† Corresponding author. E-mail:
Project supported by the Science and Technology Major Project of Guangdong Province, China (Grant No. 2015B010112001) and the Natural Science Foundation of Guangdong Province, China (Grant No. 2015A030312011).
Indium-composition fluctuations in InGaN epitaxial layers are suppressed by using periodically-pulsed mixture (PPM) of N2 and H2 carrier gas. Photoluminescence, optical transmission, reciprocal space map and space-resolved cathodoluminescence are employed to characterize the InGaN epilayers. It is shown that the lateral In-fluctuations mainly occur as hillock-like In-rich regions. Both the number and size of In-rich regions are reduced by introducing the PPM carrier gas. Moreover, the measurements first experimentally demonstrate that the H2 carrier gas has a stronger decomposition effect on the In-rich region. As the duration time of the PPM carrier gas increases, the reduction of In-content in the In-rich region reaches up to 12%, however, only 2% for the In-homogeneous region. These factors lead to the suppression of In-fluctuations.
InGaN ternary alloys are promising materials and have received enormous attention due to their direct and adjustable bandgaps in entire visible region, high theoretical electron mobility,[1] and large optical absorption.[2] These alloys offer many potential applications, such as quantum-well-based solid-state emitters,[3] light detectors, photovoltaics,[4] photoelectrochemical (PEC),[5] and thermoelectric devices.[6] For some applications, such as visible light detectors and solar cells, an InGaN film with thickness in excess of 100 nm is required to absorb more than 90% of the incident above-bandgap light.[7] However, due to the low miscibility[8] and large difference in interatomic spacing between GaN and InN (∼ 11%),[9,10] Indium-composition fluctuations (In-fluctuations) occur easily in InGaN epitaxial layers and result in poor electrical and structural properties, especially for high In-content. Moreover, InGaN epilayers which are generally deposited on GaN buffer layer suffer a large lattice constant mismatch between InGaN and GaN and a small critical layer thickness (CLT). For example, The CLT of an InxGa1−xN epilayer with x > 10% is calculated to be less than 100 nm.[11] When the thickness of InGaN epilayer exceeds the CLT, strain relaxation accompanied by the formation of dislocations may occur. Therefore, it is difficult to prepare high crystalline quality InGaN epilayers with In-content > 10% and thickness > 100 nm due to the severe In-fluctuations and high defect density.[12]
Up to now, some efforts have been made to suppress the In-fluctuations in InGaN and these efforts can be classified, in essence, as three categories: (i) strain control, it was proposed that elastic strain helps to suppress the In-fluctuations in InGaN,[9] and single-phase InGaN layers have been grown on lattice-matched ZnO substrates through compressive strain;[13] (ii) non-equilibrium growth, it was found that In-fluctuations could be suppressed by promoting growth rate, thereby converting thermodynamic conditions into non-equilibrium growth conditions;[14] and (iii) indium adlayer control, including semi-bulk methods,[7,15,16] metal-modulated epitaxy,[17,18] and indium modulation technology.[19] All these epitaxial methods use pure nitrogen as carrier gas. It is worth noting that hydrogen was often adopted as carrier gas during the growth of barriers[20,21] and the interruptions between barriers and wells in InGaN multiple quantum wells (MQWs).[22] Its effects have been studied for at least 20 years and one of the effects was that hydrogen could eliminate the inclusions, namely In-fluctuations.[20–22] For thick InGaN epitaxial layers, some investigations have been done about the effects of hydrogen on indium incorporation[23,24] and C and O impurities.[25] However, there are few reports on the effects of hydrogen on the In-fluctuations in thick InGaN layers.
In the present work, we provide another method to suppress the In-fluctuations during the growth of thick InGaN layers by periodically-pulsed mixture (PPM) of N2 and H2 carrier gases. In-rich regions, which mainly contribute to the In-fluctuations, are directly observed and evidently reduced by the PPM carrier gas. The effects of hydrogen on the In-rich region and In-homogeneous region are first experimentally investigated. It is shown that hydrogen has much stronger decomposition effect on the In-rich regions and, therefore, the In-fluctuations can be effectively suppressed by controlling the duration time of PPM.
All experimental InGaN samples were grown on c-plane sapphire substrates by low-pressure metal–organic chemical vapor deposition (MOCVD) in a closed-coupled showerhead reactor. Trimethyl indium (TMIn), tetraethyl gallium (TEGa), and ammonia (NH3) were used as metal and nitride precursors, respectively. N2 or H2 carrier gas flowed through the metal precursors bubblers. The growth temperature measured by pyrometer was converted into the wafer temperature shown here, calibrated by means of the blackbody radiation. The substrate temperature was firstly increased to 1098 °C in H2 atmosphere for 300 s to remove the surface damage and contamination. Then, a 30-nm-thick GaN nucleation layer was grown at 536 °C, followed by a 2.5-μm-thick unintentionally doped (uid) GaN layer deposited at 1069 °C. Next, an InxGa1−xN (x ∼ 14%) transition layer with nominal 60-nm thickness was grown with pure N2 carrier gas to avoid aggravating the composition pulling effect. Finally, a uid-InGaN main layer with nominal 90-nm thickness was deposited under the identical growth conditions of transition layer except for the periodically changed carrier gas between pure N2 gas and a mixed gas of H2 (400 sccm) and N2 (5842 sccm). During the deposition of all InGaN layers, the metal and nitride precursors were kept continuous. The flow rate of NH3, TMIn, and TEGa were 9 slm, 27.72 μmol/min, and 4.52 μmol/min, while the growth temperature and pressure were 765 °C and 300 mbar (1 bar = 105 Pa), respectively. Four samples were grown by PPM method with pulsed duration times of 0, 5, 10, and 15 s in each period of 45 s, respectively. The InGaN main layer consists of 55 periods in total and each periodic thickness is around 1.6 nm, corresponding to the growth rate of ∼ 2.2 nm/min. It should be noted that there were no multilayered structures observed in InGaN main layers by transmission electron microscopy, indicating that the deposited InGaN should be bulk film rather than superlattice. The flow sequences of the carrier gas during the growth of InGaN main layer are shown in the inset of Fig.
Figure
Figure
To further investigate the effects of PPM carrier gas on the deposited InGaN, we also carry out space-resolved CL measurements. The measured results are shown in Fig.
Besides the number and size of the hillocks, the variation of CL spectrum with the PPM duration time is also important for understanding the role of PPM carrier gas. The CL spectra are therefore characterized by focusing an electron beam on the selected hillocks, as shown in Figs.
It should be noted that the spatial extent of In phase separation is still under debate.[32–34] However, the dimension of the phase separation is typically on the order of a few to tens of nanometres, much smaller than the sizes of inhomogeneous regions observed in our InGaN epitaxial layers. Accordingly, strictly speaking, the effect of PPM carrier gas shown in our case is to suppress the In-composition fluctuations rather than the phase separation. The above analysis therefore demonstrates that the suppression of In-fluctuations is a major contribution of the PPM carrier gas.
The hydrogen in PPM carrier gas plays an important role in suppressing the In-fluctuations during the InGaN epitaxy by the MOCVD method. On the one hand, hydrogen could etch (or decompose) InN more easily than GaN via the reverse synthesis reaction because of its small equilibrium constant.[35,36] Indium, one of the reverse synthetic products, partially reacts with hydrogen and forms indium-hydride volatile species[37] which could be desorbed seriously due to their weaker bonds to the surface than In atoms.[4] Meanwhile, hydrogen is a by-product of deposition reactions of
The RSM measurements of asymmetric (105) reflection are performed to further clarify the mechanism of In-fluctuation suppression from the aspect of strain conditions. The broad diffraction spot shown in Fig.
Since the PPM carrier gases may also result in the decrease of In-content in InGaN layer, it is still necessary to clarify the effect of In-content reduction on In fluctuation. An InGaN control sample deposited with N2 carrier gas is prepared. Figure
In this work, a periodically-pulsed mixture of N2 and H2 carrier gas is introduced into the MOCVD during InGaN growth. The In-fluctuations are effectively suppressed by this method. A decreasing process of In-content in the In-rich region with increasing the hydrogen duration time is first clearly presented. It is found that not only the number and size of the hillock-like In-rich InGaN regions but also the In-content of the In-rich region is significantly reduced by the introduction of PPM carrier gas. These factors contribute to the suppression of In-fluctuations. The PPM carrier gas provides a way to precisely control the reaction amount and time of introduced hydrogen.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] | |
[23] | |
[24] | |
[25] | |
[26] | |
[27] | |
[28] | |
[29] | |
[30] | |
[31] | |
[32] | |
[33] | |
[34] | |
[35] | |
[36] | |
[37] | |
[38] |