Improvement of Boron Dopant Quantification Accuracy in Atom Probe Tomography via High Electric Field Analysis

The characterization (i.e., spatial distribution and concentration) of dopants within site-specific regions of a device is of critical importance to the semiconductor industry. However, this scope of investigation is complicated by the continued miniaturization and geometric complexity of individual components and structural features on devices. Atom Probe Tomography (APT) has the potential to meet this demand with the unique capability of three-dimensional, chemical mapping at the atomic-scale [1]. Although boron (B) is the most popular p-type dopant for silicon (Si)-based devices, it has traditionally suffered from poor quantification accuracy in APT. These challenges are due to the complexity of B field evaporation mechanics, which is believed to be a result of its high evaporation threshold compared to Si [2]. Consequently, B tends to be preferentially retained on the surface of an APT specimen and evaporate non-stoichiometrically, resulting in a concentration profile that is elongated in the depth direction of analysis. This has also been verified by Secondary Ion Mass Spectrometry (SIMS) measurements [3]. The retained B may also experience surface migration, which can degrade the lateral resolution and result in B detection from outside the doped region(s). Moreover, B is generally underestimated using APT as a result of its natural tendency to co-evaporate in burst events and the inability of a three-anode, delay-line detector to resolve such evaporated ions that are insufficiently separated in space and time [4]. The severity of these artifacts has been previously shown to have a dependence on the experimental conditions employed [3, 5]. For the purposes of enabling accurate quantifications using APT, this study provides a holistic investigation into the severity of observed detection losses and profile shape integrity for a wide range of electric field analysis conditions. The NIST Standard Reference Material (SRM) 2137, a B implant in Si, is used as a well-characterized baseline sample with a certified dopant dose and a defined concentration profile shape [6]. The Si Charge-State Ratio (SiCSR), defined as Si2+/Si+, is used as an indirect measure of the average surface electric field during analysis. Specimen preparation followed a standard FIB lift-out procedure for APT needles, performed on a Thermo Scientific Helios 5 UC DualBeam. APT data was acquired on a CAMECA LEAP 5000 XS at a constant detection rate of 0.005 ions/pulse (0.5%). The analysis conditions were set so that the laser pulse energy would automatically be adjusted to maintain a target CSR value and, therefore, relatively constant field evaporation conditions. A high apex electric field is shown to enhance quantification accuracy in terms of the total retained B dose measured (increasing from 53.6% to 104.7% from the lowest to the highest electric field analysis), the profile shape integrity (Figure 1), and the lateral resolution (Figure 2). Evidence of extensive surface migration at the lowest apex electric field will be presented through desorption maps, which indicate a highly inhomogeneous B emission and its localized detection around the center of the detector. In the context of routine investigations of semiconductor devices, these findings offer pathways towards best-practices to enable accurate and repeatable measurements. B concentration profiles for the lowest and highest electric field analysis conditions. 2D B concentration maps for the lowest and highest electric field analysis conditions.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.