Technical Insight
Quantum dots and GaN dominate talks in Hawaii
GaN-based materials and quantum dots were two of the major talking points at ICMOVPE-XII, which was held in Maui, Hawaii. Richard Stevenson describes some conference highlights.
The resort of Lahaina, on Maui island, Hawaii, provided a magnificent setting for the 12th International Conference on MOVPE, held on 30 May - 4 June. GaN-based materials and devices were the main subjects of discussion, and the opening talk, given by George Craford, chief technology officer of Lumileds Lighting, considered the role of MOVPE in solid-state lighting.
Current MOVPE technology was demonstrated by dazzling the audience with a GaN-based LED torch, while Craford underlined the global impact of semiconductor light sources by highlighting the installation of LEDs in traffic signals in Cambodia.
Craford is optimistic about the technology and, although he believes that improvements must be made to both substrates and growth processes, he predicts that the high efficiencies offered by LEDs will fuel the increase of worldwide production levels to 50 billion GaN-based LEDs per year by 2010. This assumes a 10% penetration into traditional markets and GaN-based LED lifetimes of 10 years. Overall it would correspond to an annual production of 25 million 2 inch wafers, or 1000 reactors yielding 68 wafers each day.
UV-LED developmentsLater in the week John Edmond, director of optoelectronic technology at US-based Cree, described developments in deep-UV-LED sources. Edmond explained that solid-state sources emitting in the 270-340 nm region will enable technological advances such as fluorescence-based biological-agent detection; non-line-of-sight communications; water purification; and industrial processing, including ink drying and epoxy curing. In addition, Edmond described the benefits of optimizing device geometries to improve light extraction, reporting output powers of 1 mW at 20 mA and a peak wavelength of 340 nm.
An alternative method to improve light extraction - by introducing a corrugated interfacial substrate - was demonstrated by Samsung Advanced Institute of Technology, Korea. Improvements of more than 60% in extraction power, compared with conventional UV LEDs, were achieved using standard photolithography and reactive-ion etching.
High prices and the limited availability of GaN substrates have led to the fabrication of the majority of GaN-based devices on sapphire and SiC substrates (see Compound Semiconductor July 2004). Silicon substrates are attractive alternatives in terms of cost and scale, although they have inherent drawbacks. In particular, GaN epilayers can exhibit bowing and cracking due to the large thermal coefficient mismatch between silicon and GaN.
An alternative growth scheme was put forward by Matthew Charles from the University of Cambridge, UK. He suggested incorporating AlN interlayers, AlGaN graded layers, AlN/GaN superlattices and SiN into substrates to reduce tensile strain and dislocation density, thus allowing the growth of 2.3 µm-thick uncracked films of GaN on silicon.
Growing AlGaN/GaN HEMTs on silicon substrates was the subject of a talk given by Shiping Guo from Emcore Corporation, Somerset, NJ. Using Emcore s proprietary buffer technology, researchers fabricated a crack-free 2 µm-thick buffer layer on which they grew the HEMT structure.
Quantum dot (QD) structures and devices were another major talking point in Maui. Most groups grow QDs by strain-induced self-ordering using the Stranski-Krastanov (SK) growth mode, but a few researchers presented work involving prior growth patterning of substrates, leading to the formation of pyramidal structures with a dot defined at the very apex of the pyramid.
Improving QD lasersPelucchi Emanuele, from the Ecole Polytechnique Fédérale de Lausanne (EPFL), Switzerland, explained that although SK growth can produce very good interfaces, site control of the dots is difficult to achieve and there is a wide distribution in their peak emission wavelengths. Emanuele believes that ordered arrays of dots with reproducible and controlled excitonic states can provide opportunities for studying fundamental physics (for example, exploring entangled and correlated excitonic and photonic states) and deliver improved performances in QD lasers.
The EPFL group has patterned [111]B GaAs substrates with hexagonal arrays of tetrahedral recesses with a pitch ranging from 300 nm to 5 µm using electron-beam lithography. Subsequent growth of GaAs/AlGaAs followed by InGaAs enables the formation of pyramidal structures with quantum wires along the edges and a dot at the top of each structure emitting at around 800 nm (figure 1). The size uniformity of the dots is sufficient for the intrinsic broadening of light emission from a single dot to outweigh any contribution from fluctuations in the size of the dots. The EPFL group believes that this will enable the fabrication of QD lasers with all dots participating in the lasing process, giving a performance that could potentially surpass that of state-of-the-art quantum well lasers.
Tung-Po Hsieh, from the National Central University in Taiwan, reported strong photoluminescent (PL) emission at 1.55 µm from InAs QDs on GaAs substrates. He believes that this is the longest wavelength emission reported for InAs-on-GaAs dots grown by MOVPE. This result was achieved by inserting an In0.25Ga0.75As strain-reducing layer between the cap and the InAs QDs.
Jin Hong Lee from the Electronics and Telecommunications Research Institute of Korea reported on the tuning of emission from self-assembled InAs QDs, this time on InP substrates. Increasing the interrupt time after deposition of the InAs QD layer red-shifted emission from 1.48 to 1.66 µm and increased PL intensity. Blue-shifting emission from 1.48 to 1.38 µm was achieved by introducing a thin InAlGaAs layer on top of the InAs QDs.
The Korean team also fabricated ridge waveguide lasers using InP substrates, an InAlGaAs matrix and seven stacks of InAs QDs. This resulted in room-temperature lasing at 1.5 µm, which Lee believes is a first for this type of structure, although InAs quantum dash structures have also been shown to lase at this wavelength.
Solar-cell-production issuesFrank Dimroth, from Fraunhofer ISE, Germany, highlighted the perils of scaling up the production of next-generation solar cells from single to multiwafer reactors. He spoke about problems such as unintentional doping that can occur during the transfer of growth of solar cells containing InGaAsN from a single-wafer to an 8 x 4 inch multiwafer platform. InGaAsN, which can be lattice-matched to GaAs, has been identified as a suitable material for next-generation solar cells containing up to six active junctions.
InGaAsN s suitability was assessed by first producing single-junction cells on an AIX200 single-wafer reactor. These cells were considered to be of suitable quality to incorporate into multijunction cells, but transfer to an 8 x 4 inch platform was problematic.
For example, triethylgallium (TEGa) - the source used for single-wafer growth - caused black deposits on the inside of the reactor lid. Dimroth believes that these deposits alter the incorporation of indium in InGaAsN, preventing run-to-run control of the composition of the quaternary material. Substituting trimethylgallium (TMGa) for TEGa introduced a new set of problems: solar cells grown with TMGa had high background-doping levels, and external quantum efficiencies in the region of interest (800-1200 nm) of less than half of those produced from a single wafer reactor. Dimroth admits that further work is required on the multiwafer reactor either to reduce background doping or to solve the problems associated with TEGa growth.
A method to fabricate 1.55 µm VCSELs by fusing a strain-compensated InAlGaAs MQW region grown on an InP substrate, and AlGaAs mirrors on GaAs substrates was reported by Rudra Alok from the EPFL. Wafer-bonding produced 7 µm-aperture devices that operate in CW mode at room temperature with output powers of 3 mW and associated side-mode suppression ratios of 30-35 dB.
Emission at a wavelength of 9 µm from a quantum cascade laser (QCL) was the subject of a talk by Andrey Krysa from the III-V growth facility at the University of Sheffield, UK. Krysa explained that, historically, MBE technology has dominated the growth of QCLs, but recently his group has produced InP/InGaAs/InAlAs QCLs via MOVPE. These lasers exhibit peak powers in excess of 1 W at room temperature, which is comparable to state-of-the-art devices grown by MBE.
Exotic structuresGeorge Wang from Sandia National Labs, US, described his work on AlGaInN nanowires. He explained that, to date, the primary growth methods used in the synthesis of GaN nanowires have been thermal evaporation and chemical-vapor deposition techniques using gallium metal or GaN powder source materials in hot-wall tube reactors. Wang believes these evaporation-based techniques suffer from a general lack of control, reproducibility and scalability, and cannot be used to produce complex heterostructures.
The Sandia group now employs an MOVPE process to synthesize GaN nanowires on 2 inch substrates coated with nickel catalysts. Various studies indicate that the nanowires are single-crystalline GaN with nickel clusters at the tips, indicating growth via the vapor-liquid-solid mechanism. The nanowires have tip diameters of 20-100 nm and lengths of up to several tens of microns (figure 2). Using MOVPE, core-shell heterostructure nanowires consisting of GaN cores and various III-nitride shell materials, including AlN, InN, AlGaN and InGaN, have also been produced.
The creation of more complex nanostructures, such as "nanotrees" (figure 3), was described by Werner Seifert from the University of Lund, Sweden. These structures are produced by seeding a GaP [111]B substrate with gold aerosol nanoparticles, which act as catalysts for GaP nanowire growth in the [111] direction. The nanowires act as the "trunks" of the nanotrees, with subsequent "branches" generated by seeding the trunks with additional gold nanoparticles. The process can be continued for a third time forming sub-branches or "leaves".
To form an ordered array of nanotrees, such as those in figure 3, a regular pattern of gold particles is deposited on the substrate by using electron-beam lithography. Growth is not restricted to GaP and researchers have already produced InP-based trees on InP substrates. The team hopes that one day nanoleaves will be able to replicate plant photosynthesis.
Current MOVPE technology was demonstrated by dazzling the audience with a GaN-based LED torch, while Craford underlined the global impact of semiconductor light sources by highlighting the installation of LEDs in traffic signals in Cambodia.
Craford is optimistic about the technology and, although he believes that improvements must be made to both substrates and growth processes, he predicts that the high efficiencies offered by LEDs will fuel the increase of worldwide production levels to 50 billion GaN-based LEDs per year by 2010. This assumes a 10% penetration into traditional markets and GaN-based LED lifetimes of 10 years. Overall it would correspond to an annual production of 25 million 2 inch wafers, or 1000 reactors yielding 68 wafers each day.
UV-LED developmentsLater in the week John Edmond, director of optoelectronic technology at US-based Cree, described developments in deep-UV-LED sources. Edmond explained that solid-state sources emitting in the 270-340 nm region will enable technological advances such as fluorescence-based biological-agent detection; non-line-of-sight communications; water purification; and industrial processing, including ink drying and epoxy curing. In addition, Edmond described the benefits of optimizing device geometries to improve light extraction, reporting output powers of 1 mW at 20 mA and a peak wavelength of 340 nm.
An alternative method to improve light extraction - by introducing a corrugated interfacial substrate - was demonstrated by Samsung Advanced Institute of Technology, Korea. Improvements of more than 60% in extraction power, compared with conventional UV LEDs, were achieved using standard photolithography and reactive-ion etching.
High prices and the limited availability of GaN substrates have led to the fabrication of the majority of GaN-based devices on sapphire and SiC substrates (see Compound Semiconductor July 2004). Silicon substrates are attractive alternatives in terms of cost and scale, although they have inherent drawbacks. In particular, GaN epilayers can exhibit bowing and cracking due to the large thermal coefficient mismatch between silicon and GaN.
An alternative growth scheme was put forward by Matthew Charles from the University of Cambridge, UK. He suggested incorporating AlN interlayers, AlGaN graded layers, AlN/GaN superlattices and SiN into substrates to reduce tensile strain and dislocation density, thus allowing the growth of 2.3 µm-thick uncracked films of GaN on silicon.
Growing AlGaN/GaN HEMTs on silicon substrates was the subject of a talk given by Shiping Guo from Emcore Corporation, Somerset, NJ. Using Emcore s proprietary buffer technology, researchers fabricated a crack-free 2 µm-thick buffer layer on which they grew the HEMT structure.
Quantum dot (QD) structures and devices were another major talking point in Maui. Most groups grow QDs by strain-induced self-ordering using the Stranski-Krastanov (SK) growth mode, but a few researchers presented work involving prior growth patterning of substrates, leading to the formation of pyramidal structures with a dot defined at the very apex of the pyramid.
Improving QD lasersPelucchi Emanuele, from the Ecole Polytechnique Fédérale de Lausanne (EPFL), Switzerland, explained that although SK growth can produce very good interfaces, site control of the dots is difficult to achieve and there is a wide distribution in their peak emission wavelengths. Emanuele believes that ordered arrays of dots with reproducible and controlled excitonic states can provide opportunities for studying fundamental physics (for example, exploring entangled and correlated excitonic and photonic states) and deliver improved performances in QD lasers.
The EPFL group has patterned [111]B GaAs substrates with hexagonal arrays of tetrahedral recesses with a pitch ranging from 300 nm to 5 µm using electron-beam lithography. Subsequent growth of GaAs/AlGaAs followed by InGaAs enables the formation of pyramidal structures with quantum wires along the edges and a dot at the top of each structure emitting at around 800 nm (figure 1). The size uniformity of the dots is sufficient for the intrinsic broadening of light emission from a single dot to outweigh any contribution from fluctuations in the size of the dots. The EPFL group believes that this will enable the fabrication of QD lasers with all dots participating in the lasing process, giving a performance that could potentially surpass that of state-of-the-art quantum well lasers.
Tung-Po Hsieh, from the National Central University in Taiwan, reported strong photoluminescent (PL) emission at 1.55 µm from InAs QDs on GaAs substrates. He believes that this is the longest wavelength emission reported for InAs-on-GaAs dots grown by MOVPE. This result was achieved by inserting an In0.25Ga0.75As strain-reducing layer between the cap and the InAs QDs.
Jin Hong Lee from the Electronics and Telecommunications Research Institute of Korea reported on the tuning of emission from self-assembled InAs QDs, this time on InP substrates. Increasing the interrupt time after deposition of the InAs QD layer red-shifted emission from 1.48 to 1.66 µm and increased PL intensity. Blue-shifting emission from 1.48 to 1.38 µm was achieved by introducing a thin InAlGaAs layer on top of the InAs QDs.
The Korean team also fabricated ridge waveguide lasers using InP substrates, an InAlGaAs matrix and seven stacks of InAs QDs. This resulted in room-temperature lasing at 1.5 µm, which Lee believes is a first for this type of structure, although InAs quantum dash structures have also been shown to lase at this wavelength.
Solar-cell-production issuesFrank Dimroth, from Fraunhofer ISE, Germany, highlighted the perils of scaling up the production of next-generation solar cells from single to multiwafer reactors. He spoke about problems such as unintentional doping that can occur during the transfer of growth of solar cells containing InGaAsN from a single-wafer to an 8 x 4 inch multiwafer platform. InGaAsN, which can be lattice-matched to GaAs, has been identified as a suitable material for next-generation solar cells containing up to six active junctions.
InGaAsN s suitability was assessed by first producing single-junction cells on an AIX200 single-wafer reactor. These cells were considered to be of suitable quality to incorporate into multijunction cells, but transfer to an 8 x 4 inch platform was problematic.
For example, triethylgallium (TEGa) - the source used for single-wafer growth - caused black deposits on the inside of the reactor lid. Dimroth believes that these deposits alter the incorporation of indium in InGaAsN, preventing run-to-run control of the composition of the quaternary material. Substituting trimethylgallium (TMGa) for TEGa introduced a new set of problems: solar cells grown with TMGa had high background-doping levels, and external quantum efficiencies in the region of interest (800-1200 nm) of less than half of those produced from a single wafer reactor. Dimroth admits that further work is required on the multiwafer reactor either to reduce background doping or to solve the problems associated with TEGa growth.
A method to fabricate 1.55 µm VCSELs by fusing a strain-compensated InAlGaAs MQW region grown on an InP substrate, and AlGaAs mirrors on GaAs substrates was reported by Rudra Alok from the EPFL. Wafer-bonding produced 7 µm-aperture devices that operate in CW mode at room temperature with output powers of 3 mW and associated side-mode suppression ratios of 30-35 dB.
Emission at a wavelength of 9 µm from a quantum cascade laser (QCL) was the subject of a talk by Andrey Krysa from the III-V growth facility at the University of Sheffield, UK. Krysa explained that, historically, MBE technology has dominated the growth of QCLs, but recently his group has produced InP/InGaAs/InAlAs QCLs via MOVPE. These lasers exhibit peak powers in excess of 1 W at room temperature, which is comparable to state-of-the-art devices grown by MBE.
Exotic structuresGeorge Wang from Sandia National Labs, US, described his work on AlGaInN nanowires. He explained that, to date, the primary growth methods used in the synthesis of GaN nanowires have been thermal evaporation and chemical-vapor deposition techniques using gallium metal or GaN powder source materials in hot-wall tube reactors. Wang believes these evaporation-based techniques suffer from a general lack of control, reproducibility and scalability, and cannot be used to produce complex heterostructures.
The Sandia group now employs an MOVPE process to synthesize GaN nanowires on 2 inch substrates coated with nickel catalysts. Various studies indicate that the nanowires are single-crystalline GaN with nickel clusters at the tips, indicating growth via the vapor-liquid-solid mechanism. The nanowires have tip diameters of 20-100 nm and lengths of up to several tens of microns (figure 2). Using MOVPE, core-shell heterostructure nanowires consisting of GaN cores and various III-nitride shell materials, including AlN, InN, AlGaN and InGaN, have also been produced.
The creation of more complex nanostructures, such as "nanotrees" (figure 3), was described by Werner Seifert from the University of Lund, Sweden. These structures are produced by seeding a GaP [111]B substrate with gold aerosol nanoparticles, which act as catalysts for GaP nanowire growth in the [111] direction. The nanowires act as the "trunks" of the nanotrees, with subsequent "branches" generated by seeding the trunks with additional gold nanoparticles. The process can be continued for a third time forming sub-branches or "leaves".
To form an ordered array of nanotrees, such as those in figure 3, a regular pattern of gold particles is deposited on the substrate by using electron-beam lithography. Growth is not restricted to GaP and researchers have already produced InP-based trees on InP substrates. The team hopes that one day nanoleaves will be able to replicate plant photosynthesis.
