make contact with the Si—SiO₂ interface with an increase in charge quantity. This scenario is known as the surface full well [12] of BCCD. In laser dazzling CCD [1, 2, 9–11], a full well is usually referred to as saturation. Hence, the surface full well of BCCD corresponds to surface saturation in the present study. The signal charges of BCCD pixels in surface saturation can be divided into two parts on the basis of their location. One part is located in the interior of the Si crystal, and the other is situated on the Si—SiO₂ interface. The CTI of the former can be neglected; by contrast, the CTI of the latter is a constant of $\epsilon$. CTI is generally treated as such in CCD devices [12]. Therefore, the CTI of the signal charges in an entire BCCD pixel is simply expressed as

3.2 Simulations

Equation (6) replaces Eq. (2). Furthermore, the model for simulating spot tail profile is programmed as an ‘$sTail$’ function. $[Q_{m0}]$, $\epsilon$, and $Q_{surfS}$ are inputted while $Q_{\text{out}n}$ is outputted. $[Q_{m0}]$ is a sequence of charge quantities in a line of pixels prior to transfer. Information from the TCD1200D CCD datasheet is inadequate with regard to $Q_{surfS}$ and $\epsilon$. Thus, a series of non-dimensional parameters are inputted into $sTail$. A simulation result is outputted that shares the same features as the experimental result, thereby validating the jump function model of CTI. These results are exhibited in Fig. 6.

 

Fig. 6 Result of tail profile simulation and the comparison to the experiment result.

Download Full Size | PDF

4. Summary

A new phenomenon involving a laser spot with an entirely saturated unilateral tail in BCCD was observed in this study. The conceptualization of the CTI in BCCD as a jump function of charge quantity was proposed as well. When charge quantity increases, some charges make contact with the Si—SiO₂ interface. Moreover, some free charges interacted with the interface states and were left behind during transfer. Hence, CTI was mutated in BCCD. A simple expression was established for CTI with varying charge quantities; this expression was used to develop a model to simulate the profile of a laser spot with a tail. The simulation displayed the same features as the experimental result, thus validating the CTI jump function in BCCD.

The laser spot tail indicates a deterioration in CTI with an increase in charge quantity. CTI is the main index for measuring the CCD transfer function; therefore, this model may help improve the transfer function of BCCD as well. Our future work aims to investigate the relationship between the CTI function and CCD parameters. The findings from this study may serve as a specific reference for improving CCD design.