After working in the optical fiber communication industry for a long time, we take it for granted that multimode corresponds to 850 nm, or 850 nm and 910 nm wavelengths, and that single-mode corresponds to 1260-1650 nm wavelengths, especially those near the 1310 nm band and the 1550 nm band.
Hey, one day we happened to learn that there are multimode lasers with a wavelength of 1550 nm, and also multimode lasers with a wavelength of 1310 nm. How does this happen? In fact, 850 nm can also be made into a single-mode laser. In the industry, single-mode refers to a single transverse mode, and multimode refers to multiple transverse modes. Wavelength actually expresses the characteristic of the longitudinal mode, which can be divided into single longitudinal mode and multiple longitudinal modes, that is, single-wavelength lasers or multi-wavelength lasers.
This wavelength specifically refers to the “amplification” wavelength. A laser is an optical amplifier based on the principle of interference amplification. So many people say that a laser emits “coherent light.” Which wavelengths do single-mode and multimode correspond to? Then the question becomes: what is the relationship between the single transverse mode or multiple transverse modes of a (stimulated emission) optical amplifier and the output range of its longitudinal modes? Answer: theoretically there is no corresponding relationship. A 1310 nm laser can be single transverse mode or multiple transverse modes. Similarly, 850 nm and 1550 nm lasers can be designed as single transverse mode or multiple transverse modes.
However, we should consider that a specific industrial chain has already formed in practical applications, and in some special scenarios, which approach has the highest cost-effectiveness. For example, in application scenarios based on optical fiber communication, multi-transverse-mode lasers at 850 nm have a low-cost advantage, while single-transverse-mode lasers based on 1310 nm or 1550 nm have excellent transmission performance with low dispersion/low loss in optical fiber.
For example, in the LiDAR industry, for edge-emitting lasers, there is no need to consider the characteristic of a single transverse mode in TOF or AM scenarios. Multi-mode EEL edge-emitting long-wavelength lasers can be made to increase power and reduce eye-safety risk. Light has one transmission direction, and the mode along the transmission direction is the longitudinal mode, which is based on the frequency-related characteristic of “time interference” in the laser.

The light transmission direction of a laser
The optical field distribution on the cross-section perpendicular to the transmission direction is a transverse mode, which is based on some distribution characteristics formed by “spatial interference” in the laser.
People who have done laser design know that there are two kinds of solutions in laser formulas — temporal solutions and spatial solutions.

The temporal and spatial dimensions of a laser
Transverse modes are based on the energy distribution in space. This is a distribution that can be seen in real life, which we will discuss later. Longitudinal modes are based on the energy distribution in time and the transmission of light at the speed of light, with several parameters: wavelength, (light) speed, wave time, and distance. Wavelength is invisible to the naked eye, so it needs to be tested with a spectrometer, which is actually a spectrum analyzer. Light frequency × light wavelength = the speed of light in vacuum C, where C is a constant. Therefore, a spectrum analyzer can be used to test the frequency (for example: ~300 THz corresponds to a 1.3 μm wavelength, ~200 THz corresponds to a 1.5 μm wavelength) and calculate the wavelength.

The transmission of light is the transmission of waves
The wavelength mentioned here specifically refers to a wavelength range.

Wavelength is a range. The wavelength range is determined by the gain material. Within this range, selecting a specific wavelength for interference is the function of the resonant cavity, and interference can play an amplifying role. The resonant cavity has a cavity length. Both DFB and VCSEL gratings have “extremely small” cavity lengths. The cavity length and the wave speed of light can be used to calculate the reflection time of light, and time-based interference is the longitudinal mode. In our context, these longitudinal modes have different emphases. The resonant cavity and the gain are made together, which is the choice of most lasers in optical transceivers, such as VCSEL, DFB, EML and other chips. The resonant cavity (the cavity for interference amplification) determines which wavelengths are amplified. The resonant cavity and the gain material choose the same type of material, such as VCSEL using GaAs material and DFB using the InP material system.
The resonant cavity and the gain material use different material systems, such as the gain material using the GaAs gallium arsenide material system, and the resonant cavity using the silicon material system. This is Huawei’s process of epitaxially growing gallium arsenide on a silicon substrate and making gallium arsenide quantum dots at OFC2023. Using InP as the gain and silicon as the resonator is Intel’s silicon-based integrated laser. Gain material: when the resonator and the gain material are manufactured independently, we have an external-cavity laser, that is, a laser composed of a resonator “independent of the gain part.” Back to the transverse mode, single-mode and multimode in the industry specifically refer to the transverse mode, that is, the energy distribution in space. When there are multiple energy peaks, it is multimode; when there is only one energy peak, it is single-mode. Spatial fluctuation can be seen with the eyes. When our laser pointer shines on a screen, the distribution of bright spots can be seen, which is the embodiment of the transverse mode.

The transverse mode of the laser spot distribution. The VCSEL in an optical transceiver is a multimode laser, that is, there are multiple bright spots. VCSEL is also called a vertical-cavity surface-emitting laser. Vertical cavity refers to the relative coordinates of the wafer. The resonant cavity is perpendicular to the wafer surface, called a vertical cavity. The direction of the resonant cavity is related to the output direction of the laser, which determines the light output direction of the laser, without special processing. It can achieve emission perpendicular to the wafer surface, either surface emission or bottom emission. Single-mode lasers generally emit from the side, either from the left or right side, or from both sides simultaneously. For edge emission, the output direction of light is realized by the transverse resonant cavity, without other additional processing. The output wavelength range is realized by the gain material, and the transverse spot is determined by the spatial structure of the output waveguide.

Two common waveguide structures for confining single mode
The space of light also “fluctuates,” and as long as the width is appropriate, multiple bright spots will not appear. For the design of most optical communication lasers, the waveguide width of the InP material system is about 2 μm, which is a common single-mode confinement structure. Since it is common, it is not the only one. If the spatial confinement structure is widened, multiple transverse modes will appear. This was very common in early lasers. Later, making single mode in the optical communication field became easy, and people forgot about this.
In recent years, because of the need to increase output power, people have begun to discuss widening the waveguide again. For example, some use multimode to realize DFB+SOA cascade, and then use a virtual waist to realize single mode. The 5 μm-wide laser that Huawei has made in recent years is essentially also multi-transverse-mode, with the purpose of increasing output power. Another example is the TOF method in the LiDAR field, which is more inclined to put high power into single-mode and multimode. They do not need to transmit in optical fiber, so there is no need to consider this aspect. The figure below is a multi-transverse-mode laser whose waveguide width is stretched very wide to achieve edge emission.

Edge-emitting multimode laser for LiDAR
In summary, as far as edge-emitting lasers are concerned, optical fiber communication requires single mode, and single mode is achievable. Some application scenarios have partial multimode designs, and we need additional processing to restore single mode. This behavior is essentially to meet the needs of optical fiber communication. The logic behind it is that the communication technology of commercial solid glass optical fiber has an optimal transmission frequency band. (Hollow-core fiber is not affected by this band; OFC2023 described the progress of hollow-core fiber.) The only material that can satisfy this wavelength range and achieve single mode is InGaAsP/InP, or AlGaInAs/InP. This material system cannot achieve single-mode output in a vertical-cavity surface-emitting structure, or even achieve high-reliability large-scale industrial manufacturing. Therefore, people have to choose the unfriendly edge-emitting optical path output structure.

The waveguide structure determines whether it is single-mode or multimode
Surface emission is very cheap and can theoretically achieve single-mode output, but the industry cannot achieve large-scale, long-term, stable and reliable manufacturing. This is caused by the trench etching in the process flow. Trench etching is for lateral oxidation, used to guide the current path.

VCSEL process flow
Peel back those airtight coatings on the surface, and you will see an etched trench. The groove forms an internal current input and emits light through lateral oxidation. If the light structure is shrunk to a diameter of 1.x μm, it can be confined to “single mode” in the spatial dimension. However, in this hole there is still a DBR layer, that is, the resonant cavity, which needs to control light reflection and interference amplification after resonance. The DBR layer requires dozens of layers. The smaller the island, the easier it is to achieve single mode, but it brings hidden dangers to the reliability of the resonant cavity. In the 50G and 100G VCSELs seen this year, the diameter of the confinement hole is about 5-6 μm.

Single-mode or multimode is determined by the waveguide structure
The many methods above show that VCSEL can theoretically achieve single mode, and in the current multimode state, there are also many auxiliary processing modes that can achieve single mode. However, the industry’s expectation is low cost and large-scale manufacturing, so multimode is the industry’s choice. In summary, technically, single-mode or multimode and wavelength are designed independently. The transverse mode is generally limited by the pinhole diffraction spatial structure. The wavelength is jointly affected by the gain layer and the resonant cavity, with the gain material being the main reason.
In terms of application, the industry has requirements for both wavelength and mode. Single-mode InP lasers, especially those at distances above 500 meters / within thousands of kilometers, have performance advantages, but InP edge-emitting lasers do not have a cost advantage. Multimode VCSEL lasers, although low in cost, have performance inferior to single-mode lasers, with a use distance of 180 meters, and no advantage in wavelength.
End
……………………………………………… Related topics:
- Do You Really Understand ITU-T G.65X Single-Mode Fiber?
- Master Multimode Fiber OM1/2/3/4/5 Parameter Comparison in One Article!
- Center Wavelength/Frequency: What Spacing Is Appropriate?
Enjoy this article?
Subscribe to get new optical communication insights delivered to your inbox. Free, unsubscribe anytime.
