The transmittance and reflectance of a 700 nm BiOI film on solution processed NiOx on glass was
measured by UV-visible spectrophotometry (UV-vis), using an integrating sphere to account for scatter due to sample roughness. Note that NiOx coated glass was used as the substrate due to
improved coverage relative to glass substrates, as will be discussed later in the chapter (section 6.3.1). From this, the absorption coefficient was calculated using the following derivative of the Beer-Lambert law.249 𝛼 = − ln (𝑇 𝑇𝐵𝑖𝑂𝐼 𝑁𝑖𝑂𝑥(1 − 𝑅𝐵𝑖𝑂𝐼)) 𝑑 (6.23)
Where α is the absorption coefficient (cm-1), T
BiOI is the transmittance of BiOI, TNiOx is this
transmittance of NiOx on glass and RBiOI is the reflectance of BiOI and d is the BiOI thickness
(cm), measured by profilometry.
The absorption coefficient as a function of wavelength is shown in Figure 6.2. It has a slow onset at ~1.9 eV and rises to > 1 x 105 cm-1 beyond 2.6 eV. Band structure simulations have
shown that BiOI is an indirect semiconductor;182 this is responsible for the slow onset of
absorption. An absorption coefficient >105 cm-1 is large for an indirect semiconductor.2 It should
be noted that the absorption coefficient determination by this method is prone to sources of error. For example, there were differences in film thickness/uniformity across the film, whilst an average value for the thickness was used to determine α from equation 6.23. Additionally, TBiOI
and RBiOI were not measured at the exact same point on the film. It would therefore be ideal to
repeat the measurement at multiple locations across the film, allowing average and standard deviation values for α to be determined at each wavelength.
It is possible to determine the bandgap of BiOI using α and the Tauc equation:250,251
(𝛼ℎ𝜈)1𝑟 = 𝐴(𝐸𝑔 − ℎ𝜈) (6.24)
where Eg is the bandgap (eV), hv is the energy of the incident photon (eV), A is a constant and r
is representative of the type of transition, (r =2 for an indirect transition, r = ½ for a direct transition).186 As BiOI is an indirect semiconductor, (αhv)1/2 was plotted against energy to give
the plot in the inset of Figure 6.2. From equation 6.24, the band-gap energy was determined from the intercept with the x-axis (i.e. (αhv)1/2 = 0):
85 Hence,
𝐸𝑔 = ℎ𝜈 (6.26)
From Figure 6.2 (inset), the bandgap of BiOI was 1.91 eV. This is in good agreement with literature values for BiOI.185,186 The bandgap energy is large for a single junction solar cell
material; the Shockley-Queisser limit is < 25% at this energy.31 However, for a two-junction
tandem with a silicon bottom cell a bandgap of 1.9 eV is optimum, for which a theoretical maximum efficiency of ~ 45% has been calculated for a four-terminal device.32
The absorption depth was obtained from the inverse of the absorption coefficient and is plotted as a function of wavelength in Figure 6.2b. The bandgap is labelled at 1.9 eV (~650 nm). The absorption depth was < 1 μm from 400 – 600 nm, after which it increased to 3 μm at the band-edge (650 nm). The implication for devices is that films of micron-scale thickness are required to absorb all incident light. This necessitates carrier diffusion lengths > 1 μm for good current collection in devices, placing stringent requirements on the defect characteristics of BiOI.2
The photoluminescence spectrum of BiOI was also recorded, shown in Figure 6.2c. Here, quartz was used as the substrate to avoid emission peaks from the substrate, as it is non-emissive. BiOI gave weak emission, typical for an indirect semiconductor due to the change in electron momentum required for an indirect transition. Weak PL emission has been reported for many other bismuth-based absorber compounds, which tend to have indirect bandgaps.1 A broad
emission peak centred at ~ 1.9 eV was present, in agreement with the value measured by UV-vis photo-spectroscopy. This means that emission occurred from the band-edge as opposed to a defect state.2 However, the PL had a long tail towards lower energies, indicating a defect tail at the band
86
Figure 6.2 a) The absorption coefficient, α, as a function of wavelength for BiOI (orange),
calculated from uv-visible spectrophotometry data of a 1 μm BiOI film on a glass|solution processed NiOx substrate. The inset shows the Tauc plot for BiOI considering an indirect
transition (i.e. a plot of (αhv)1/2 against wavelength), b) the absorption depth of BiOI as a function
of wavelength, with the band-gap energy labelled at 1.9 eV. c) shows the photoluminescence emission spectrum of a 1 μm BiOI film on quartz.
Time correlated single photon counting (TCSPC) was used to measure the lifetime of BiOI. A 1 μm BiOI film was grown onto a quartz substrate, again to avoid emission from the substrate. Samples were excited by a laser of wavelength 532 nm with a fluence 400 nJ cm-2, and the photons
emitted between 640 and 700 nm were counted, yielding the data in Figure 6.3. The laser pulse occurred at 0 s.
a)
b
)
87 A bi-exponential decay function:
𝑦 = 𝑦
0+ 𝐴
1𝑒
−𝑡 𝜏 1 ⁄+ 𝐴
2𝑒
−𝑡𝜏 2 ⁄ (6.27)where y is intensity, y0 is an off-set, A1 and A2 are constants of amplitude, t is time and τ1 and τ2
are values of the lifetime, was fitted to the data using Origin software. Lifetimes τ1 = 0.40 ns and
τ2 = 2.75 ns were extracted from the fit. The shorter lifetime, τ1, is often attributed to surface
recombination, whilst the τ2 refers to the lifetime of the bulk.2,252 Screening through the lifetime
of a series of establish light absorbers, Jaramillo et al previously deduced that materials with a lifetime > 1 ns in early stage research are worth exploring for photovoltaics.50 By this metric, by
possessing a bulk lifetime of 2.75 ns, CVD-deposited BiOI is identified as a promising photovoltaic material.1,2 This motivated the fabrication of devices from BiOI, the development
and characterisation of which will make up the rest of this chapter.
Figure 6.3 TCSPC spectrum of a 1 μm BiOI film on quartz. The sample was excited at 532 nm
at a fluence of 400 nJ cm-2 and photons were counted between 640 and 700 nm at 0.008 ns
88