INTRODUCTION
In recent years, the cryogenic technology has played a vital role in the field of astronomical observation1, energy2, quantum computing3-9, sensing10-12. and some fundamental researches of the condensed-matter physics13-18, which leads to a research focus of cryogenic refrigeration. Besides, the requirement for the cryogenic technologies of different fields is different, thus appropriate cooling method should be selected according to the demands. In the field of the astronomical observation, the space detectors with high observation accuracy need to operate at the liquid helium temperature zone or even the milli-Kelvin temperature zone so as to overcome the interference of spurious thermal vibration19. In the field of the quantum information, quantum computing could be used to solve computational problems more efficiently than classical computers, utilizing the quantum bits (qubits) operating at the temperature close to 0 K for the purpose of preventing quantum decoherence4-7. In the field of energy storage, the development of electrolytes and active materials in temperature at minus tens of degrees Celsius could overcome the shortcomings of the lithium-ion batteries in low temperature2. In the fundamental physics research, many novel physical phenomena can only be observed at very low temperatures, such as superconducting17, Bose-Einstein condensation16, superfluid in liquid helium14 and so on. Hence, the development of cryogenic technology is of great significance for the progress of science and technology.
In this review, we concentrated on the cryogenic refrigeration technologies for cryogenic electronics. Superconductor electronics, which is functional electronic circuits including active nonlinear and linear elements that are superconducting below the critical temperature, is an important part of cryogenic electronics20. The cryogenic superconducting circuits could exhibit unsurpassed performance that cannot be achieved by conventional semiconductor electronics, but the practical uses are limited by the cooling to and operating stably at cryogenic temperatures21. Besides, cryogenic semiconductor electronics is also a vital part of the cryogenic electronics22. There are many advantages of cryogenic semiconductor electronics, compared with that in room temperature, such as higher operational speed, lower power dissipation, shorter signal transmission time and increased integration density etc23. The cryogenic electronics is endowed with its unique advantages when applied in the quantum information processing since it uses qubits and two-state quantum-mechanical systems that can be in coherent superpositions of both states at the same time, which would improve the computing speed greatly24.
For the millikelvin temperature environmental implementations, dilution refrigerators and adiabatic demagnetization refrigerators are two main conventional candidates19,25-28. The dilution refrigerators are used widely in the quantum technology4-7,9,29, which show the advantages of working continuously and high efficiency, while it is gravity-sensitive and needs the expensive helium. The adiabatic demagnetization refrigerators have become the mainstream ultra-low temperature refrigeration technology for space applications30, which are endowed with the outstanding advantages of gravity-independent, not relying on the scarce helium as well as compact and high efficiency. However, they also have some disadvantages of the requirement for magnetic field, working discontinuously. Besides, room temperature magnetic refrigeration is also a highly efficient and environmentally protective developing technology31-33. Superconducting junction refrigeration is an emerging subkelvin refrigeration method34-41, which can be integrated with the cryogenic quantum chips. Besides, thermoelectric refrigerators42,43 are suitable candidates for the small and portable refrigerators. In recent decades, optical refrigeration, which mainly includes laser cooling of semiconductors and rare-earth-doped solids, has attracted more and more interest due to its advantages of small size, no vibration, no refrigerant and so on44-51. The application of laser cooling of rare-earth-doped solids in radiation balanced lasers is gradually expanding. The special energy level structure of rare-earth ions50-52 makes them excellent cooling materials. By using high-efficiency optical refrigeration and controlling parameters such as laser wavelength, power and scanning rate, radiation balanced lasers can be achieved52,53. Besides, laser cooling of semiconductors has been realized experimentally in the past decades, which will have a wide range of applications in fields such as infrared detection chips54, quantum information processing4,6, on-chip lasers55,56, and more. In this review, refrigeration technologies were categorized into non-optical refrigeration techniques and optical refrigeration techniques, as shown in Table 1, and a brief introduction was provided to their principles, advantages and disadvantages, current research status, as well as perspectives.
Table 1. A comparison of different refrigeration technologies. |
| Refrigeration technologies | The lowest temperature | Cooling power | Cost | |
|---|---|---|---|---|
| Non-optical refrigeration technologies | Dilution refrigeration | ∼ 1.75 mK57 | 20 µW @10 mK57 | High |
| Adiabatic demagnetization refrigeration | ∼ 22 mK from 2 K136 | NA | High | |
| Superconducting tunnel junction refrigeration | ∼ 30 mK from 150 mK 35 | ∼ 40 pW35 | High | |
| Thermoelectric refrigeration | ∼ 120 K135 | NA | Low | |
| Optical refrigeration technologies | Laser cooling of the rare-earth doped solids | ∼ 91 K134 | ∼ 50 mW134 | Low |
| Laser cooling of semiconductors | Net cooling by 40 K from 290 K47 | ∼ 180 mW @514 nm47 | Low | |
SEVERAL NON-OPTICAL REFRIGERATION TECHNOLOGIES
Dilution refrigeration
The dilution refrigeration is a kind of subkelvin refrigeration with relatively large cooling power, which uses circulating 3He to produce cooling, and it has been widely used in the fields of the high-sensitivity sensors, quantum information technology, and fundamental physics researches of the quantum mechanical effects in superfluids, superconductors and semiconductors13-17. Dilution refrigeration utilizes the properties of 3He-4He mixtures for realizing cooling. The 4He atoms in the mixture transform from helium I into superfluid helium II below 0.7 K since the 4He atoms are bosons and obey Bose-Einstein statistics60. The 3He atoms in the mixture are fermions and obey Fermi-Dirac statistics, thus the 3He-4He mixture can be regarded as a weakly-interacting Fermi gas and the superfluid 4He only behaves as an inert superfluid background. When the temperature is below 0.87 K, the 3He-4He mixture can be separated into a concentrated phase and a dilute phase of 3He, and the concentrated phase floats on top of the dilute phase. The concentration of 3He in the dilute phase is 6.6% at saturated vapor pressure when the temperature is close to 0 K, while the concentrated phase is almost pure 3He, which allows the dilution process to occur at very low temperatures25,61. Hence, the dilution refrigerators can reach temperatures below several millikelvin.
The structure of the typical wet dilution refrigerator is shown in Fig. 1b60, it is composed of a 3He circulation pump at room temperature, a 4 K helium bath, an 1 K bath, a main impedance, a heat exchanger to the still, a secondary impedance, a still, heat exchangers, and a mixing chamber. The minimum temperature achieved by the wet 3He-circulating dilution refrigerator is 1.75 mK until now57, which is reported by Cousins et al. in 1999. The advances of Gifford-McMahon (GM) cryocooler lead to the removal of the 4 K liquid helium bath20,62,63. In 1990, P. Pari invented the dry dilution refrigerator (Fig. 1c), where the 4 K liquid helium bath was unnecessary64. In the dry dilution refrigerator, a two-stage Gifford-McMahon cryocooler combined with a 4He Joule-Thomson refrigerator is used to cool the incoming 3He to 4 K. However, Gifford-McMahon cryocoolers introduce a lot of mechanical vibrations due to the moving mass of the regenerator. In 2000s, Gifford-McMahon-type plus tube refrigerators with lower mechanical vibration were applied in the dry dilution refrigerators and provided the pre-cooling temperature around 4 K. The commercial dry dilution refrigerators are provided by many companies, such as Leiden Cryogenics, Oxford Instruments and Bluefors etc., which is widely used in the investigation of the cryogenic CMOS chips (Fig. 1a)4. However, the disadvantage of dry dilution refrigerators is also obvious, that is, the size of dry dilution refrigerators is bulky, which may limit the applications of the cryogenic CMOS chips. Hence, reducing the size of the refrigerators may be the future research directions.
Fig. 1. Dilution refrigerators. a, Left panel: The location of the cryogenic quantum chips inside the dilution refrigerator and the position of the cryo-controller. Right panel: The top and bottom views of the 3-K plate, showing the mounted chip enclosure and the fixed holder for the enclosure, respectively. Reprinted with permission from ref.4. © 2021 Nature Publishing Group. b, Schematic diagram of the low-temperature part of a wet dilution refrigerator (the dilution unit). c, Schematic diagram of a dry dilution refrigerator with two-stage pulse tube refrigerators. Reprinted with permission from ref.60. © 2022 Elsevier. |
Adiabatic demagnetization refrigeration
Adiabatic demagnetization refrigeration is based on the magnetocaloric effect, which is the reversible temperature change of a magnetic material when a magnetic field is applied or removed27,28,31,32,58,65-74. In order to simply illustrate the typical magnetic refrigeration cycle, the process was separated into four stages, as shown in Fig. 2a. In the first stage, when adiabatically applying a magnetic field, the orientation of the magnetic moments are in parallel with that of the magnetic field, leading to the reduction in the magnetic entropy which is compensated by an increase in lattice entropy, and an increase in the temperature of the lattice to T+δTad. In the second stage: a heat transfer fluid is employed to cool the magnetocaloric material back to its initial temperature of the first stage. In the third stage: the magnetic field is removed adiabatically, the magnetic entropy of the sample decreaseswith a decrease in temperature of the lattice decreases to T-δTad. In the fourth stage: heat is extracted by passing a heat transfer fluid through the magnetocaloric materials so as to cool the contents in the refrigerator. Therefore, a magnetic refrigerator can be constructed by conducting the above four steps cyclically. Consequently, it can be concluded that the magnetocaloric materials are of vital significance for the magnetic refrigeration from the cooling process. There are several families of the magnetocaloric materials, including crystalline materials containing rare-earth metals, rare-earth-free crystalline materials, amorphous alloys, and nanostructured materials. The maximum magnetic entropy changes for ΔH = 5 T versus peak temperature for different families of magnetocaloric materials is shown in Fig. 2b32. In order to achieve further development of magnetic refrigerators, it is of great significance to conduct further design and growth of the magnetocaloric materials with high magnetic entropy change. Compared with other refrigeration technologies, the magnetic refrigeration exhibits several advantages as follows. Firstly, the magnetocaloric materials is environmentally friendly. Secondly, the magnetic entropy density of magnetocaloric material is higher than that of refrigerant gas and no compressors is needed. Thirdly, the coefficient of performance (COP) value of the magnetic refrigerators could reach 1.8 at room temperature and the magnetic refrigeration efficiency can be 30 to 60% of Carnot cycle efficiency, while the efficiency of the conventional refrigerators is only 5 to 10% of that. However, there are also some disadvantages that needs to be overcome65. Firstly, the magnetic field strength of permanent magnets is restricted, and the price of electromagnets and superconducting magnets is high. Secondly, the magnetocaloric materials for rectilinear and rotational magnetic cooling at higher frequencies remain to be further developed.
Fig. 2. Adiabatic demagnetization refrigeration. a, The four processes of a magnetic refrigeration cycle: adiabatic magnetization, remove heat, adiabatic demagnetization, and cool refrigerator. b, The maximum magnetic entropy changes for ΔH = 5 T versus peak temperature Tpeak for different families of magnetocaloric materials. Reprinted with permission from ref.32. © 2012 Annual Reviews Inc. |
In conclusion, adiabatic demagnetization refrigeration technology is an environmentally friendly, high-efficiency and promising refrigeration technology, which promises wide applications in the cooling of the electronics, food industry, medicine, and so on.
Superconducting tunnel junction refrigeration
Refrigeration based on superconducting junctions is a solid-state refrigeration for sub-Kelvin temperatures34-41,75-81, which is an urgent demand for the emerging quantum technology, such as quantum information processing, quantum communications, quantum sensing and so on. The basic element of the superconducting junction is composed of a normal-insulator-superconductor (NIS) junction, the energy level diagram of which is shown in Fig. 3a79. The basic cooling process can be depicted as follows: the electrons in the normal electrode are cooled by the tunneling of electrons through a biased NIS junction, which is resulted from the preferential tunneling of the hot electrons to the superconducting electrodes. In this process, the micron-sized superconducting tunnel junction coolers can cool an external object at temperatures below 300 mK, which is realized by bulky macroscopic adiabatic demagnetization or 3He-4He dilution refrigerators traditionally79. A device micrograph of the superconducting tunnel refrigeration system is shown in Fig. 3b, where the NIS refrigerator is grown on the thermally isolating membrane made of suspending SiNx/SiO2 island supported by eight legs. In order to realize enhanced cooling, the structure of the NIS refrigerator is composed of two symmetrically placed NIS junctions, forming an NISIN double junction, as shown in Fig. 3c, which was first proposed by Leivo et al. and is mostly used today82. The cooling result is shown in Fig. 3d, which shows 34 mK-cooling of a Cu stage by quantum-mechanical tunneling of electrons through the biased superconducting tunnel junctions. Recently, a stand-alone, on-chip quantum-circuit refrigerator, which can locally cool superconducting circuits based on photon-assisted tunneling using a pair of parallel NIS junctions, has been invented, it has potential applications in the initialization of quantum on-chip electric devices38. In addition, a single-junction quantum-circuit refrigerator based on photon-assisted quasiparticle tunneling has been proposed, which could simplify the quantum-circuit refrigerator39. Hence, superconducting tunnel junction refrigeration is attractive for the emerging quantum cryochips sicne it is small and convenient to be integrated.
Fig. 3. Refrigeration based on superconducting tunnel junctions. a, Energy level diagram of the NIS junctions. Red area: occupied states; green states: available states. b, Micrograph of a 370nm thick membrane cooled by NIS junctions. c, Cross-sectional device schematic showing critical nanoscale dimensions. d, The measured temperature of the suspended Cu stage plotted versus time in hours. Reprinted with permission from ref.79. © 2013 American Institute of Physics. |
Thermoelectric refrigeration
Thermoelectric refrigeration is based on the Peltier effect to convert electrical energy into a temperature gradient43,83-95. When current flows across an interface, heat is absorbed (cooling) or rejected (heating) owing to the difference in the thermal energy transported by charge carriers (electrons or holes) in the n-type junction and p-type junction, which is the basic principle of the Peltier effect. The solid-state thermoelectric coolers without refrigerants are much more environmentally friendly. Additionally, thermoelectric coolers have been widely used in the portable and small-cooling-power refrigerators for cooling detectors, carrying medical supplies, and temperature control of car seats etc. The basic element of thermoelectric refrigerator consists of an n-type junction, a p-type junction, and conducting strips (in general made of copper) which connect the junctions in series, as shown in Fig. 4a. Once a low-voltage DC power source is applied to a thermoelectric cooler, heat is transferred from one side of the thermoelectric cooler to the other side. Hence, one side of the thermoelectric cooler is cooled, while the opposite side is heated. There are several parameters that determine the quality of the thermoelectric cooler, such as the electric current, the temperature of the cold and the hot side, the electrical contact resistance, the thermal and electrical conductivities and so on. The figure of merit Z and the COP are both important parameters for the performance of the thermoelectrical cooler. The figure of merits Z quantifies the thermoelectric performance of an ideal unicouple, which can be measured, and COP can be quantified by the amount of cooling divided by the electrical power input. The COP is dependent on the temperature difference between the cooling-side temperature Tc and the heating-side temperature Th, as shown in the calculated solid curves and experimental symbol of the Fig. 4b96, where, ZTm is the average of the ZT value between Tc and Th. As shown in Fig. 4b, compared with refrigerators and air conditioners based on the vapour compression cycle, thermoelectric cooling systems exhibit much lower COP values96. In order to compete with the conventional refrigerators, it is of great significance to improve the ZTm value of the thermoelectrical cooler, whereas it is challenging to be realized in the foreseeable future. Hence, instead of being regarded as a competitor of the conventional methods, thermoelectric cooling could be considered as a complementary technology.
Fig. 4. Thermoelectric refrigeration. a, Schematic diagram of a thermoelectric refrigerator. b, A comparison of the COP between thermoelectric cooling systems (red symbols) and vapor compression refrigerators (blue symbols). The solid lines are calculated by assuming that the hot-side temperature is 300 K, and the COP is calculated using constant ZTm values. Reprinted with permission from ref.96. © 2021 Nature Publishing Group. |
EMERGING OPTICAL REFRIGERATION TECHNOLOGIES
The fields of laser cooling of solids
Laser cooling of condensed matter was first proposed and predicted by Peter Pringsheim in 192997 and it can be realized by optical irradiation accompanied by spontaneous anti-Stokes emission, where the average energy of the photons emitted by the solid is larger than that of the ones it absorbs98. In addition, the nonradiative decay rate of the laser-pumped states is negligible compared with their radiative decay rate, which is also a crucial additional requirement. A typical system of laser cooling is composed of a ground-state manifold and an excited-state manifold well separated from one another, with at least one of these manifolds split into two or more levels, such as rare-earth ion doped solids, as shown in Fig. 5b99,100. The first experiment on laser cooling of rare-earth ion doped solids was demonstrated by Epstein's research team in 1995101, where a temperature drop of 0.3 K below room temperature was measured for a 43-mm3 rare-earth doped heavy-metal fluoride glass sample in the shape of a rectangular parallelepiped. In 1999, Steven Bowman proposed to use the optical refrigeration within the laser medium to balance the heat generated by the Stokes shifted stimulated emission in a high-power solid-state bulk laser named radiation-balanced or athermal laser102, and it was realized experimentally in 2002104. In 2013, Jun Zhang and his colleagues first experimentally demonstrated the net laser cooling of the cadmium sulphide semiconductor nanobelts due to the strong coupling between excitons and longitudinal optical phonons47. At present, the fields of laser cooling of solids can be roughly divided into three main areas: laser cooling of rare-earth doped solids50,104-106, laser cooling in semiconductors46,47,107,108 and radiation-balanced lasers52,53,103.
Fig. 5. Principles of optical refrigeration of rare-earth ion doped solids. a, Three kinds of main energy engaged in the process of the optical refrigeration of rare-earth ion doped solids. 1: Photons from the pump laser; 2: Phonons from the host crystal; 3: Electrons of the doped ions. b, Energy levels and major transitions of Yb3+ doped in ZBLAN. c, The four-level energy model for optical refrigeration consisting of two pairs of levels in the ground states (E0 and E1) and excited (E2 and E3) states. Reprinted with permission from ref.99. © 2007 The American Society of Mechanical Engineers (ASME). |
Laser cooling of the rare-earth doped solids
The fundamental energy carries involved in the laser cooling of rare-earth doped solids include photons from the pump laser, phonons from the host crystal and electrons of the doped ions, as shown in Fig. 5a99. The systems of laser cooling of rare-earth doped solids consist of a host crystal lattice which can be seen as transparent to the pumping laser ideally, and optically active doped rare-earth ions, such as Yb3+51,105,106,109. In this system, the ion can be represented by an effective transition dipole moment μe, which can be thought as the coupled states of the initial state ψi and the final states ψf. The electromagnetic field may interact with effective transition dipole moment on the condition that the polarization vector of the electromagnetic field eα and the effective transition dipole moment μe are not orthogonal. The diagram of the energy of the of 4d electrons in a rare-earth ion Yb3+ is shown in Fig. 5b, and the absorption occurs at essentially the same energy as the corresponding emission in the case of the zero Stokes shift, where the laser cooling of the solids couldn't be realized. The principles of the laser cooling of the rare-earth doped solids can be depicted in the frame of the four-level model, as shown in Fig. 5c, which is generally applicable to any such two-manifold cooling system99. The model includes a two-level ground state with a separation that denotes the width of the ground-state manifold, and a two-level excited state with an energy splitting corresponding to the width of the excited state manifold. The electrons in the level E1 are assumed to be excited to the level E2 by the laser excitation. The electron-phonon interaction of the 4d electrons establishes Boltzmann quasi-equilibrium on a relatively fast timescale. The excitation then decays into ground state by radiative and nonradiative relaxations with the corresponding rates of Wr and Wnr, respectively. According to the following assumptions, the equations of the density populations N0, N1, N2 and N3 corresponding to the energy E0, E1, E2 and E3 can be obtained. Besides, equal degeneracy for all of the four levels is assumed. Then the cooling efficiency ηc of the system can be obtained:
where, ηq = ηeWr/(ηeWr + Wnr) is the external quantum efficiency, which is equal to the internal quantum efficiency for unity fluorescence extraction efficiency ηe, ηabs denotes the absorption efficiency, h is Planck's constant, ν and νf are the frequency of the pump photons and the scattering photons, respectively. Besides, we can get that the temperature-dependence of cooling in a physically transparent manner in the frame of the four-level system. The cooling efficiency decreases with lowering temperature, which is mainly ascribed to the red-shifting of the mean fluorescence wavelength and the reduction of resonance absorption. The minimum achievable temperature, where ηc=0, is set to Tm, can be achieved by means of improving purity (decreasing the background absorption), increasing the quantum efficiency, and enhancing the resonant absorption and choosing a material with a narrow ground state manifold. To improve the cooling efficiency of the laser cooling of the rare-earth doped solids, further investigations on the above methods or challenges remain to be conducted.
The Yb3+: YAG crystal is one of the most widely adopted active media in the solid-state lasers of high power due to its excellent mechanical, thermal and optical prosperities105,106. However, the heat deposition in the active medium damages the beam quality of the laser, and limits the maximum laser power available. Laser cooling of the rare-earth solids by anti-Stokes fluorescence provides a good method to solve this problem. However, the previous experimental work reported the laser cooling of Yb3+: YAG crystal with a little temperature drop110, which needs further investigation to achieve lower temperature. In 2021, Biao Zhong et al. experimentally demonstrated the optical refrigeration of the Yb3+-doped YAG crystal close to the thermoelectric cooling limit105. The experimental setup is shown in Fig. 6a. The 3% Yb3+: YAG crystal sample with the size of 2 × 2 × 5 mm3 was placed inside a vacuum chamber of pressure about 1 × 10-4 Pa and irradiated by a fiber laser beam with a varying wavelength from 1010 to 1080 nm. The absorption spectra (blue line) and emission spectra (red line) of 3% Yb3+: YAG crystal at the temperature of 300 K is shown in Fig. 6b, which indicates that the mean fluorescence wavelength is 1011.6 nm and the cooling tail is at 300 K, respectively. As shown in Fig. 6c, the overall cooling effect was the best when the sample was pumped at a wavelength of 1030 nm, where the absorption and the fluorescence emission both exhibit the local maximum intensity. Besides, the cooling window of the 3% Yb3+ was obtained: YAG crystal with simulation based on their experimental results, indicating that crystal sample can be potentially cooled to the temperature of about 180 K at a pump wavelength of 1030 nm, providing a good method to investigate high-power on-chip laser55,111,112. In this section, the laser cooling of Yb3+was mainly reviewed: YAG crystal. In addition, the optical cryogenic refrigeration has been obtained on the other rare-earth-doped crystals, such as Yb3+: LLF113-116, Yb3+: YLF117-119 and so on.
Fig. 6. Progress of optical refrigeration of rare-earth ion doped solids. a, Experimental schematics of the single-pass optical cooling setup. b, Emission spectra (red line) and absorption spectra (blue line) of 3% Yb3+-doped in YAG crystal at 300 K, in which the dash line and the blue shaded region denote the mean fluorescence wavelength and the cooling tail, respectively. c, The temperature dependence on the pump power of the laser 1030 and 1048 nm, respectively. d, The cooling window of the 3% Yb3+-doped YAG crystal, which depicts the dependence of cooling efficiency on the temperature and pump wavelength. The blue and red regions denote the cooling and heating regimes, respectively. Reprinted with permission from ref.105. © 2021 American Institute of Physics. |
Laser cooling of semiconductors
Laser cooling of semiconductors is more interesting than that of the rare-earth doped solids and has attracted much more attention due to the fact that semiconductor coolers are endowed with the advantages of providing more efficient pump-light absorption, the potential for much lower temperatures, and the opportunity for direct integration into electronics and photonics120-125. The main difference between the two cooling methods is the cooling cycles. In the semiconductors, the cooling involves transitions between extended valence and conduction bands of a direct-bandgap semiconductor (Fig. 7), while the cooling transitions in the rare-earth doped solids occurs in the localized donor ions within the host material (Fig. 5c). Hence, the cooler of semiconductor exhibits more advantages than that of rare-earth doped solids. At first, the maximum cooling power density can be decided by NkBT/τr, where N is the density photo-excited carries, which is comparable between the two systems, and τr is the radiative recombination time100. However, the radiative recombination rates in semiconductors are much faster than those in the rare-earth doped solids, thus five to six orders of magnitude in cooling power density can be gained. In addition, the indistinguishable charge carries in the semiconductors (electrons and holes) enable semiconductors to be much colder than rare-earth doped solids, this is mainly ascribed to the fact that the highest energy levels of the ground-state manifold in the rare-earth doped systems tend to be less populated with decreasing temperature according to Boltzmann statistics, leading to the lowest temperature of 60 K achieved by laser cooling in this system, whereas such limitation does not exist in the pure semiconductors59,100.
Fig. 7. Schematic diagram of the process for laser cooling of a semiconductor where a pump laser photon with energy ℏωp is absorbed followed by emission of an up-converted fluorescence photon with energy ℏωf (corresponding to the process 1). Process 2: Electrons and holes undergo non-radiative recombination and multiphonon emission happens. |
There are many theoretical investigations on laser cooling of semiconductors, including the microscopic theory for laser cooling which provides valuable predictive capabilities for low-temperature operations, conditions for net cooling based on material properties and light management, some possible enhancements of laser cooling of semiconductors and so on44,126-130. However, the experimental results on laser cooling of semiconductors are limited. In 1999, E. Finkeißen and his colleagues first reported the thorough experimental effort on laser cooling of semiconductors, while no net cooling was realized in spite of the realization of an external quantum efficiency of 96% in GaAs131. In 2013, Jun Zhang et al. reported a net laser cooling of CdS nanobelts by 40 K46,47. The anti-Stokes photoluminescence spectra of a CdS nanobelt is shown in Fig. 8b, strong anti-Stokes fluorescence with a peak position of 506 nm could be identified, which was facilitated by resonant annihilation of multiple longitudinal optical phonons (LOPs), and much stronger fluorescence up-conversion was achieved by the 514-nm pump laser owing to stronger band-tail absorption. A pump-probe luminescence thermometry (PPLT) technique was adopted to measure the laser cooling with the sample image, as shown in Fig. 8a. The CdS nanobelts were suspended on the hole-patterned SiO2/Si substrates to improve the luminescence extraction efficiency and decrease the thermal conduction. The pump laser was a solid-state, 532-nm laser and an argon ion laser (variously 514, 502 and 488 nm) and the pump beam was a solid-state, 473-nm laser to measure the cooling or heating. In this experiment, the temperature change ΔT caused by laser cooling was deduced by the peak shift of the photoluminescence spectrum of the CdS nanobelt. Fig. 8c shows the Stokes photoluminescence evolution on continuous 6.3-mW, 514-nm laser pumping starting from 290 K. Based on the temperature calibration curves, ΔT is plotted in Fig. 8d, showing net laser cooling of 40 K for 514-nm pump laser, and 20 K for 532-nm laser, both at 290 K. In the current work, the larger bandgap made the CdS exhibit a much smaller surface recombination velocity and lower Auger non-radiative recombination coefficient compared with GaAs. Consequently, this work provided a method to the design of materials for the laser cooling of semiconductors. In the subsequent work, Jun Zhang and his colleagues proposed the concept of exciton optomechanics (Fig. 8e, f, h)45,132,133, where the exciton resonance is regarded as the optical cavity in the frame of cavity optomechanics. When the decay rate of the optical mode is smaller than the frequency of phonons, the systems reach the resolved sideband regime and resolved sideband cooling can be realized. Fig. 8g and Fig. 8i show the resolved-sideband Raman cooling of LOP in ZnTe nanobelts and a van der Waals semiconductor WS2, respectively.
Fig. 8. Progress of laser cooling of semiconductors. a, Laser cooling pump-probe luminescence thermometry. The top panel: False color scanning electron microscope image of a single CdS nanobelt suspended on a SiO2/Si substrate. The bottom panel: Measurement set-up with pump and probe laser beams. b, Multiple-LOP-assisted up-conversion spectra of CdS nanobelts. c, Evolution of the pump-probe luminescence thermometry spectra starting from 290 K, pumped by a 514-nm laser with a power of, 6.3 mW. Solid curves: cooling cycle; dash curves: pump laser is switched off. d, Temperature change, ΔT, versus time pumped by four laser lines (532, 514, 502 and 488 nm), using data extracted from the pump-probe luminescence thermometry around 290 K. Reprinted with permission from ref.47 © 2013 Nature Nature Publishing Group. e, The measured PL spectra of exciton transitions and Raman spectra of LOP of 120-nm ZnTe nanobelts at 225K. f, Schematic diagram of the principle of the resolved-sideband cooling LOP with red-detuned laser and LOP phonon modes. g, Occupation number and the temperature of the LOP versus pump power. Reprinted with permission from ref.133. © 2016 Nature Publishing Group. h, Schematic diagram of the principle of the resolved-sideband cooling optical phonons in 2-demonsional van der Waals semiconductors. i, Occupation number and the temperature of the phonons versus pump power. Reprinted with permission from ref.45. © 2022 American Chemical Society. |
In conclusion, laser cooling of semiconductors is emerging as a promising and novel research field. To our knowledge, laser cooling of semiconductors could be only realized in the systems of 1D semiconductor nanobelts47, thus it is far from the preparation of semiconductor optical cooler, which needs further investigations. Besides, silicon-based semiconductors are the cornerstone of the field of integrated circuits, while it is difficult for silicon-based materials to emit light. Several works have realized emitting light in the direct band gap of silicon nanocrystals, thus laser cooling of the silicon-based semiconductors may be achievable, which promises important implications in the development of cryogenic chips.
CONCLUSIONS AND PERSPECTIVES
In cryogenic temperature, cryogenic electronics could exhibit unique characteristics and advantages which are not available in room temperature and it has great application prospects, such as quantum information processing. Hence, cryogenic environment is of vital importance for the development and applications of cryogenic electronics. In the current work, the developments status and prospects of the non-optical refrigeration were introduced, such as conventical dilution refrigerator, thermoelectrical refrigerator, magnetic refrigerator and the emerging superconducting tunnel junction refrigeration at first. Then, we made the introduction of the optical refrigeration, including laser cooling of rare-earth doped solids and laser cooling of semiconductors, later it was illustrated that optical refrigeration is a promising refrigeration technology for cryogenic chips in the future.
Dilution refrigeration and other non-optical refrigeration technologies will continue to be the mainstream cooling techniques in the field of cryogenic chips for the current and even the foreseeable future. However, these technologies suffer from drawbacks such as complex cooling processes and expensive equipment, which drives us to seek new types of refrigeration technologies, and optical cooling technology is an emerging and promising refrigeration technique that offers advantages such as small size, no vibration, no refrigerant, etc. Nevertheless, achieving a stronger anti-Stokes scattering than Stokes scattering of phonons in semiconductors is not an easy task due to tha fact that heating is inevitable, and net laser cooling of semiconductors by 40 K has been only achieved in 1D semiconductor CdS. Hence, new semiconductor material systems and structures need to be explored, and engineer densities of electronic and photonic states need to to achieve net cooling and it is desirable to cool larger volumes for practical applications in cryogenic chips. Although various emerging refrigeration technologies may employ different physical principles and operational mechanisms, they are still based on the fundamental principles of the Carnot cycle. By optimizing the processes within the cycle, and improving materials and techniques, these new refrigeration technologies can enhance cooling efficiency, reduce energy consumption, and provide a wider range of cooling toolboxes for cryogenic chips.
MISCELLANEA
Acknowledgments J. Z. acknowledges the National Natural Science Foundation of China (12074371), CAS Interdisciplinary Innovation Team, Strategic Priority Research Program of Chinese Academy of Sciences (XDB28000000). We thank C. Zhang for the discussion.
Declaration of Competing Interest The authors declare no competing interests.
