Introduction

Luminescent signal-based applications continue to attract significant attention because of their potential in fields of biosensing, biomedical imaging, therapies, energy conversion, and also security. In the early days of luminescent materials, organic dyes and quantum dots (QDs) were reported for use in numerous applications15. Organic dyes have a good biocompatibility, relatively high fluorescence intensity and facilitate the easy modification of the surface for covalent conjugations. However, photobleaching, chemical degradation, and broad emission spectra with long tailing limit their practical applications6. Due to the rapid development of nanotechnology, QDs have emerged as an alternative fluorescent material that possesses the unique properties of an intense and size-tunable emission, a good photostability, an excitation in the ultraviolet (UV) region, and narrow emission bands. These QDs have led to the various applications in fluorescence resonance energy transfer (FRET) based analysis, tissue imaging, and electrochemical energy conversion. Unfortunately, due to the inherent cyto-toxicity of heavy metal-based QDs, their applications have been limited7. Moreover, QDs typically require UV radiation, which is potentially harmful to living cells and biomolecules, including proteins and nucleic acids.

In contrast to QDs, lanthanide-doped upconversion nanoparticles (UCNPs) can convert near-infrared (NIR) light into UV, visible, or NIR light through a nonlinear optical phenomenon that is known as anti-Stokes emission, wherein two or more low-energy photons are absorbed leading to a high-energy emission. Different from traditional fluorescent materials,

UCNPs possess the following powerful advantages: 1) an outstanding signal-to-noise ratio offers improved detection sensitivity, owing to less autofluorescence in NIR diagnostic window range of 700 – 1100 nm. 2) Further, the utilization of the NIR excitation light allows for deeper light penetration and reduced photodamage effects, and 3) the lack of toxicity enables long-term use in biological environments. Additional advantages of UCNPs are, namely: narrow emission peaks, large anti-Stokes shifts, and good chemical- as well as photo-stability8. Over the past decade, these unique properties of UCNPs have led to successful applications, including background-free biological sensing, high contrast biomedical imaging, energy conversion, and security9. Moreover, UCNPs with various dopants (e.g., Gd, Sm, etc) can combine luminescent imaging with other imaging modalities that pave the way for novel theranostic applications such as magnetic resonance imaging (MRI) as well as computed tomography (CT)10. In this review, we focus on the recent advances in design and applications of UCNPs. We categorized our review to cover upconversion luminescence mechanism, synthesis methods, and the luminescence signal-based analysis of target markers. Moreover, it elucidates the diagnosis and therapeutic applications. Solar light-induced energy conversion applications have also been addressed along with luminescence signal-based applications for security (See Figure 1).

Figure 1
figure 1

Schematic illustration of upconversion nanomaterial applications in diverse fields.

Upconversion Luminescence Mechanism

Upconversion is a nonlinear optical process of converting low energy (e.g., NIR light) electronic transitions into higher energy electronic transitions (e.g., visible or UV light) and can, therefore, be ascribed as an anti-Stokes mechanism. Upconversion luminescence (UCL) emissions are generated by optical transitions within the partially filled 4f orbitals of lanthanide ions11. The theoretical possibility of this phenomenon in ion-doped solid-state infrared quantum counters was first introduced by Nicolaas Bloembergen in 195912. The mechanism of upconversion process has been extensively studied in recent years and can be divided into five main classes, which include excited state absorption, energy transfer upconversion (ETU), photon avalanche, cooperative sensitization upconversion, and energy migration upconversion. It should be noted that among these five proposed mechanisms the ETU mechanism of energy transfer strongly dominates13. Thus, we will focus on the description of the ETU process here.

Energy Transfer Upconversion

The ETU process takes place between the sensitizer and activator system with a non-radiative energy transfer pathway that exists between two neighboring ions as illustrated in Figure 2A. In a typical ETU process, the photon excites an ion known as the sensitizer from the ground state (G) to enter a long-lived intermediate, metastable excited state of the sensitizer ion and energy transfer then occurs from the sensitizer ion to the metastable level 1 state (E1) of the activator ion. Continuously, the photon then promotes this ion from the activator E1 state to the higher, metastable excited state 2 (E2). When the sensitizer ion relaxes back to G state, the activators populated in the E2 state radiatively relax to the G state and emit upconversion photons. The efficiency of the ETU process depends on the distance between the sensitizers and activators, which relies on the concentrations of dopant ions. As a result, doping is the most widely implemented strategy in the design of upconversion nanomaterials.

Figure 2
figure 2

Schematic illustration of energy transfer process for (A) ETU and (B) Energy transfer mechanism of NaYF4:Yb,Er/Tm co-doped upconversion nanomaterials under 980 nm excitation.

Selection of Host Materials, Sensitizers and Activators for Tunable Emission

Upconversion nanomaterials are mainly composed of a host matrix, a sensitizer ion and an activator ion. Appropriate selection of host materials is essential for an efficient upconversion process. Recently, the upconversion process has been reported in numerous nanoscale host matrixes such as fluorides, oxides, phosphates, and vanadates1417. Among these types of host materials, fluorides have been shown to be a superior host material for the upconversion process, since fluorides show low phonon energies and high chemical stability18. Thus, NaYF4 has become the most popular host material for high-quality UCNPs.

The most commonly used lanthanide dopant ion combinations for the fabrication of upconversion nanomaterials are Yb3+/Er3+ and Yb3+/Tm3+. Note that other ions with partially filled 4f orbitals can also be used as dopants ion. As shown in Figure 2B, Yb3+ is employed as the sensitizer for Tm3+ or Er3+. Codoped Yb3+ systems have a large absorption cross-section in the NIR region at 980 nm. NIR photons are firstly absorbed by the Yb3+ ion sensitizer and then are transferred to a near activator emitting center (e.g., Er3+ and Tm3+) via the ETU process. Radiative relaxation of the excited activator ions emits photons each red, green, and blue emissions (Figure 2)19,20. The emission intensity of upconversion nanomaterials can be affected by numerous factors, such as dopants concentrations, particle size, morphology, host material, crystalline phase and method of preparation2123. Upconverting processes of multiple-step between lanthanide ions commonly have a relatively low conversion quantum yield. Therefore, the enhancement of the upconversion intensity can be achieved by annealing treatment or shell formation for surface passivation which minimizes luminescence quenching.

Synthesis of Upconversion Nanomaterials

To obtain high-quality upconversion nanomaterials, many efforts have been devoted to the exploration of developing facile synthetic methods such as co-precipitation and thermal decomposition. One of the main goals in the many synthetic methods is the reproducibility of monodispersed upconversion nanomaterial. These particles should exhibit a narrow size distribution, be phase-pure, and be highly soluble in aqueous solutions for biological applications. In the following section, we discuss the common synthetic methods of co-precipitation and thermal decomposition. Synthetic process of upconversion nanomaterials is briefly described in Figure 3.

Figure 3
figure 3

Synthesis methods of upconversion nanomaterials. (A) Thermal decomposition method (B) Co-precipitation method.

Thermal Decomposition Method

In a typical thermal decomposition procedure for fluoride-based nanomaterials, a rare earth metal trifluoroacetate (RE(CF3COO)3) is thermally decomposed to provide the corresponding metal and fluoride ions. Novel synthesis of pure-phase and monodisperse LaF3 triangular nanoplates from a single-source precursor of RE(CF3COO)3) was first reported by the Zhang et al. through the thermal decomposition24. This method was further applied to alkali lanthanide tetrafluoride-based UCNPs (NaREF4). Mai et al. reported a general synthetic strategy for crystalline, monodisperse NaREF4 UCNPs via metal/RE-trifluoroacetate precursors (Na(CF3COO) and RE(CF3COO)3)25. The solvent for this procedure consists of a coordinating solvent and a non-coordinating solvent. To provide a high-temperature environment, 1-ocatadecene (ODE) was used as the non-coordinating solvent (b.p.315 °C), and the reaction was performed under a protective layer of argon gas. As coordinating solvents, oleic acid (OA) and olelyamine (OM) were used to prevent aggregation by capping the surface of nanoparticles. Using a single solvent trioctylphosphine oxide (TOPO) Shan et al. reported a new thermal decomposition method such that the energy barrier of the cubic to hexagonal phase transition was considerably reduced by using TOPO as the solvent, providing a more efficient way to form hexagonal phase UCNPs a rather than OA/OM/ODE combined solvent system26.

Co-Precipitation Method

Co-precipitation method is the easiest way for synthesizing lanthanide-doped upconversion nanomaterials. Yi et al. first reported the co-precipitation method for synthesizing NaYF4:Yb,Er with the aid of ethylenediamine tetraacetic acid (EDTA)27. The cubic-phase of α-NaYF4:Yb,Er resulted from that co-precipitation procedure via a rare earth (RE) complex of EDTA with a NaF solution. To increase the UCL emission intensity, an annealing treatment was applied that resulted in the phase transition from a cubic to hexagonal phase. This transition is critical, because of the weak luminescence emission intensity of the cubic-phase upconversion nanomaterials (α-NaYF4). However, aggregation of UCNPs after the annealing process limits their potential applications. Moreover, the capping ligands EDTA could be carbonized after the annealing process, reducing the hydrophilicity of the upconversion nanomaterials. Therefore, further surface modifications are required to improve the water-solubility of upconversion nanomaterials. For example, polyethylenimine (PEI), an organic polymer surfactant, is used as a surface ligand to improve water-solubility and prevent particle aggregation.

Exploiting Applications of Upconversion Nanometerials

Detection and Analytical Platform

Upconversion nanomaterial signal-based detection assays have the powerful property of low signal-to-noise levels in the NIR-region, leading to an outstanding limit of detection (LOD) in biological environments. Therefore, these materials have been utilized in sensitive and specific methods for the detection of biomolecules, pH changes, metal ions and temperature. The assay methods are categorized as heterogeneous assays and homogeneous assays. Heterogeneous assays are commonly carried out in a solid-phase system that requires several wash steps or sample preparation steps, while a homogeneous assay is generally performed in a liquid-phase, where an optical measurement utilizes a simple mix and readout system.

Heterogenous Assay

To detect target analytes, the heterogeneous assay method utilizes highly specific recognition with binding affinity between the target analytes and functionalized UCNPs, immobilized on a solid-phase substrate. Based on the upconversion nanomaterials, heterogeneous assays have been successfully explored and exhibited improved LOD2831. One of the most promising assays platforms, the lateral flow assay (LFA) platform has advantages, providing point-of-care and on-site monitoring extensively established for the detection of various analytes. As shown in

Figure 4A, an avian influenza virus (AIV) assay has a LOD 10-times lower than that of the commercial gold nanoparticle-based AIV LFA kit, even with opaque samples32. The LOD of the AIV LFA kit for low pathogenicity avian influenza (LPAI) H5N2 and high pathogenicity avian influenza (HPAI) H5N6 viruses was 102 and 103.5 EID50/mL, respectively32. Recently, a microplate assay based on UCL has been developed to detect trace amounts of analytes in fluid samples. For example, an assay for the biomolecule alpha-fetoprotein, a biomarker of liver cancer, has been developed with a dynamic working range of 0.01–40 ng/mL and a LOD of 5.3 pg/mL33. The UCNP signal is based on enzyme-linked immunosorbent assays (ELISA). This method can be employed for other analytes that are commonly detected by conventional ELISA. For example, the micropollutant diclofenac was detected in solution with LOD of 0.05 ng/mL, using ELISA type based UCNPs signals34.

Figure 4
figure 4

Upconversion nanomaterial based heterogeneous and homogeneous assay (A) Lateral flow immunoassay for detection of AIV32, (B) Competitive immunoassay for detection of progesterone38. Images reproduced from the references with permission.

Homogeneous Assay

In contrast to heterogeneous methods that may require laborious separation or washing steps, homogeneous assays do not require substrate immobilization or washing steps and are performed by a simple mix and readout procedure directly performed in the liquid phase. The homogenous assay is commonly based on the distance-dependent energy transfer (e.g., FRET or LRET) between the energy donor and the energy acceptor35. To achieve a sensitive assay with a low LOD, the factors to be considered for efficient energy transfer are the distance between the donors and acceptors and spectral overlap between the emission of the donor and the absorption of acceptors. In a typical LRET-based homogeneous assay, upconversion nanomaterials are first excited by NIR irradiation and then non-radiative energy transfer occurs to the spectrally overlapping acceptors, leading to the luminescence quenching of UCL emission signal. The LRET-based homogeneous assays have greatly improved LOD and use a simple readout process36,37. As shown in Figure 4B, progesterone was detected by the sensor platform based on a NIR-to-NIR signal LRET system via competitive immunoassay. The LOD was calculated to be 1.40 pg/mL, which is 35 times lower than that of commercial progesterone ELISA kit38. The LRET-based homogeneous assay has further been used for the detection of heavy metals Hg2+ and Pb2+ in aqueous solutions and to successfully detect pH changes3942.

Diagnosis and Treatment

Biomedical Imaging

Upconversion nanomaterials play a critical role in biomedical techniques, such as monitoring and treatment as well as MRI and CT imaging due to the upconversion process where NIR photons are converted into higher energy photons like visible emission. Upconversion nanomaterials have the advantages of low toxicity, deep light penetration and low autofluorescence. Lanthanide-doped (Gd3+/Yb3+/Nd3+/Er3+) UCNPs can be used as multifunctional contrast agents with magnetic, optical and x-ray properties43. Yb3+- doped UCNPs have been widely explored but applications are limited due to rising of local temperatures and overheating, which could induce undesired death or injury to a normal cell. To resolve these limits, Nd3+ has been employed as a photosensitizer (PS) for reducing the overheating, and to adjust the excitation wavelength at 980 nm to a more biocompatible wave-length of 808 nm (Figure 5A)44,45. In recent research, Gd3+ has been used in UCNPs to enhance the T1-weighted MRI signal and Dy3+ has been used as a contrast agent for T2-weighted MRI and enhancing CT imaging due to high k-edge values (Figure 5B)43,46. Further, Sun et al. reported in vivo multimodal imaging (UCL, MRI, CT) using NaGdF4:Yb,Er UCNPs and it showed strong contrast at the tumor site46. By synthesizing appropriate lanthanide-doped UCNPs, multimodal imaging can offer highly accurate information and image-guided therapy44.

Figure 5
figure 5

(A) Schematic of multimodal MR/Upconversion luminescent imaging of hepatocellular carcinoma (HCC) tumor using transcatheter hepatic intra-arterial (IA) targeted UCNPs44, (B) T2-weighted MR images acquired before and after UCNP injection and T2 maps (insets) of HCC tumor regions; (right) intraoperative upconversion luminescent imaging44. Images reproduced from the references with permission.

Therapeutics

The non-invasive phototherapy is mainly divided into photodynamic therapy (PDT) and photothermal therapy (PTT). PDT can produce reactive-oxygen species (ROS) by activating the PS47. The penetration ability of UV-visible light is limited, but NIR light has a deeper-penetrable ability. So, the UCNPs absorb NIR light and emit UV-visible light, which can activate the PSs to produce cytotoxic ROS48. The generation of ROS can cause cancer cell apoptosis or necrosis46,47. Also, ROS have been implicated in Alzheimer’s disease. For example, Lee et al. reported that NaYF4:Yb,Er UCNPs emit visible light under the NIR irradiation and photoexcited rose Bengal (RB) as a PS generates ROS, which can suppress the aggregation of β-amyloid (Aβ)48. In PTT, NIR light is converted into localized heat energy, inducing cellular hyperthermia46. PTT relies on photothermal agents to generate heat and increase temperature upon NIR light irradiation49. Various NPs, such as carbon nanomaterials, transition metal sulfides and oxides, have been proposed as photothermal agents. In particular, plasmonic nanomaterials can be easily used in PTT due to the localized surface plasmon resonance. For example, Sun et al. reported that gold nanorod (NR) dimers serve as strong PTT agents and the UCNPs serve as a PDT agent. The tumor treatment with only PTT or PDT is not enough to induce cell death completely. However, the combination of PTT and PDT can provide superior effects like sufficient or complete tumor elimination46.

Solar-to-Chemical Energy Conversion

Water Splitting

In terms of energy, sunlight reaching the Earth’s surface is approximately 53% IR (above 700 nm), 42% visible (400 to 700 nm), and 3% UV (below 400 nm)50. Solar-to-chemical energy conversion is one of the ultimate goals for researchers because it can allow sustainable production and the carbon-neutral use of chemicals51. Therefore, photon upconversion of infrared light is a promising approach to efficient light harvesting. Kim and co-workers fabricated a photoe-lectrochemical water splitting device using a versatile layer-by-layer (LbL) assembly method52. UCNPs, Ag nanoparticles, and polyoxmetalate catalysts have been used for NIR light emission, plasmonic effects, and water oxidation catalytic activity, respectively (Figure 6A). Figure 6B shows that the upconversion emission was significantly improved after the deposition of LbL-assembled hybrid films, which synergistically enhanced the performance of water oxidation devices. The photocurrent density at 1.23 V versus reversible hydrogen electrode under 980 nm NIR light irradiation was improved by about five fold after the LbL deposition.

Figure 6
figure 6

Solar-to-chemical energy conversion by upconversion luminescence process. (A) Upconversion nanoparticle-based photoelectrochemical water splitting device52, (B) Photographs of upconversion nanoparticle complex52, (C) Cofactor regeneration under NIR illumination54. Images reproduced from the references with permission.

Cofactor Regeneration

Green plants use nicotinamide cofactors [NAD(P)H], redox chemicals carrying electrons, as a reducing power for conversion of carbon dioxide to carbohydrates in the Calvin cycle51. Therefore, nanobiocatalysis-based artificial photosynthesis aims to reassemble man-made PSs, electron mediators, electron donors, and redox enzymes for photochemical synthesis of valuable chemicals through cofactor regeneration process53. Figure 5C shows NIR-light-driven biocatalytic artificial photosynthesis using efficient photon-conversion through FRET with silica-coated UCNPs as donor and RB as acceptor54. NIR-excited electrons are transferred to an electron mediator (i.e., a rhodium(III) catalyst) through FRET and reduced NAD+ to NADH, resulting in the synthesis of l-glutamate by cofactor-dependent l-glutamate dehydrogenase. The photoenzymatic conversion of substrate to product is facilitated with upconversion complexs through NIR light-driven photochemical cofactor regeneration. The results show that UCNPs are promising light harvesters for NIR light-to-chemical conversion processes.

Security Printing

Anti-Counterfeiting

UCL-based ink is promising as an anti-counterfeiting application, since UCL ink is invisible under ambient conditions and consequently difficult to duplicate or fake in contrast to organic dyes55. A QR code can hold more data than the traditional barcode but is not suitable for security applications. For anti-counter-feiting applications, invisible QR codes can be printed using UCL ink that can be protected by coating with an opaque protective layer without adversely affecting functionality56. Hiding NIR luminescent letters or symbols in a color print enhances the security level of color prints. UCL ink can be printed onto various surfaces such as paper or plastic substrates. Prints of letters “UCNP” and a football are hidden into a non-luminescent color print of a butterfly, enhancing the anti-counterfeiting strength57.

Latent Fingerprinting

Finger marks are considered as one of the most valuable types of physical evidence for effective identification at crime scenes. Conventional latent fingerprinting methods that have been achieved using metallic, magnetic or fluorescent powders are ineffective on certain substrates like wet objects. To improve contrast, sensitivity, and selectivity, UCL-based latent fingerprinting was developed using NIR-to-visible UCNPs (NaYF4:Yb,Er)58. Fingerprints on a variety of substrates can be performed without necessitating a background signal noise interference and the minute features, including sweat pores can be observed clearly; this is because of NIR excitation and nanosized UCNPs. In addition, latent fingermarks aged for up to 1 year can be developed with clear ridges59. For effective fingerprint imaging in full room light, NIR-to-NIR UCNP can be used, which is significantly brighter than NIR-to-visible UCNP (Figure 7). Images obtained at 800 nm suffer less from the background signal noise interference on substrates; this is because numerous inks cannot absorb at the 800 nm emission wavelength of NIR-to-NIR UCNP60.

Table 1 Upconversion nanomaterial-based detection tools for numerous analytes.
Figure 7
figure 7

Latent fingerprints developed with NIR-to-NIR (top row) and NIR-to-visible (bottom row) UCNP60. Images reproduced from the references with permission.

Conclusions and Future Perspective

Over the past decades, a variety of studies has been identified not only efficient synthetic approaches that have enabled the synthesis of high quality UCNPs but also upconversion optical properties such as a large anti-Stokes shift, narrow emission spectra and long excited-state lifetime. In this review, we highlighted the overview of UCNPs and their recent applications. Contrast to the traditional fluorescent materials (e.g., organic dyes and quantum dots), the UCNPs overcome the traditional challenges such as photo-bleaching, surface modification and toxicity. Consequently, due to the unique optical properties of UCNPs (especially less autofluorescence and deeper penetration upon NIR irradiation), UCL-based platforms have turned out to be a promising tool for detecting analytes, bioimaging, and therapies. In particular, the NIR-to-NIR signal based detection system called the diagnostic window (650 ~ 800 nm) was able to detect signals in opaque samples and serum-based samples. Furthermore, the utilization of NIR has broadened their applications in energy harvesting and conversion as well as security printing. However, despite the great potential of UCNPs, several challenges exist that must be addressed. 1) The relatively low quantum yield of UCNPs attributes to their low extinction coefficient (an intrinsic feature of 4f optical transitions in lanthanide dopant ions). Therefore, UCNPs only play a secondary role in the solar energy conversion. 2) The UCNPs are generally excited under 980 nm irradiation owing to the overlapping with the absorption of sensitizer Yb3+ ions. However, it well-matched with the absorption of water, which could cause the overheating effect in biological environments. In the future, systematic studies to overcome these limitations of UCNPs can be a great stimulus for the next generation of luminescence-based applications.