Violet Laser Diode Enables Lighting Communication

Violet laser diode (VLD) based white-light source with high color rendering index (CRI) for lighting communication is implemented by covering with Y3Al5O12:Ce3+ (YAG:Ce) or Lu3Al5O12:Ce3+/CaAlSiN3:Eu2+ (LuAG:Ce/CASN:Eu) phosphorous diffuser plates. After passing the beam of VLD biased at 70 mA (~2Ith) through the YAG:Ce phosphorous diffuser, a daylight with a correlated color temperature (CCT) of 5068 K and a CRI of 65 is acquired to provide a forward error correction (FEC) certified data rate of 4.4 Gbit/s. By using the VLD biased at 122 mA (~3.5Ith) to excite the LuAG:Ce/CASN:Eu phosphorous diffuser with 0.85-mm thickness, a warm white-light source with a CCT of 2700 K and a CRI of 87.9 is obtained at a cost of decreasing transmission capacity to 2.4 Gbit/s. Thinning the phosphor thickness to 0.75 mm effectively reduces the required bias current by 32 mA to achieve the same CCT for the delivered white light, which offers an enlarged CRI of 89.1 and an increased data rate of 4.4 Gbit/s. Further enlarging the bias current to 105 mA remains the white-light transmission capacity at 4.4 Gbit/s but reveals an increased CCT of 3023 K and an upgraded CRI of 91.5.

data rate of 4.5 Gbit/s over 1.5-m free-space transmission was demonstrated by Wang et al. 30 . The high-order CAP modulation and hybrid post equalization additionally helps the red/blue/green/yellow (RBGY)-LED based WDM-VLC to approach 8-Gbit/s free-space transmission over 1 m 31 . More recently, Bamiedakis et al. employed a μLED with 3-tap feed-forward pre-equalized 4-level pulse amplitude modulation (PAM-4) to demonstrate a 2-Gbit/s VLC link over 0.6 m 32 . Sun et al. fabricated an aluminum-doped zinc oxide (AZO) LED with maximal output power of 42 mW to carry 3-Gbit/s 32-QAM OFDM data at chip level 33 . By employing bit-rate adaptive OFDM scheme, Chun et al. used RC R-LED and B/G-μLEDs to demonstrate WDM-VLC transmission up to 11.28 Gbit/s 34  As compared to the currently available LEDs with limited modulation bandwidth of several tens MHz [42][43][44] , the laser diodes (LDs) with higher coherence, narrower spectral linewidth and higher relaxation oscillation frequency is capable of handling data bandwidth up to several GHz 45,46 , which is more suitable for next-generation lighting communications. Recently, Watson et al. used a BLD with 1.4-GHz modulation bandwidth for error-free OOK transmission at 2.5 Gbit/s 47 . Later on, a RLD based bidirectional 2.5-Gbit/s VLC over 20-km single-mode fiber (SMF) and 15-m free space was reported by Chen et al. 48 . With low-noise amplification and equalization at receiving end, Chang et al. demonstrated a 100-Gbit/s (12.5 Gbit/s/channel × 8 channels) MIMO-VLC over 5 m by employing vertical-cavity surface-emitting lasers (VCSELs) 49 60 .
Note that either the RGB mixed lights 61,62 or the yellow phosphor combined blue light 63,64 has been widely realized for white-light generation. Therein the white light generated by illuminating the blue light through the yellow phosphor becomes a more compact and cost-effective solution than the RGB mixed white-light source with higher transmission capacity. Fortunately, such a drawback can be alleviated by carrying high-spectral-efficiency data formats such as high-level quadrature amplitude modulation (QAM)-OFDM 65,66 . However, the BLD based white-light source inevitably suffers from insufficient CRI 67 with the lack of violet color, thus limiting its applications in some high-end lighting circumstances such as museums, galleries and medical field. The use of violet LD (VLD) for exciting the fluorescent phosphor would be helpful, which enables the generation of white light with larger luminous efficiency and CRI than that demonstrated by using the BLD 68 . Nonetheless, the VLD based lighting communication has yet to be discussed to date.
In this work, for the first time, the phosphorous diffuser covered VLD at 405 nm is demonstrated for both white-lighting and QAM-OFDM data transmission in free space, in which two different fluorescent phosphors of Y 3 Al 5 O 12 :Ce 3+ (YAG:Ce) and Lu 3 Al 5 O 12 :Ce 3+ /CaAlSiN 3 :Eu 2+ (LuAG:Ce/CASN:Eu) are employed for comparison. At beginning, the VLD based point-to-point (PtP) VLC system for free-space transmission over 7 m is demonstrated. By exciting the phosphorous diffusers with the light of the VLD at different bias currents, the white-lighting performances including CCT, CRI and luminous flux are discussed. Furthermore, the optimization on allowable encoding bandwidth and transmission performance of the white light carried QAM-OFDM data over 0.5 m in free-space is performed.

Results
To perform high-quality data stream, the VLD needs not only large modulation bandwidth and small relative intensity noise (RIN), but also flat throughput intensity and high extinction ratio. The VLD optimizes its bias current to 75 mA for achieving a data rate of 11.2 Gbit/s under PtP transmission over a 7-m free-space link, which allows a 16-QAM OFDM data bandwidth of 2.8 GHz to meet the forward error correction (FEC) criterion, and the related error vector magnitude (EVM), signal-to-noise ratio (SNR) and bit-error-rate (BER) of 17.1%, 15.3 dB and 3.4 × 10 −3 are observed, respectively, as shown in Fig. 1. To further upgrade the transmission capacity, increasing the OFDM data bandwidth induced high-frequency SNRs degradation can be compensated by introducing the pre-leveling technique 69 . In detail, the OFDM subcarrier pre-leveling technique is to pre-amplify the OFDM subcarriers by multiplying a rising exponential function 70 , which slightly sacrifices the subcarrier SNRs at low frequency to compensate the high-subcarrier-frequency SNRs for improving the average SNR of the transmitted QAM-OFDM data. This enlarges the allowable OFDM bandwidth to 3 GHz at a raw data rate of 12 Gbit/s, and the subcarrier SNRs and related constellation plot are also shown in Fig. 1 to reveal an average EVM of 17.2%, an average SNR of 15.2 dB and a BER of 3.8 × 10 −3 . Note that the transmission capacity as high as 12 Gbit/s is guaranteed for the VLD based 7-m PtP VLC system.
For white-light generation, the VLD at different bias currents is employed to excite the YAG:Ce or LuAG:Ce/ CASN:Eu fluorescent phosphor, and the corresponding optical spectra are shown in Fig. 2(a). By exciting the YAG:Ce or the LuAG:Ce/CASN:Eu fluorescent phosphor, the violet laser beam not only broadens its spectral linewidth but also emits blue-yellow or yellow-orange light, respectively. These excited light components are mixed with the divergent violet laser beam to generate the white light, as shown in Fig. 2(b). Note that the LuAG:Ce/ CASN:Eu fluorescent phosphor enables to deliver the spontaneous emission light with longer wavelength and stronger intensity than the YAG:Ce does, indicating that the LuAG:Ce/CASN:Eu fluorescent phosphor provides higher exciting efficiency. Obviously, increasing the bias current of the VLD gradually broadens its spectral linewidth and enhances the excited light intensity for all cases.
Moreover, the Fig. 2(c) displays the achievable CCTs of the phosphorous diffuser covered VLD output at different bias currents, in which the link distance is 0.5 m. By increasing the bias current from 50 to 80 mA, the YAG:Ce fluorescent phosphor generated white light exhibits daylight appearance with its CCT reducing from 6750 to 4750 K. Such a variable CCT range entirely covers the International Commission on Illumination (CIE) defined standard illuminants of D50 (5000 K), D55 (5500 K) and D65 (6504 K). In contrast to the YAG:Ce fluorescent phosphor, the 0.85-mm-thick LuAG:Ce/CASN:Eu phosphorous diffuser covered VLD output enlarges its CCT from 1750 to 2700 K with increasing the bias current from 70 to 122 mA, and the proposed warm white-light source is comparable to commercial incandescent light bulbs. By thinning the phosphor thickness to 0.75 mm, it effectively reduces the demanded VLD bias current by 32 mA to deliver the white light with the same CCT of 2700 K. Enlarging the bias current to 105 mA slightly increases the CCT of the generated white light to 3000 K, which meets the CIE defined standard illuminant of F12. At the same bias current of 70 mA, the LuAG:Ce/ CASN:Eu phosphorous diffuser covered VLD output clearly shows warmer white light than the YAG:Ce one, which verifies its higher exciting efficiency. In addition, the CRI responses of the white light generated from the VLD illuminated phosphor diffuser at different bias currents are shown and compared in Fig. 2(d). Without adding the phosphorous diffuser, the violet laser beam decreases its CRI from 62 to 10 by increasing the bias current from 50 to 122 mA. With the YAG:Ce fluorescent phosphor adhering to the front of the VLD, increasing the bias current from 50 to 80 mA effectively raises the CRI of the generated white light from 64 to 71. With the use of the 0.85-mm-thick LuAG:Ce/CASN:Eu fluorescent phosphor, an increased CRI from 72 to 88 is observed for the generated white light as the VLD bias current increases from 70 to 122 mA. At the same bias current of 70 mA, the 0.75-mm-thick LuAG:Ce/CASN:Eu phosphorous diffuser covered VLD output exhibits the highest CRI of 81 among all cases. Note that the CRI can further be increased up to 91.5 by continuously increasing the bias current of the VLD to 105 mA.
After 0.5-m propagation in free space, the angular radiation patterns of the phosphorous diffuser covered VLD output are shown in Fig. 2(e), where the 90° and 0° represent the normal and grazing incidences, respectively. Note that the highest luminance is obtained at an orientation angle of 90°, and the proposed white-light source significantly extends its optical field distribution by increasing the VLD bias current. Three cases show similar lambertian-like shape with uniform optical field distribution, which verifies the strong scattering effect of these phosphorous diffusers. By enlarging the bias current from 50 to 80 mA, the luminous flux of the VLD illuminating YAG:Ce phosphorous diffuser is increased from 0.4 to 1.8 lux with its optical field distribution ranging from 30° to 150°. At the same bias current of 70 mA, changing the phosphorous diffuser from YAG:Ce to the 0.85-mm-thick LuAG:Ce/CASN:Eu essentially enhances the luminous flux of the generated white light to 2.6 lux, which is slightly lower than that of 2.67 lux for the 0.75-mm-thick fluorescent phosphor delivered one. By increasing the bias current of the VLD to 122 mA, the luminous flux can be increased to 5.1 lux with its optical field distribution ranging from 25° to 155°. Thinning the thickness of the LuAG:Ce/CASN:Eu fluorescent phosphor to 0.75 mm and decreasing the bias current of the VLD to 105 mA make the generated white light remain a similar luminous flux of 5.2 lux. However, such a thickness induces relatively low scattering effect to exhibit a slightly focused optical field distribution ranged from 45° to 135°. As compared to the brightness of conventional white-light LEDs, although the luminous flux of proposed white light is somewhat low for lighting purpose, a LD array covered with phosphorous phosphor could help to solve this problem. Apparently, the residual violet light power is a simple multiplication with the number of LD chips in the array, as the generated white light has to preserve its CCT a constant during the multiplication of its illuminance. Assuming that the pupil diameter of the human eye is 7 mm and the free-space distance is 0.5 m, an irradiated angle of 0.8° (0.014 rad) with a related solid angle of 1.54 × 10 −4 sr can be obtained for subsequent calculation 60 . After passing through the 0.85-mm-thick LuAG:Ce/CASN:Eu phosphorous diffuser, the divergent white light (without collimating and refocusing) of the VLD biased at 122 mA exhibits a residual power of only 0.43 μW (within an accepting aperture diameter of 1.2 mm) as mesured at a distance of 0.5 m from the source. This corresponds to a related irradiance of 3.8 mW/m 2 for the residual violet light, indicating a radiance of 0.025 kW/m 2 sr to satisfy the IEC 62471 risk group-0 (RG0) criterion of < 0.1 kW/m 2 sr. That is, even a LD array with up to 400 chips can still pass the upper limitation of the RG1 criterion (0.1-10 kW/m 2 sr).
To optimize the operational parameter for transmission after covering the VLD with the YAG:Ce and LuAG:Ce/CASN:Eu phosphorous diffusers, the VLD at different bias currents is directly encoded by two 16-QAM OFDM data with raw data rates of 4 and 2 Gbit/s covering bandwidths of 1 and 0.5 GHz, respectively. Indeed, the standard deviation of an OFDM signal can be determined to evaluate the clipping noise, which helps to optimize the operation condition of transmitters 71,72 . In addition to the aforementioned standard deviation of the OFDM signal, the other parameters of the transmitter including its modulation bandwidth, throughput intensity, relative intensity noise, modulation depth and on/off extinction ratio should also be taken into account. As a result, the BERs, constellation plots and subcarrier SNRs of the 0.5-m free-space transmitted 16-QAM OFDM data carried by the phosphorous diffuser covered VLD at different bias currents are directly employed for optimization, as shown in Fig. 3. In principle, the insufficient bias current of the VLD induces inadequate modulation bandwidth, strong RIN and data clipping effect to distort the carried data and degrade its receiving performance 73,74 . In contrast, the over biasing inevitably suffers from the declined throughput intensity at low frequency and the decreased data extinction, which also degrades the received data quality. When exciting the YAG:Ce and LuAG:Ce/CASN:Eu fluorescent phosphors with the divergent VLD beam, the same optimized bias current of 70 mA is observed to reveal the clearest constellation plot with the lowest BERs of 1 × 10 −3 , 8.63 × 10 −7 (0.75-mm case) and 1.9 × 10 −3 (0.85-mm case) for the carried 4-and 2-Gbit/s data, respectively. In detail, for the 4-Gbit/s data carried by the white light generated from the VLD excited YAG:Ce fluorescent phosphor, the average EVM of 14.9% and SNR of 16.5 dB are obtained. By using the 0.85-mm-thick LuAG:Ce/CASN:Eu fluorescent phosphor, the white light carried 2-Gbit/s data exhibits an average EVM of 15.9% and an SNR of 16 dB. In contrast, thinning the LuAG:Ce/CASN:Eu fluorescent phosphor to 0.75 mm effectively improves the EVM to 9.5% and the SNR to 20.5 dB.
For white-lighting communication, the VLD covered with YAG:Ce phosphorous diffuser changes its bias current to 50, 70 and 80 mA to obtain the corresponding allowable data rates, and the subcarrier SNRs and related constellation plots of received QAM-OFDM data are shown in Fig. 4(a). To meet the FEC demanded BER of 3.8 × 10 −3 at a VLC bias current of 50 mA, the delivered white light successfully carries the 0.8-GHz 16-QAM OFDM data to achieve a raw data rate of 3.2 Gbit/s with an average EVM of 17.25%, an SNR of 15.3 dB and a BER of 3.6 × 10 −3 . Enlarging the bias current to 70 mA effectively extends the allowable OFDM data bandwidth to 1.1 GHz at a raw data rate of up to 4.4 Gbit/s, and the related EVM, SNR and BER of 16.4%, 15.7 dB and 2.5 × 10 −3 are observed, respectively. By continuously increasing the bias current to 80 mA, although the declined throughput intensity of the VLD at low frequency is inevitably introduced, a raw data rate of 4 Gbit/s can still be achieved with an average EVM of 15.6%, an SNR of 16.1 dB and a BER of 1.6 × 10 −3 .
In contrast, by using the VLD to excite the 0.85-mm LuAG:Ce/CASN:Eu fluorescent phosphor for white-light generation, the allowable OFDM data bandwidth is slightly decreased from 1.1 to 0.6 GHz at the same bias current of 70 mA, which only achieves a raw data rate of 2.4 Gbit/s with corresponding EVM, SNR and BER of 16.8%, 15.8 dB and 3 × 10 −3 , respectively, as shown in Fig. 4(b). Enlarging the VLD bias current to 95 mA maintains the carried OFDM data bandwidth at same raw data rate of 2.4 Gbit/s with an EVM of 17%, an SNR of 15.4 dB and a BER of 3.3 × 10 −3 . By increasing the bias current up to 122 mA, the declined throughput of the VLD seriously decreases the allowable OFDM data bandwidth of the generated white light to 0.5 GHz, which limits the data rate to 2 Gbit/s with corresponding EVM, SNR and BER of 16.8%, 15.8 dB and 3.5 × 10 −3 , respectively.
For the 70-mA biased VLD covered with the 0.75-mm-thick LuAG:Ce/CASN:Eu fluorescent phosphor, a raw data rate as high as 4.8 Gbit/s with 16-QAM OFDM data covering 1.2-GHz bandwidth is observed with an EVM of 17.3%, an SNR of 15.2 dB and a BER of 3.7 × 10 −3 , as shown in Fig. 4(c). Increasing the bias current to 90 mA slightly decreases the allowable data rate to 4.4 Gbit/s, and the related EVM, SNR and BER of 16.9%, 15.5 dB and 3 × 10 −3 are observed, respectively. When the bias current is further increased to 105 mA, the proposed white light maintains its transmission capacity at 4.4 Gbit/s with corresponding EVM, SNR and BER of 17.1%, 15.3 dB and 3.4 × 10 −3 , respectively. The last technique employed to compensate the high-frequency intensity declination and to extend the OFDM data bandwidth for upgrading the transmission capacity of the generated white light is the OFDM subcarrier pre-leveling technique. To pass the FEC criterion, the allowable OFDM data bandwidths (BW) and related raw data rates (D) of the VLD covered with phosphorous diffusers before and after pre-leveling are summarized in Table 1.
Owing to the different operational conditions such as the bias current of the VLD and the various frequency responses induced by different phosphors, some of the cases cannot increase their raw data rates by using the pre-leveling technique. By pre-leveling with the same power-to-frequency slope of 0.8 dB/GHz, two cases of the YAG:Ce covered VLD biased at 50 and 80 mA effectively increase the allowable OFDM data bandwidths to 0.9 and 1.1 GHz at raw data rates of 3.6 and 4.4 Gbit/s, respectively. In contrast, at bias currents of 70 and 122 mA for the 0.85-mm-thick LuAG:Ce/CASN:Eu fluorescent phosphor covered VLD, the pre-leveling slopes of 0.2 and 0.6 dB/GHz help to increase the raw data rates of the generated white light to 2.8 and 2.4 Gbit/s, respectively.
In principle, the overall bandwidth may considerably be affected by both the spontaneous emission lifetime of the phosphor and the modulation bandwidth of the laser diode, depending on which mechanism dominates the systematic performance. To further clarify the effects of these phosphorous diffusers on the transmission capacity  Fig. 4(d). By fitting with an exponential function, a lifetime of 60 ns is obtained for the YAG:Ce fluorescent phosphor. In contrast, the 0.85-and 0.75-mm-thick LuAG:Ce/CASN:Eu fluorescent phosphors show their lifetimes of 130 and 120 ns, respectively. Note that these lifetimes are too long to support high-speed data transmission beyond Gbit/s, which means that the transmission capacity of the VLD based white-light source is irrelevant to the used phosphor but is strongly correlated with the residual violet laser beam. After collimating the beam for transmission and refocusing it for receiving, the residual VLD radiance within the proposed white light is calculated to estimate the induced hazard on the human eye. The power-to-current (P-I) responses of the residual violet light power are obtained without and with the YAG:Ce or LuAG:Ce/CASN:Eu phosphorous diffuser adhering to the VLD, as shown in 4(e). The VLD covered with phosphorous diffusers significantly decreases its P-I slope (I > I th ) from 50 to around 2 mW/A as most of the violet laser light contributes to emit yellow-orange light. As compared to other cases, the 0.85-mm-thick LuAG:Ce/CASN:Eu phosphorous diffuser covered VLD at a bias current of 122 mA exhibits the maximal power of 0.07 mW with a related irradiance of 0.62 W/m 2 for the residual violet light. This indicates a radiance of 4.026 kW/m 2 sr to satisfy the RG1 criterion.
For comparison, the white-lighting and data transmission performances of the VLD covered with YAG:Ce and LuAG:Ce/CASN:Eu phosphorous diffusers at different operation conditions are summarized in Table 2. As a result, the use of YAG:Ce fluorescent phosphor enables to generate the white light with its color temperature fitting the daylight appearance, and the corresponding CCT can be decreased to provide warm white light by increasing the bias current of the VLD. When biasing the VLD at 70 mA, the generated white light with a CCT of 5068 K and a CRI of 65 achieves 4.4-Gbit/s raw data rate. In comparison with the YAG:Ce fluorescent phosphor, adhering the LuAG:Ce/CASN:Eu fluorescent phosphor to the VLD ensures the delivered white light a warm-white color with lower CCT, comparable transmission capacity and higher CRI. To achieve a CRI of 2700 K, the VLD illuminating 0.85-mm-thick LuAG:Ce/CASN:Eu phosphor doped diffuser is required to increase its bias current to 122 mA, and the generated white light with a CRI of 87.9 can support a raw data rate of 2.4 Gbit/s. The thinner LuAG:Ce/CASN:Eu fluorescent phosphor enables the white light to exhibit higher CCT, CRI and transmission capacity. At the same bias current of 70 mA, the VLD illuminating 0.75-mm-thick LuAG:Ce/CASN:Eu fluorescent phosphor provides the largest transmission capacity of 4.8 Gbit/s, the highest CRI of 81.1 and the lowest CCT of 2235 K when comparing with other cases. For the LuAG:Ce/CASN:Eu phosphorous diffusers with different thicknesses, the thinner one has higher SNR to achieve higher data rate than the thicker one. To increase the CCT to 3000 K for competing with the commercial incandescent light bulb, the VLD needs to enlarge its bias current to 105 mA for the generated white light with its CRI enlarging up to 91.5 at a cost of slightly decreasing the transmission capacity to 4.4 Gbit/s. The use of VLD combined with the different fluorescent phosphors and the pre-leveling technique effectively implements the low-CCT and high-CRI white-lighting communication, which is a powerful candidate for next-generation indoor lighting industries.

Methods
The infrastructures for implementing the VLD based PtP and white-lighting communications were illustrated in Fig. 5, in which the used VLD with a TO-can package exhibits a peak wavelength of 404 nm, a threshold current (I th ) of 34 mA and a power-to-current slope of 1.23 W/A (I bias > I th ). A pair of plano-convex lenses with the same focal length of 10 mm and numerical aperture of 0.3 helps to control the laser beam divergence, the collimation during free-space transmission and the convergence of beam prior to the receiver. A Si p-i-n photodiode (PD, Thorlabs, FDS025) with a 3-dB bandwidth of 3.3 GHz was employed for PtP data receiving. For white-light generation, either the YAG:Ce 56 or the LuAG:Ce/CASN:Eu 75 luminescent phosphor was individually adhered to the VLD, and an avalanche photodiode (APD, Hamamatsu, S12023-02) with a cut-off frequency of 1 GHz was used for receiving the optical data. To suppress the parasitic capacitance and inductance induced declination on frequency response, the pin lengths of the used VLD and PDs are shortened to 3 mm before connecting with the SubMiniature version A (SMA) jacks. A homemade temperature controlling system with electrical isolation was assembled to stabilize the output dynamics of the VLD and PDs 54 . For directly encoding the VLD, the electrical QAM-OFDM data with various bandwidths are delivered by importing a homemade MATLAB programing data into an arbitrary waveform generator (AWG, Tektronix, 70001 A) with a sampling rate of 24 GSa/s. Afterwards, the output waveform after passing through an electrical amplifier (Amp, Picosecond 5828 A with a gain of 10 dB for the PtP case and Picosecond 5865 with a gain of 26 dB for the white-lighting case) was carried by the VLD for data transmission. The optical transmission distances for the PtP and white-lighting communications were set as 7 and 0.5 m, respectively. After receiving with the p-i-n PD for PtP case and the APD for white-lighting case, the optoelectrically converted QAM-OFDM data was amplified by another amplifier with 18-dB gain (New focus, 1422) for PtP case and 40-dB gain (Mini-circuit, ZKL-1R5+) for white-lighting case, respectively. Then the amplified waveform was captured by a 100-GSa/s digital serial analyzer (DSA, Tektronix, 71604 C), and a decoded MATLAB program was employed to evaluate the QAM-OFDM data qualities including constellation plot, EVM, SNR and BER.