Complete Report - University of New South Wales

ARCPHOTOVOLTAICSCENTRE OFEXCELLENCE2010/11ANNUAL REPORT(a) Conductivities measuredthroughout the entire voltagerange. (b) ln(σ) vs 1/T 1/2 plotfrom 120 to 220 K, for the 0.54at.% Sb-doped sample. Thesolid line is least square fitto the data. (c) Temperaturedependentconductivities forthe undoped and 0.54 at.%Sb-doped samples.Figure 4.5.13The temperature dependent conductivity dataof Fig. 4.5.13(a) indicate that the conductivity of0.54 at.% Sb-doped SRN film (~2.8×10 -2 S/cm)is six orders of magnitude higher than that forundoped material (~7.3×10 -8 S/cm), which could beattributed to an increase in carrier concentration.Furthermore, n-type electrical behaviour withcarrier concentration of 2.4×10 16 cm -3 and mobilityof 2.94 cm 2 /V s in the doped films as observed fromHall measurements was attributed to free carriergeneration due to the effective Sb doping.The temperature-dependent conductivities ofboth undoped and Sb-doped samples are shownin Fig. 4.5.13(b). As can be seen, doping of theSi-NCs strongly influences the electronic transportproperties of the films. Arrhenius-like temperaturedependence is observed in the T range between220 and 320 K, attributable to thermally activatedconduction. From the slope of the Arrhenius plot,we found that the activation energy E Adecreases to0.182 eV for the Sb-doped Si-NCs film, suggestingeffective n-type doping of the Si-NCs. Nevertheless,the extracted E Ais much larger than that of bulk Siwith the same dopant. One possible explanation forthe large E Aobserved would be the deeper donorlevel expected in Si-NCs. Also, it is possible that thenumber of free carriers available for conductionmay be limited due to a trap density distributedwithin the bandgap that is comparable to thedoping density.Interestingly, the conductivity for the Sb-dopedsample at T < 220 K deviates from Arrheniusbehavior and is best described by the expression, σ= σ 0exp [-(T 0/T) 1/2 ] as illustrated in Fig. 4.5.13(c). Thecharge transport mechanism can be explained wellby a percolation-hopping model.4.5.2.3.2 Si QDs in SiO 2/Si 3N 4hybrid matrixReserachers:Dawei Di, Ivan Perez-Wurfl, Gavin ConibeerTo improve the current transport properties in thevertical direction and to obtain better size control ofSi quantum dots, we proposed a new design basedon Si QDs embedded in a SiO 2/Si 3N 4hybrid matrix[4.5.27, 4.5.28]. By replacing the SiO 2tunnel barrierswith the Si 3N 4layers, the new material managesto constrain the growth of doped Si quantumdots effectively and enhances the apparent bandgap. Also electrical characterisation on Si QD/c-Si hetero-interface test structures indicates thenew material possesses improved vertical carriertransport properties.Doped and undoped samples with differentbarrier dielectrics (SiO 2and Si 3N 4) are compared.Samples were fabricated using the co-sputteringmultilayer technique with layers containing excessSi consisting of SRO for both materials, with an SROthickness of 4 nm. Samples were either undoped orco-sputtered with either P 2O 5or B for n- and p-typedoping respectively. After annealing at 1100⁰Cthe crystalline properties were characterised byX-ray diffraction (XRD), as shown in Fig. 4.5.14. Forsamples with Si 3N 4barriers, the measured Si NCsizes are 4.3 nm when undoped, 3.5 nm when Bdoped and 5.0 nm when P 2O 5doped. For sampleswith SiO 2barriers, grain sizes are 7.7 nm whenundoped, 7.2 nm when B doped, and 15 nm whenP 2O 5doped. The Si NC sizes vary significantlyalthough all samples have the same silicon-tooxygenratio in the SRO layers and the same asdepositedlayer thickness.It can be observed from the experiment thatsamples with Si 3N 4barriers contain generallysmaller Si QDs than samples with SiO 2barriers.This is because Si 3N 4, a denser and stiffer materialthan SiO 2, acts as a better diffusion barrier thanSiO 2[4.5.27]. It could also be thermodynamicallyrelated to the interface free energy between68