Ultraviolet (UV) position-sensitive photodetectors (PSD) have become an essential building block for information technologies, including laser monitoring^[1], wireless communication^[2], flame sensing^[3], and lithography^[4]. Traditionally, CMOS imaging sensors^[5], charged coupled devices (CCDs)^[6], and photodetector arrays^[7] hold the advantages of high resolution when used as pixel-type PSDs. However, they usually cannot provide continuous information recording and suffer from slow response due to the massive signal extraction process from the large pixel array. Although lateral-effect PSDs^[8] are handling continuous photogenerated signals with no internal discontinuities, they have a relatively low position resolution and a slow response. In comparison, quadrant photodetectors (QPDs) can realize position analysis with high resolution, fast response, high sensitivity, and low cost^[9,10].

Currently, Si-based QPDs are the mainstream choice for UV-band spot positioning applications. However, they suffer from inherent sensitivity to visible light and can easily degrade under harsh operation environments. Among various semiconductor materials^[11–14] explored to fabricate UV QPD, 4 H-SiC draws particular interest for its wide bandgap, visible-blindness, relatively mature processing technology, and potential to operate in harsh environments^[15,16]. Rafi et al. fabricated segmented 4 H-SiC quadrant diodes and demonstrated good performance at variable temperatures^[17]. Romijn et al. demonstrated a collimating quadrant sun position sensor microsystem based on 4 H-SiC CMOS technology^[18]. So far, the study on SiC-based QPD is still in its early stage. Systematic characterizations and positioning applications of SiC QPD have rarely been reported in the literature.

In this Letter, we demonstrate UV spot position measurement based on a 4 H-SiC QPD. The fabricated QPD shows low dark current, high UV responsivity, and excellent uniformity. Subsequently, the device is adopted into the UV spot position measurement system, where key performance parameters, including positioning error and resolution, are evaluated. As has been unveiled, the inherent nonlinearity of the QPD would reduce the accuracy of position measurement, especially for a wide measuring range. With a detailed study of the origins of the positioning inaccuracy, we develop a calibration method utilizing the Boltzmann function, which has achieved significant improvement in the positioning accuracy of the system.

The schematic diagram of the 4 H-SiC QPD fabricated in this work is shown in Fig. 1(a). The epi-structure consists of a 4-µm n−-type layer (n∼3×1014cm−3) and a 1-µm n+-type layer (n∼1×1019cm−3) from top to bottom, which is grown on a 4-inch n+-type 4 H-SiC substrate^[19]. At the beginning of the process, the device is passivated with a SiO2 layer on the 4 H-SiC surface. Next, the p+-type doping region is formed using the Al ion implantation process, where the processing parameters are similar to those reported in fabricating the 4 H-SiC p-i-n UV photodiode^[20]. The ohmic contact on the surface p+-type implanted region and the n+-type substrate is formed by depositing a Ni/Ti/Al/Au metal stack. Finally, Ni/Au metal bilayers is deposited and patterned as contact pads to facilitate the subsequent packaging and characterization. The fabricated QPD has a photosensitive area of 8mm×8mm, which is divided into four quadrants (4mm×4mm) by a cross gap with 0.1 mm in width. For system applications, the QPD is hermetically sealed in a TO-8 package by wire bonding. Figure 1(b) shows a photograph of the packaged QPD.

Figure 1(c) shows the room temperature dark current-voltage (I−V) characteristics of the QPD with typical rectification behavior. The dark current of each quadrant remains consistent with a similar value, as low as ∼18pA at a reverse bias of 20 V, which ensures a high signal-to-noise ratio (SNR) in position measurement. Figure 1(d) shows the spectral response of the QPD under zero bias. A high responsivity of 0.111 A/W at 275 nm is obtained, corresponding to an external quantum efficiency (EQE) of 49.95%. The UV/visible (275/400 nm) rejection ratio is over three orders of magnitude, validating the preferable visible blind performance of the 4 H-SiC detector. Additionally, both the dark current and the responsivity of the four quadrants exhibit minor variation. This uniformity is especially important for the position-sensitive application, where near-ideal symmetry promises simple signal processing with low positioning error.

The schematic of the UV spot position measurement system is shown in Fig. 2. A 275 nm UV LED is used as the light source, where the light is then collimated and focused on the active region of the QPD through two convex lenses. An aperture is located closely in front of the QPD to regulate the spot size. The incident light power density is about 1mW/cm2 on the surface of the photovoltaic-mode QPD, which is sufficient to maintain a high SNR. The photocurrent generated in each quadrant is then fed into four independent signal processing channels, where each channel consists of a trans-impedance amplifier with a gain of 106V/A and a buffer amplifier. The output voltages of the processing circuit are converted into digital signals by an analog-to-digital converter (ADC), which will be subsequently transmitted to the computer for data processing. The QPD and the related processing circuits are mounted on a two-axis (horizontal and vertical directions) displacement motion platform.

The zero point of coordinates is defined as the geometric center of the QPD. The horizontal and vertical directions are defined as the x- and y-directions, respectively. Based on the output voltages of all four channels, the relative position of the light spot with respect to the zero point can be determined by {x=k·σx=k·(V1+V4)−(V2+V3)V1+V2+V3+V4y=k·σy=k·(V1+V2)−(V3+V4)V1+V2+V3+V4,(1)where σx and σy are the output signals in the x- and y-directions, respectively. k is the linear fitting parameter, which is commonly related to the size and shape of the spot. V1, V2, V3, and V4 are the output voltages of each quadrant^[21].

The position detection performance of the QPD is investigated by changing the relative position of the detector to the light spot using the two-axis displacement motion platform. At first, a one-dimensional measurement is conducted to evaluate the linearity between the output signal and the actual position of the light spot. The motion platform is set to move along the x-direction from (−1000μm, 0) to (1000 µm, 0) with a step of 100 µm. Figure 3 shows the relation of the output signal σx and the spot position x under different spot radii r. Excellent linearity could be observed in the dependence of σx on the x position within a displacement range, while the nonlinearity will then dominate as the displacement increases, especially for small spot sizes. In other words, nonlinearity increases as the value of x/r increases. Similar nonlinearity characteristics can be obtained in the y-direction, which is not shown here.

Considering that the linear region is closely related to the spot radius, a spot with a 1 mm radius is chosen to evaluate the two-dimensional positioning accuracy across a larger measurement area. The spot position is changed line by line from (−800μm, −800μm) to (800 µm, 800 µm) with a step size of 100 µm in both directions, which covers about 90% of the effective measurement area under this spot size. The corresponding output position values in the UV spot position measurement system are determined by Eq. (1). Figure 4(a) shows the position pattern of the measured points. It can be seen that the points near the center exhibit a regular grid pattern. However, the density of points increases as the spot leaves the center, suggesting an increase in nonlinearity. Moreover, the positioning error is calculated and its spatial distribution is shown in Fig. 4(b). The mean positioning error is ∼126μm, and the error increases sharply as the spot moves away from the center, indicating a decrease in measurement accuracy due to the nonlinearity. This behavior is similar to the results above for the one-dimensional testing, where the physics will be further discussed in the following section.

Position resolution is also a key figure of merit for position measurement, and it is defined as the minimum detectable displacement of a light spot^[22]. In this experiment, the spot is located at the zero point of the QPD, and the motion platform is kept stationary. Then, a total of 4500 detected positions are recorded. As can be seen from Fig. 5(a), the measured positions are basically distributed around the zero point with a small amplitude disturbance. Figures 5(b) and 5(c) exhibit the histograms of measured positions along the x- and y- directions, respectively, where the position points follow a normal distribution. 3Σ (standard deviation) is defined as the position resolution of the system in this work, which dictates that the system can accurately resolve spot displacements beyond 3Σ. The calculated position standard deviation Σ is 0.1 µm, corresponding to a position resolution of 0.3 µm. It demonstrates the promising position-sensitive detection capability of 4 H-SiC QPD for applications like precise alignment and laser guidance.

To improve the positioning accuracy of the system, the physics origin of the nonlinearity is analyzed, and the corresponding calibration approach is implemented to reduce the measuring error. Due to the consistency of the photo-response across four quadrants, the Vi (i=1,2,3,4) in Eq. (1) is proportional to the incident optical power on each quadrant Pi, which can be expressed as Pi=∬SiI(x,y)dxdy,(2)where Si is the area of each quadrant and I(x,y) is the optical power density. Besides, the irradiance on the surface of the QPD is considered approximately uniform owing to the large distance between the LED and the collimating lens^[23]. The light spot geometry on the QPD is shown in Fig. 6(a). I(x,y) in Eq. (2) can be then determined as I(x,y)={I0,(x−x0)2+(y−y0)2≤r20,(x−x0)2+(y−y0)2>r2,(3)where I0 is a constant and r is the spot radius.

According to Eqs. (1)–(3), the theoretical output of the QPD can be calculated with full consideration of the spot shape, i.e., circular with a radius of r. Figure 6(b) shows the relationship curve between the calculated σx and x/r when the spot radius r is 1 mm (black line). The corresponding experiment is carried out as well, with the spot position read as the red dots in Fig. 6(b). The theoretical analysis agrees well with the experimental results, revealing that the systematic nonlinear error in the “S-curve” between σx and x should have been introduced by the geometric shape itself. To simplify the calculation with error calibration, a normalized Boltzmann-sigmoidal function is applied to approximate the nonlinear response^[24]. The spot position by the Boltzmann function can then be expressed as {x=k′·ln(1+σx1−σx)y=k′·ln(1+σy1−σy),(4)where k′ is the fitting parameter obtained in the same way as Eq. (1).

To verify the reliability of the calibration method, the results of the two-dimensional spot position measurement in the previous section are reprocessed, as shown in Fig. 6(c). The calibrated points exhibit a more regular pattern as compared to those obtained by the linear fitting method, indicating the excellent improvement of linearity. Then, the positioning error is calculated and its spatial distribution is shown in Fig. 6(d). The positioning errors after calibration are significantly alleviated, with a mean positioning error reduced to ∼28.5μm, which is only ∼22.6% of that obtained with the linear fitting method. In comparison with the calibration methods like polynomial fitting^[25], the calibration method proposed here achieves evident improvement in accuracy without much complexity added, ensuring its practicality in actual UV positioning applications.

For a better comparison, Table 1 presents a summary of the state-of-the-art QPDs based on various semiconductor materials, concerning the key parameters for QPD application. It can be observed that the 4 H-SiC QPD as demonstrated in this work exhibits evidently lower dark current with preserved EQE and responsivity, which will be especially beneficial for the weak UV light spot analysis. Meanwhile, the positioning error as achieved in this work, which has rarely been explored in previous works, should be highly meaningful in leading this QPD technique into practical applications.

In conclusion, UV spot position measurement based on the 4 H-SiC QPD is realized. The fabricated QPD exhibits a low dark current, a high UV responsivity, and a near-ideal uniformity across four quadrants. The two-dimensional position measurement exhibits excellent positioning performance with a slight inherent nonlinearity. The uniform circular spot profile has been recognized as the origin of the nonlinearity. Correspondingly, the Boltzmann function calibration method is proposed to reduce the positioning error, which has achieved significant improvement in accuracy. This work thus demonstrates a prototype UV positioning system based on a 4 H-SiC QPD with substantial potential in precise UV non-contact sensing applications, including LIDAR alignment, laser auto-collimation, and pose measurement.