Improving Analytical Capability via Simultaneous Voltage and Laser Pulsing in Atom Probe Tomography

Voltage and laser pulsing (V+L) in atom probe tomography (APT) [1] has been proposed as a means to improve sensitivity and compositional accuracy [2,3]. In this pulsing scheme, background reduction [2,4] and mass spectral peak sharpening [3] may be achieved, especially for materials with poor thermal diffusivity. Recently, background reduction was achieved for a variety of materials, including aluminum (∼8X), stainless steel (∼4X), silicon (∼15X), and silicon dioxide (∼2X) [4]. This work further explores the phenomenon described in [3] with respect to using a voltage pulse to improve the peak shapes from a specimen in order to improve mass resolving power (MRP) and thus improve compositional accuracy. In APT, a laser pulse (Fig. 1a) is applied onto a DC voltage and background noise is generally inversely related to the magnitude of the pulse (red arrow in Fig. 1a). The application of concurrent voltage and laser pulses decreases background by reducing the probability for field evaporation to occur between laser pulses, Fig. 1b. If the falling edge of the voltage pulse is properly aligned with the thermal response of the specimen, Fig. 1c, (as proposed in [3]), then the thermal tail of the mass peak (ion emission occurring after maximum apex temperature has been achieved) is minimized. Results of using the configurations shown in Fig. 1 are illustrated in Fig. 2 which shows spectra from BaTiO3 focused on the 138Ba2+ peak located at 68.95 Da. The laser-pulsing only (black curve) shows a mass peak with a thermal tail, common in a material with relatively poor thermal diffusivity. This tail is cropped with the application of V+L mode (red curve) and is further improved by optimizing the relative timing of the voltage and laser pulses (blue and green curves). These improvements are quantified in Fig. 3 for the MRP measurements at half (FWHM) and hundredth (FW0.01M) maximum. For the optimal condition V+L (green curve), the FWHM resolution is not degraded and the FW0.01M measurement shows an improvement of ∼2.6X. By improving MRP, especially in the region of peak tails, compositional quantification may be improved, as shown in Fig. 4. Here the measurement of a Zr-related-peak located at 54.5 Da in between other closely spaced ZrO peaks in a Zr-based oxide is examined. In Fig 4a, this peak (black arrow) is obscured by general background noise level and the thermal tail from a preceding peak, while in Fig 4b this peak is much more distinct (spectra normalized by total counts). Since the expected intensity is not known for this complex ion peak, comparisons of expected uncertainties in these two measurements are made. We measure background-corrected ionic percentages of 0.0107 ± 0.0006 and 0.0091 ± 0.0001 for the laser-only case and for the optimal V+L case, respectively. For the latter case, the one sigma error estimate is improved by ∼4.5X in the V+L data due to the reduction in both general background (∼2.5X) [4] and the thermal tail of the preceding peak. Evaporation versus time plots depict how standing voltage and pulse shapes affect mass spectrum quality. (a) Background noise generated in APT, (b) reduction of background noise using V+L pulsing, and (c) positioning of the voltage pulse optimizing MRP. Mass spectrum showing the Ba2+ peak in BaTiO3 comparing MRP for laser-only (black), V+L-pulsing (red), and two different offsets of the V+L pulses (blue and green). MRP at FWHM (dashed black) and FW0.01M (solid black) for the same four cases presented in Fig. 2. Conditions: T=50K, laser energy=25pJ and voltage pulse fraction=30%. Normalized comparison of the quantification of the 54.5 Da peak in a zirconium-based oxide comparing the (a) laser-only case vs the (b) V+L case.