|Home | About | Journals | Submit | Contact Us | Français|
We demonstrate a novel method to fabricate an axial p–n junction inside <111> oriented short vertical silicon nanowires grown by molecular beam epitaxy by combining ion implantation with in-situ doping. The lower halves of the nanowires were doped in-situ with boron (concentration ~1018cm−3), while the upper halves were doubly implanted with phosphorus to yield a uniform concentration of 2 × 1019 cm−3. Electrical measurements of individually contacted nanowires showed excellent diode characteristics and ideality factors close to 2. We think that this value of ideality factors arises out of a high rate of carrier recombination through surface states in the native oxide covering the nanowires.
In order to make use of silicon nanowires (Si NWs)  in nano-devices, selective doping to form pn junctions or p and n wells is a necessity. Till date, a host of devices with selectively doped Si NWs have been demonstrated [1,2]. Out of them, axial pn and pni junction in Si NWs have shown the potential to be used as solar cells. However, axial pn junctions in NWs grown by the vapor–liquid–solid (VLS) technique have mostly been fabricated by purely in-situ doping [3,5,6]. It has been observed that a pure in-situ doping to fabricate an axial junction may result in unwanted lateral doping  due to unavoidable dopant incorporations through the NW sidewalls by vaporsolid (VS) growth. On the other hand, ion implantation  which is the most widely used doping technique in very large scale integration (VLSI) fabrication can form well-confined doped regions when appropriately used with masking. Ion implantation has been used to fabricate the doped source and drain contacts  as well as the channel  in Si NW-based field effect transistors (FETs). But one of the principal reasons for not extensively using ion implantation to fabricate axial junctions in vertical NWs is possible irrecoverable implantation damages  that were observed in other low dimensional structures such as a FinFET . However, we have shown  that by choosing appropriate ion doses and energies, it is possible to uniformly dope vertical Si NWs of diameter in the range of 100 nm without leaving any residual structural defects in them. Separately, we have also demonstrated in-situ doping of molecular beam epitaxy (MBE)-grown Si NWs . S. Hoffmann et al.  have realized a pn junction in a Si NW purely by ion implantation. However, co-diffusion of acceptors and donors during annealing after such dual implantations of different ions (boron and phosphorus) often lead to the formation of acceptor–donor complexes [15,16] that can anomalously increase the solubility of the donors in the acceptor-rich segments, thus affecting the p– and n– profiles.
In this paper, we demonstrate a novel approach to form an axial p–n junction in a Si NW by combining the above-mentioned ex-situ and in-situ doping techniques. First, we do a modulated in-situ doping with boron by homogeneously doping the lower half of the NW to make it p-type. The upper half of the NW is kept intrinsic (i-type) by simply switching off the boron source. This intrinsic upper half is subsequently converted to n-type by implanting it with phosphorus. We present the details of the fabrication process, the expected dopant profiles in the NW, and electrical characterization of individual NWp–n diodes and explain their typical current–voltage (I–V) curves.
Basics of the growth process including the mechanism of Si NW growth by MBE using Au seeds have already been reported earlier . The NWs were grown on 5″ p-type (boron doped, 5–10 Ω-cm) Si <111> wafers. A B-doped (B concentration ~1018 cm−3) Si buffer layer was grown first on a RCA-cleaned wafer at 525°C in order to provide a clean surface for NW growth and increase the density of the NWs. Afterward, a 1–2-nm thick Au film was deposited in-situ at the same temperature. The Au film subsequently broke into Au droplets to serve as the NW growth initiator . Immediately after this step, Si and B were co-evaporated for 45 min (B concentration ~1018 cm−3). At 45 min, the B source was switched off, while the Si source was kept on 45 min longer. Such a recipe should result in B-doped—intrinsic (pi) type NWs, since the boron diffusion in silicon is negligible at 525°C , i.e., the B atoms incorporated in the lower half will not diffuse into the upper half of the NW. Immamura et al.  verified this with Raman measurements on NWs grown by chemical vapor deposition (CVD) following a similar recipe as ours.
Figure Figure11 illustrates the different steps, accompanied by scanning electron microscope (SEM) images, to fabricate Si NWp–n diodes from the as-grown p–i NWs. The average length of our as-grown p–i NWs amounted to 300 nm and diameter to 125 nm (see Fig. Fig.1a,1a, ,1b).1b). As gold can act as an effective phosphorus ion stopper because of its heavy mass compared to phosphorus, the Au caps on top of the NWs were removed (Fig. (Fig.1c)1c) by an aqueous solution of KI and I2, a standard Au etchant. This resulted in the reduction in the average length of the NWs to around 260 nm (Fig. (Fig.1d).1d). Before the implantations, the samples were spin coated using a spin-on-glass (SOG) silicon dioxide (Silicafilm, Emulsitone Co.) for 30 s at 3,000 rpm. This thereby protected the substrate (Fig. (Fig.1e)1e) and the B-doped lower segment of the NWs from being implanted with P ions from the side. This step effectively eliminates the possibility of lateral doping that is almost unavoidable in purely in-situ doping of NWs successively by two different dopants . A two-step implantation of phosphorus ions at room temperature was used to obtain a rectangular dopant profile. The implantation energies were 45 and 25 keV corresponding to doses of 1.3 × 1014 and 3.2 × 1013 cm−2, respectively. The NWs were tilted by 7° with respect to the impinging ions (Fig. (Fig.1e)1e) to reduce ion channeling . The implanted NWs were subsequently annealed by rapid thermal annealing (RTA) at 850°C for 30 s in Ar atmosphere. Afterward, the SOG was removed using an HF-dip resulting in the NW pn diodes as shown in Fig. Fig.1g1g (corresponding SEM image in Fig. Fig.1h1h).
We illustrate the formation of an axial p–n junction in a NW in Fig. Fig.2.2. Figure Figure2a2a shows the SEM image of a p–n junction NW. Figure Figure2b2b shows the expected phosphorus concentration profile along the length of the NW simulated by the transport of ions in matter (TRIM) code  as well as the expected B concentration profile that results from the in-situ doping. The B profile was taken from the secondary ion mass spectrometry (SIMS) measurements reported earlier . For simplicity, we considered the pn junction formed in Fig. Fig.2b2b to be an abrupt one. Assuming full activation of the dopants, i.e., number of donors (ND) = 2 × 1019 cm−3 (the peak P concentration) and number of acceptors (NA) = 1018 cm3 (the peak B concentration), we calculated the depletion width  as 40 nm. This value is significantly smaller than the average NW length of 260 nm implying that the pn junction should be confined within the length of the NWs.
To confirm the diode behavior of the NWs, we measured their current–voltage (I–V) characteristics by contacting them with a Pt/Ir tip mounted to a micro-manipulator inside an SEM. Details of the measurement system can be found elsewhere . The measured I–V curves of three different pn junction NWs along with an unimplanted pi NW are shown in Fig. Fig.3a.3a. The details of these NWs are listed in Table Table1.1. The inset of Fig. Fig.3a3a shows the I–V curve of the substrate of the NW. As can be seen from Fig. Fig.3a,3a, all the pn NWs show excellent rectifying characteristics with an ON/OFF current ratio of 168, 120, and 50, respectively, at ±1 volt (see Table Table1).1). In comparison, the unimplanted pi NW shows a quasi-Ohmic behavior implying the lack of a significantly rectifying junction. The p-type substrate of the implanted NWs (inset of Fig. Fig.3a)3a) showed an Ohmic (linear) behavior. This confirmed that the phosphorus implantation is indeed forming a pn junction within the NWs as illustrated in Fig. Fig.2,2, and it did not extend to the substrate.
In forward bias, the diode current (I) in the lower voltage range can be written as 
where IS is the saturation current, V the applied voltage, kB the Boltzmann constant, T the temperature, and n the ideality factor of the diode.
We plotted the I–V curves of Fig. Fig.3a3a in a semi-log scale in Fig. Fig.3b.3b. From the slope (S) of the linear parts of these curves in the lower voltage range (−0.2 to −0.6 volt), we extracted the values of n by using
The values of n for the three p–n junction NWs were 2.0, 1.8, and 1.7, respectively (see Table Table1).1). Sah et al.  have found that the ideality factor of a pn diode can vary from 1 to 4 (or even higher in special cases) depending on what kind of current conduction mechanism is dominating. A value close to 2 for the ideality factor indicates that recombination across the pn junction through the surface states is dominant in the carrier transport mechanism. Our pn junction NWs are always covered with a 2–3 nm thick native silicon oxide with an estimated surface state density of 1.1 × 1010 cm−2. These surface states are in direct contact with the pn junction. Therefore we think that surface recombination is indeed playing a major role in the current conduction mechanism across the pn junction resulting in the extracted ideality factors close to 2.
In conclusion, we have demonstrated a novel method to form p–n junction NW diodes by combining two well-established doping techniques—in-situ doping and ion implantation, in succession. The measured NWs showed excellent diode characteristics with a high ON/OFF ratio. The ideality factors of the p–n junctions were close to 2 which points to significant carrier recombinations through the surface states.
The authors thank Mr. A. Frommfeld, Mr. K. U. Assmann, Ms. S. Hopfe, and Ms. C. Muenx for technical support. The authors acknowledge the financial support from the FP6 EU project ‘Nanowire based One Dimensional Electronics’ (NODE).