Energy losses inherent to the conversion of sunlight to electricity in solar cells are mainly due to the so-called spectral mismatch: low energy photons are not absorbed while the energy of high energy photons is only partly used by the solar cell. The losses can be significantly reduced by adapting the solar spectrum. A promising avenue is the use of a downconversion material where one higher energy visible (blue-green) photon is “cut” into two lower-energy near-infrared photons that both can be used by the solar cell. Here the efficiency of downconversion for the $\left({\text{Nd}}^{3+},{\text{Yb}}^{3+}\right)$ couple in ${\text{YF}}_3$ is studied to investigate if efficient two-step energy transfer occurs from the ${{}^4\text{G}}_{9/2}$ level of ${\text{Nd}}^{3+}$ (situated around $21000{\text{cm}}^{-1}$ or 470 nm) exciting two neighboring ${\text{Yb}}^{3+}$ ions to the ${{}^2\text{F}}_{5/2}$ level (around $10000{\text{cm}}^{-1}$ or 1000 nm). Optical measurements of ${\text{YF}}_3$ doped with ${\text{Nd}}^{3+}$ and ${\text{Yb}}^{3+}$ show that there is efficient energy transfer from ${\text{Nd}}^{3+}$ to ${\text{Yb}}^{3+}$, but downconversion from the ${{}^4\text{G}}_{9/2}$ level does not occur due to fast multiphonon relaxation. Relaxation from this level to lower-energy levels populates the ${{}^4\text{F}}_{3/2}$ level of ${\text{Nd}}^{3+}$ from which efficient one-step energy transfer to ${\text{Yb}}^{3+}$ occurs. Analysis of the luminescence decay curves for different ${\text{Yb}}^{3+}$-concentrations using Monte Carlo simulations reveals a high nearest neighbor transfer rate $\left(3.3\times {10}^5{\text{s}}^{-1}\right)$ through a dipole-dipole interaction mechanism. Downconversion is observed from the ${{}^4\text{D}}_{3/2}$ level (situated in the UV, around $28000{\text{cm}}^{-1}$ or 360 nm) with an estimated quantum efficiency up to 140%. For application in solar cells this UV to 2 NIR downconversion will only result in a marginal reduction of spectral mismatch losses.

• Received 9 October 2009

DOI:https://doi.org/10.1103/PhysRevB.81.035107