Electrical characterization of single nanometer-wide Si fins in dense arrays

Introduction

The transition from planar to three-dimensional transistor architectures such as the fin field-effect transistor (finFET) [1] has raised the need for measuring the electrical properties of nanometer-wide conducting features [2]. Recently, it has been shown that the micro four-point probe (μ4pp) technique, which is commonly used for sheet resistance measurements on blanket materials or relatively large pads (larger than 80 × 80 µm²) [3-5], provides a solution to this requirement [6]. The μ4pp technique was demonstrated to provide (sheet) resistance measurements in single fins without the need for dedicated Kelvin resistor or transmission line structures [7]. However, the results demonstrated in [6] focused on isolated fins whereby the fin pitch was larger than the contact size of the μ4pp electrodes such that only one single fin was contacted at a time. Intuitively, this suggests that the technique developed therein fails when trying to measure dense structures where a fin pitch smaller than the apparent contact size of the electrodes is used (see below in Figure 2). In that case, the μ4pp technique appears to be of limited value in routine semiconductor manufacturing where state-of-the-art chips use much smaller fin pitches [8].

In this paper, we describe further developments of the μ4pp technique, as implemented by the CAPRES A300 tool, which enable the electrical characterization of single nanometer-wide fins in dense fin arrays (pitch < 200 nm) with high precision and repeatability. First, we describe the general concept of how to establish and control the electrical contact between the metallic (Ni-coated) μ4pp electrodes and the semiconducting (Si) fins. Next, we show that, by carefully controlling this process, the electrical contact can be confined to one single fin such that the resistance of individual fins in dense arrays can be measured with a high precision. Finally, we use the technique to determine the electrical resistance of individual fins in a dense array and we demonstrate that the measured resistance correlates with the geometrical width of the fins, as measured with transmission electron microscopy (TEM). Due to the demonstrated high precision, a critical dimensional sensitivity of ca. 0.5 nm could be achieved.

Experimental

Before discussing the electrical contact between the μ4pp electrodes and an individual fin, a general description of a μ4pp measurement on large blanket semiconducting samples is needed. The μ4pp electrodes comprise four Ni-coated Si cantilevers with a spacing of 8 µm and a contact size d_contact ≈ 300 nm [6,9,10]. In a μ4pp measurement, the electrodes are landed on the sample surface after which a current I_in is injected into the investigated sample via two of the electrodes while the induced voltage drop V is measured between the other two electrodes. Initially, however, the native oxides present both on the semiconducting material and the Ni-coated electrodes act as highly resistive barriers and therefore prevent any electrical contact [11]. To establish the electrical contact, the μ4pp technique uses the so-called punch-through current, i.e., a short current pulse of magnitude I_pulse applied between two electrodes, which causes the breakdown of the native oxide barrier [12-14] and hence creates the conductive path required to inject I_in into the investigated material. Empirically, it is observed that the magnitude of I_pulse must be chosen larger than a certain threshold current (I_threshold, typically >100 µA for blanket materials) in order to reduce the contact resistance R_C between the electrodes and the sample and hence activate the required electrical contact.

The given description of the punch-through mechanism is also valid for more confined structures, such as fins. This does, however, require some additional considerations, starting with the distinction between isolated and dense fins. First, for isolated fins (Figure 1a), i.e., fins are separated by a distance (= pitch) larger than d_contact, the procedure is identical to the previously described case of blanket materials. The electrical contact is indeed created, i.e., contacts j = 1, 2, 3, 4 are activated, when I_pulse ≥ I_threshold and the electrical resistance R_fin of the region of the fin included between the two inner contacts is readily obtained from the ratio R_fin = V/I_in [6]. Secondly, in the more complex case of dense fins, i.e., fin pitch < d_contact, the μ4pp electrodes can physically contact multiple fins at the same time. For simplicity, this paper only considers the case of two fins physically contacted by the electrodes (Figure 1b). In this situation, electrical contact is formed on both fins, i.e., contacts j = 1, 2, …, 8 are activated, when the magnitude of I_pulse is similar as used on blanket materials. The measured resistance is then determined by the ratio between the two currents I_in1 and I_in2 injected into the two electrically connected fins. Since this ratio depends on the contact resistances R_Cj (j = 1, 4, 5, 8), this leads to a high measurement variability, i.e., a loss in precision [3]. As a consequence, in order to precisely determine R_fin in a dense fin array, I_pulse should be carefully controlled (I_pulse < 2 × I_threshold) to only allow for the formation of electrical contact to one single fin, i.e., only contacts j = 1, 2, 3, 4 or j = 5, 6, 7, 8 are activated. On top of that, to make sure that all four electrodes indeed form electrical contact with the same fin, the punch-through mechanism between electrode pairs must be sequenced properly. For example, when first applying the punch-through mechanism on the top two electrodes, which then form electrical contact with the left fin in Figure 1b, i.e., contacts j = 1 and j = 2 are activated, the next electrode pair should use an already activated contact to make sure the contacts on the same fin, i.e., j = 3 or j = 4, are activated next. Note that, however, the exact behavior of the electrical contact formation on dense fins is still not fully understood and a more thorough description would also include pulse duration, peak voltage, and material properties.

Results and Discussion

The experimental demonstration of using the punch-through current I_pulse to individually contact single Si fins in dense arrays is shown in Figure 2a, where the measured R_fin is plotted as a function of the fin width W_fin after using a high (100 µA) or low (25 µA) punch-through current to form the electrical contact. To highlight the impact of the fin pitch, we have additionally separated the isolated and dense fins, assuming the approximately 300 nm physical contact size of the electrodes as measured with scanning electron microscopy (SEM) [9]. It can be observed that, while I_pulse does not affect the precision on isolated fins (red and blue triangles), for dense fins a major improvement in precision can be achieved by decreasing the punch-through current from 100 μA (red diamonds) to 25 μA (blue diamonds). Based on the previous theoretical considerations, the improvement in precision is achieved by restricting the electrical contact to one single fin despite the electrode being in physical contact with two fins. To show this improvement more clearly, the relative standard deviation of the measured values of R_fin can be plotted against the fin pitch, as shown in Figure 2b. Excitingly, the precision of the 25µA punch-through current measurement remains stable at around 3%, making the measurement feasible even for fin pitch much smaller than d_contact. Note that Figure 2 also shows that the fin width has no impact on the measurement precision.

The ability to probe individual fins in dense arrays allows us to exploit the high precision of the μ4pp tool [15] to electrically characterize nanometer-wide fins regardless of the fin pitch. To demonstrate this, Figure 3 shows that we can now measure variations in fin resistance induced by nanometric variations in fin width in a dense array of narrow Si fins. For this, we used an array of ten ca. 20 nm wide Si fins implanted with B (3 × 10¹⁵ cm⁻², 5 kV) and laser-annealed three times at 1150 °C. Note that W_fin is assumed constant, i.e., the very small tapering of the fins along the shallow (ca. 60 nm) implant depth [6] is ignored. These fins, having a pitch of 200 nm, were measured individually by using a punch-through current of 25 µA to restrict the electrical contact to a single fin. Moreover, by running the μ4pp measurement over the fin array with a step size of ca. 25 nm, we could assign the measured values of R_fin to each specific fin. As can be observed for the four out of ten fins shown in Figure 3a, R_fin varies in accordance with the fin width measured by TEM. Note that the error in Figure 3a is 3.0% for each fin, which was obtained by taking the lowest precision achieved out of all ten measured fins. Since R_fin is obtained by taking the average of several subsequent measurements, the precision includes the variation in the exact position of the electrical contact points for each landing of the electrodes, i.e., a variation in contact spacing s, which may result both from a variation in the electrode positioning itself and from the exact location of the small electrical contact under the wider electrode. Additionally, using the widths measured with TEM and the relation R_fin = R_s^fin × s/W_fin (using s = 8 µm), the inset of Figure 3b shows that all ten fins have the same sheet resistance R_s^fin ≈ 200 Ω/sq, indicating that the observed variations in R_fin are indeed caused by variations in fin width. This allows us to evaluate the sensitivity of the technique by plotting the measured values of R_fin as a function of the fin width (Figure 3b) and subsequently fitting the data with a constant sheet resistance R_s^fin using the relation R_fin = R_s^fin × s/W_fin. By comparing the slope of the fitted curve at W_fin = 20 nm (ca. 4 kΩ/nm) to the achieved precision (ca. 2.3 kΩ, Figure 3a), we can deduce that the technique has a sensitivity to variations in fin width down to about 0.5 nm, opening the prospects for its use in electrical critical dimension metrology. As also interestingly shown in the inset of Figure 3b, the measured R_s^fin is higher than the sheet resistance R_s^pad = 135 Ω (dashed red line) measured in a large 80 × 80 μm² pad having undergone the same implantation and annealing treatment. This increase in sheet resistance when going to nanoscale elongated geometries was expected and understood to originate from the presence of interface states as well as defects at the fin sidewalls [6].

Conclusion

This paper demonstrates the capability of μ4pp to electrically characterize individual nanometer-wide Si fins in dense arrays regardless of fin pitch. By carefully controlling the electrical contact, we were able to measure the resistance of individual ca. 20 nm wide fins in dense arrays even though the μ4pp electrodes physically contact more than one fin. Thanks to the high precision of the measurements, the correlation between measured resistance and nanometer-scale variations in fin width could be demonstrated with a sensitivity as small as 0.5 nm.