- Email: [email protected]
The design of rapid turbidity measurement system based on single photon detection techniques
Optics & Laser Technology 73 (2015) 44–49
Contents lists available at ScienceDirect
Optics & Laser Technology journal homepage: www.elsevier.com/locate/optlastec
The design of rapid turbidity measurement system based on single photon detection techniques Yixin Yang a,d, Huanqin Wang a,n, Yangyang Cao a,d, Huaqiao Gui b, Jianguo Liu b, Liang Lu c, Huibin Cao b, Tongzhu Yu b, Hui You a a State Key Laboratory of Transducer Technology, Institute of Intelligent Machines, Chinese Academy of Sciences, 350 Shu Shang Hu Road, Hefei, Anhui 230031, China b Key Laboratory of Environmental Optics and Technology, Anhui Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, 350 Shu Shang Hu Road, Hefei, Anhui 230031, China c Key Laboratory of Opto-Electronic Information Acquisition and Manipulation of Ministry of Education, Anhui University, Jiulong Road 111#, Hefei 230601, China d Department of Automation, University of Science and Technology of China, 96 Jin Zhai Road, Hefei 230026, China
art ic l e i nf o
a b s t r a c t
Article history: Received 8 January 2015 Received in revised form 30 March 2015 Accepted 12 April 2015 Available online 14 May 2015
A new rapid turbidity measurement system has been developed to measure the turbidity of drinking water. To determinate the turbidity quantitatively, the total intensity of scattering light has been measured and quantified as number of photons by adopting the single photon detection techniques (SPDT) which has the advantage of high sensitivity. On the basis of SPDT, the measurement system has been built and series of experiments have been carried out. Combining then the 90° Mie scattering theory with the principle of SPDT, a turbidity measurement model has been proposed to explain the experimental results. The experimental results show that a turbidity, which is as low as 0.1 NTU (Nephelometric Turbidity Units), can be measured steadily within 100 ms. It also shows a good linearity and stability over the range of 0.1–400 NTU and the precision can be controlled within 5% full scale. In order to improve its precision and stability, some key parameters, including the sampling time and incident light intensity, have been discussed. It has been proved that, to guarantee an excellent system performance, a good compromise between the measurement speed and the low power consumption should be considered adequately depending on the practical applications. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Turbidity Scattering Photodetection
1. Introduction Turbidity is a measure of the degree to which the water loses its transparency due to the presence of suspended particulates [1]. These suspended particulates, which induce the turbidity in water, are the potential insecurity factors for human health, because of their adsorption of bacteria, viruses, parasites and many other toxic organic compounds and pesticides. The World Health Organization (WHO) establishes that the turbidity of drinking water should not be more than 5 NTU (Nephelometric Turbidity Units), and should ideally be below 1 NTU. Therefore, the demand for low turbidity of water has increased and the method of fast measurement draws more and more attention in the world. It is well-known that the Mie scattering has been widely employed in turbidity measurement. The scattering intensities will be very low for media with low turbidity according to the principle of Mie n
Corresponding author. Tel.: þ 86 551 65591137; fax: þ 86 551 65592420. E-mail address: [email protected] (H. Wang).
http://dx.doi.org/10.1016/j.optlastec.2015.04.005 0030-3992/& 2015 Elsevier Ltd. All rights reserved.
scattering. In the traditional turbidity measurement system, there are four types of detectors usually adopted: photomultiplier tubes, vacuum photodiodes, silicon photodiodes, and cadmium sulfide photoconductors [2]. Since their disadvantages of low sensitivity, lots of time has to be spent in waiting for the response of system, accumulating weak signal and averaging multi-measurements to ensure accurate measurements. For example, the precision of 1720E (Hach) is 75% over the range of 40–100 NTU, while the response time is 60 s and the minimum measurement time is 6 s [9]. Recently, some new techniques have also been introduced in the turbidity measurement. Fiber-optic probes, for instance, are commonly used for in-situ measurement. A. Kramer and Th.A. Paul have analyzed the performance of various fiber-optic probe designs, in which the light exits under an angle of 10–20° [3] and its precision is 70.1 NTU over the range of 0.1–60 NTU. The application of fiber optic has presented a flexible measurement, allowing measurements to be made online. However, when turbidity is higher than 60 NTU, the probe sensitivity decreases gradually which will result in nonlinearity. Although a bi-exponential calibration fit function is well suited to describe the relation of
Y. Yang et al. / Optics & Laser Technology 73 (2015) 44–49
backscattered signal versus turbidity, the number of fit parameters has to be reduced to facilitate calibration. In our previous works [4], a real-time turbidimeter based on time-correlated single photon counting (TCSPC) was developed to measure low level turbidity for drinking water. The drawback of the design is its nonlinearity at the range of 0.1–400 NTU. In addition, in order to obtain the turbidity, a statistical histogram should be obtained by sorting the echo photons according to their arrival time, and the peak of the histogram should be calculated. The complex signal processing method is not convenient for engineering applications, especially for the fast on-line monitoring. Moreover, the electronics for sorting photons by arrival time is costly. In this paper, a rapidly straightforward turbidity measurement method for drinking water was proposed on the basis of the single photon detection techniques (SPDT). Since the Geiger mode APD has advantages of high sensitivity and short response time compared with the normal detectors, a high signal to noise ratio (SNR) can be acquired in a short measurement time. Therefore, high precision can be guaranteed while maintaining rapid measurement. Owing to the quantum statistical properties of SPDT, the fluctuation of scattering intensity can be reduced in the process of measurement, which can finally improve the stability of turbidity measurement. The experimental results show that a good linearity can be guaranteed over the range of 0.1–400 NTU, which significantly reduces calibration effort (enabling 2 points calibration). Moreover, the signal processing method is concise. The turbidity can be calculated by quantifying the total intensity of 90° Mie scattering light as the number of photons and utilizing a linear conversion model. In a word, the measurement system has the advantages of rapidity, high precision, high stability and linearity.
2. Principle and setup 2.1. The principle of SPDT based turbidity measurement system The turbidity measurement principle consists of the 90° Mie scattering and the SPDT. The 90° Mie scattering, which has been identified as the turbidity quantitative determination method by the international water quality standard ISO7027 [5], has been adopted in our turbidity measurement system. In this paper, an 850 nm low power semiconductor laser, which is used as the light source, has been modulated and its modulated light has been projected into measured medium. According to Mie scattering theory, when the distance between the particle and the test point is r, the total intensity of the scattering light Ir can be expressed as [6]
Ir = T
I0 r
2
σ=T
I0λ2 ⎡ ⎣i1(k, m, θ ) + i2(k, m, θ )⎤⎦ 8π r 2
Ir = KS TI0
light has been measured and quantified as the number of photons by avalanche photodiode (APD) in the Geiger model. Moreover, by utilizing the high avalanche gain inside device, the Geiger mode APD can directly generate large amplitude signal without multistage amplification, which has advantages of shot-noise-limited detection of single-photon events [8]. In addition, the lack of A/D conversions and multistage amplification can remove quantization errors and the usual non-idealities associated with these components. As shown in Fig. 1, the simplest photon counter consists of a detector, followed by a discriminator and a counter [7]. In this paper, APD is used as a photoelectric detector. A series of singlephoton pulses have been produced by the detector when a faint ray of light illuminates to the detector and the avalanche effect occurred. A pre-amplifier can, but need not, be used behind the detector to obtain pulses of sufficient amplitude at the discriminator input. The discriminator with an adjustable threshold, which is set to discriminate the single-photon pulses against the background noise, has been adopted in the photon counter. The discriminator threshold is set well above the noise level, but below the peak amplitude of the photon pulses delivered by the detector. When a single-photon pulse exceeds the selected threshold, the discriminator delivers a pulse of a defined duration and a defined logic level. The discriminator output pulses are counted by the subsequent counter. The photons are acquired for a given time interval, after which the result is read from the counter. And then, a faint ray of 90° Mie scattering light has been quantified as the number of photons. At last, the turbidity of the sample T can be calculated by Eq. (2). 2.2. Experiment setup To implement the turbidity measurement principle, the experimental setup has been built and its block diagram is shown in Fig. 2. In the turbidity measurement system, an 850 nm laser diode (L850P010, Thorlabs) has been modulated (fs ¼1 MHz, square waves, duty ratio¼20%) by a pulse driving circuit, and then pulse laser has been generated. The synchronous signal, which is so called as start-signal, has been generated by the oscillator at the same time. The pulse laser has been projected into sample cell
(1)
where Ir is the intensity of scattering light, I0 is the intensity of incident light, λ is the wavelength of the incident wave, σ is the scattering coefficient of a single particle, i1(k,m,θ) and i2(k,m,θ) are scattering light intensity functions, T is the turbidity of the sample, θ is the angle between the scattering light and the incident light, r is the distance between the particle and the test point of the scattering intensity, that is the optical path of the scattering light. When the angle between the scattering light and the incident light θ is 90°, the Eq. (1) can be simplified as
Fig. 1. The block diagram of SPDT based signal processing [7].
(2)
where Ks is the proportionality coefficient. Thanks to the SPDT, which is rarely reported in the field of turbidity measurement, the total intensity of 90° Mie scattering
45
Fig. 2. The block diagram of SPDT based turbidity measurement system.
46
Y. Yang et al. / Optics & Laser Technology 73 (2015) 44–49
PS = k‵ΩIr = k‵ΩKS I0⋅T = k ‵‵ T
(4)
where k ‵‵ = k‵ΩKS I0 . Therefore, the photon counting of the scattering light NS can be obtained by inserting Eq. (3) into Eq. (4), and be stated as
NS =
ηλk ‵‵ tS T = k ‵‵‵⋅T ⋅tS mhc
(5)
The symbol k ‵‵‵ used in the equation is defined k ‵‵‵ = ηλk ‵‵/m⋅hc = ηλk‵ΩKS I0/m⋅hc . Assuming that the photon counting of noise in unit time Nb is a constant, and then the background noise photon counting NB can be expressed as
NB = Nb tS
Fig. 3. The experimental setup.
with turbidity solution by a special aspherical lens, and been scattered by turbidity solution. The 90° Mie scattering light which together with background light has been detected by an APDbased single-photon detector (SPCM-AQRH-14, PerkinElmer) with low dark count (no more than 30 counts). The width of the output signal from the detector, which is so called as stop-signal, is 15 ns. In order to improve the SNR, before the detector, an 850 nm narrowband pass filter with half bandwidth of 10 nm has been used to filter the background light and an occluder with 1 mm aperture diameter has been adopted to limit the field of view. Besides, all light paths are in the dark box, although the near-infrared light (850 nm) just occupies a small proportion in the ambient light. A FPGA chip (EP2C8Q208N, Alter) has been adopted as a counter and the count time can be adjusted suitably via programming. By counting the rise edge of the stop-signal, the light detected by an APD-based single-photon detector has been quantified as the number of photons. And then, the photon count value has been sent to the PC for display by the RS232 interface. The picture of experimental setup is shown in Fig. 3. In this paper, we calibrate the system by using the standard formazin turbidity solution (particle size is 0.6 μm) and different turbidity samples can be obtained by diluting 400 NTU standard formazin turbidity solution.
3. Model and experiment results 3.1. Turbidity measurement model based on SPDT
(6)
The total photon counting of the scattering light can be expressed as
N = NS + NB
(7)
It can be simplified using Eqs. (5) and (6):
N = (k ‵‵‵⋅T + Nb ) tS
Since the photon counting of noise in unit time Nb is a constant, the background noise photon counting NB (equal to Nb tS ) will be a constant when different turbidity of samples have the same sampling time (tS ). For the different turbidity of sample Ti , Eq. (8) can be simplified as
mhc
[Ti] = ηλk‵ΩK I t [Ni] + N0 = K ·[Ni] + N0 S 0 s
(mNS )⋅hc /λ ηtS
3.2. Experiment result 3.2.1. Linearity On the basis of the above setup and model, the experiment (I) has been done to show its linearity over the range of 40– 400 NTU when the measurement interval is 40 NTU and the sampling time is 100 ms. To testify the stability and repeatability of the proposed system, 1000 repeated tests have been done. The results are shown in Fig. 4.
(3)
where tS is the sampling time when the photon count value is NS , and η is the efficiency of detector. Moreover, the power of scattered light PS is related to the scattered light intensity Ir , the solid angle Ω and the conversion factor between luminous flux and light power k‵. When the incident light is I0, and the turbidity of the sample is T , the scattering light power can be stated as
(9)
where Ni is the total photon counting of the scattering light for the sample i , and Ti is the turbidity of sample i , [Ni ] and [Ti ] are the matrixes of Ni and Ti respectively. The symbol K used in the equation is defined as K = mhc /ηλk‵ΩKSI0ts , and the symbol N0 used in the equation is defined as N0 = − Nb ⋅mhc /ηλk‵ΩKS I0 .
A turbidity measurement model, which combines the 90° Mie scattering theory with the theory of SPDT, is established. The SPDT considers the light composed of many photons. The relationship between the energy of a single photon ε and the frequency of light υ can be expressed as ε = hυ = hc/λ . According to our previous works [4], assuming that the energy of m photons can generate one photon count pulse by the detector, then the optical power corresponding to NS received photon counting pulses can be calculated by
PS =
(8)
Fig. 4. The result of experiment (I) and its linear fitting.
Y. Yang et al. / Optics & Laser Technology 73 (2015) 44–49
47
Fig. 5. The result of experiment (II) and its linear fitting: (a) the linear fitting of experiment (II), and (b) the absolute error of experiment (II).
As shown in Fig. 4, the results of the 1000 repeated tests have been shown by red circles. A linear fitting between the turbidity of sample and the photon counting has been done. As shown in Fig. 4, the photon counting is increasing constantly with the increase of turbidity. Moreover, the functional form after fitting is y = a + bx , the value of b is negative, the Adj. R2 is 0.99016 and Pearson's correlation coefficient is 0.99507. It shows an excellent fitting result and good linearity, which demonstrates the correctness of the model we proposed. In addition, the stability and repeatability of the proposed system have been testified by the 1000 repeated tests, since the absolute error can be controlled within 5% full scale over the range of 40–400 NTU. 3.2.2. Precision Since we are mainly focusing on the turbidity of drinking water, which should ideally be below 1 NTU, the precision is tested in the regime of low turbidity (0.1–1 NTU). In experiment (II), the measurement interval is 0.1 NTU and the sampling time is also 100 ms. It is worth mentioning that the system precision has been defined as the absolute error. The results are shown in Fig. 5. As shown in Fig. 5(a), the results of 1000 repeated tests have been shown by red circles. A linear fitting between the turbidity of sample and the photon counting has been done. In Fig. 5(b), the xaxis and the y-axis represent the turbidity measurement points and the absolute error, respectively. As shown in Fig. 5, the Adj. R2 is 0.98355 and Pearson's correlation coefficient is 0.99174. It shows an excellent fitting result and good linearity. Additionally, the stability and repeatability of the proposed system have been testified by the 1000 repeated tests and the maximum mean error is less than 0.04 NTU. It is indicated that a turbidity, which is as low as 0.1 NTU, can be measured steadily within 100 ms.
4. The system performance The performance of the turbidity measurement system is reflected in several factors, among which we mainly focus on the precision and the stability in this paper. According to Eq. (9), the absolute error of turbidity sample ΔT, which represents the measurement precision, can be expressed as
ΔT = K ⋅ΔN =
m⋅hc ⋅ΔN ηλk‵ΩKS I0 ts
(10)
where ΔN is the absolute error of the photon counting. The uncertainty of the turbidity sample μT , which is a representative of the system’s measurement stability, can be expressed as:
μT =
K ⋅μN =
m⋅hc ⋅μN ηλk‵ΩKS I0 ts
(11)
where μN is the uncertainty of the photon counting. According to Eqs. (10) and (11), to improve the precision and stability of the system, the value of K should be decreased. As is shown in Eq. (9), the value of K can be decreased by increasing the sampling time ts and incident light intensity I0. 4.1. The sampling time Experiment (III) has been added in the regime of low turbidity (0.1–1 NTU) to demonstrate the benefits of increasing sampling time on system performance. In this experiment, the sampling time ts is 2 ms, 10 ms, 20 ms, 40 ms, 60 ms, 80 ms, 100 ms, 120 ms, 140 ms, 160 ms, 180 ms, and 200 ms, while maintaining the same light intensity. The results have been shown in Fig. 6. As shown in Fig. 6(a), the x-axis and the y-axis represent the photon counting and the turbidity measurement points, respectively. The linear fitting has been done on the 1000 repeated tests. As shown in Fig. 6(b), the x-axis and the y-axis represent the sampling time and the value of K, respectively. A nonlinear fitting has been done on the value of K according to Eq. (9). The Adj. R2 is 1 and the value of b is 0.9929, which show an excellent fitting result, demonstrate the correctness of the proposed model. Hence, it shows the benefits of increasing sampling time on decrease the value of K. However, the measurement speed is an important parameter on the practical applications. On one hand, faster measurement speed can be achieved when lower measurement precision and poor stability can be accepted on some practical applications. On the other hand, a better performance of the system can be obtained when more sampling time consumed. Hence, a trade-off between the measurement speed and the system performance should be considered adequately depending on the practical applications.
48
Y. Yang et al. / Optics & Laser Technology 73 (2015) 44–49
4.2. The intensity of incident light Experiment (IV) has been added in the regime of low turbidity (0.1–1 NTU) to demonstrate the benefits of increasing the incident light intensity on system performance. In experiment (IV), eight different intensities of the incident light I0 have been generated by adopting a circular variable neutral density filter which attenuate light as the filter is rotated. The eight kinds of relative incident light intensity are 0.07, 0.12, 0.17, 0.24, 0.37, 0.53, 0.89 and 1, while maintaining the same sampling time (ts ¼100 ms). The results have been shown in Fig. 7. As shown in Fig. 7(a), the linear fitting has been done on the 1000 repeated tests. As shown in Fig. 7(b), a nonlinear fitting has been done on the value of K. The Adj. R2 is 0.99877 and the value of b is 0.9745, which show an excellent fitting result, demonstrate the correctness of the proposed model. Hence, the value of K can be decreased by increasing the incident light intensity I0. However, to obtain a better performance, we cannot increase the light intensity infinitely. As we all know, a very important parameter of the APD-based single-photon detector is dead-time which means once triggered APD must be reset to enable the detection of another event. Assuming that more than two photons successively arrive to the APD-based single-photon detector
within the reset-time a.k.a. dead-time, there is only one photon that can be recorded according to the nature of APD-based singlephoton detector itself. This phenomenon will appear when the intensity of 90° Mie scattering light is too strong which is caused by high turbidity. In this case, the system performance cannot be improved and end up in poor performance. That is to say, a better performance can be obtained by setting a stronger intensity of incident light in the regime of low turbidity. Meanwhile, the strong intensity of incident light will result in a poor performance in the regime of high turbidity because of dead-time. Therefore, we can conclude that the dynamic range of turbidity measurement will be degraded with the increase of the incident light intensity. Moreover, although a better performance can be obtained when increasing the incident light intensity properly, it is quite uneconomic for us to improve the system performance by increasing the intensity of the incident light I0. Nowadays, the low power design is very popular in some special applications, such as hand-held turbiditymeter or remote online monitoring under a harsh environment. Hence, a trade-off between the low power consumption and the performance of the system should also be considered adequately depending on the practical applications. All in all, when the low power consumption is more important than the measurement speed, the performance of the system can
Fig. 6. The result of experiment (III). (a) The results of 1000 repeated tests while the sampling time ts ¼ 2 ms, 10 ms, 20 ms, 40 ms, 60 ms, 80 ms, 100 ms, 120 ms, 140 ms, 160 ms, 180 ms, and 200 ms, and (b) the value of K versus the sampling time.
Fig. 7. The result of experiment (IV). (a) The results of 1000 repeated tests while the relative intensity of incident light I0 ¼0.07, 0.12, 0.17, 0.24, 0.37, 0.53, 0.89 and 1, and (b) the value of K versus the relative intensity of the incident light I0.
Y. Yang et al. / Optics & Laser Technology 73 (2015) 44–49
be improved by increasing the sampling time properly. On the other hand, the performance of the system can be improved by increasing the incident light intensity properly, when the measurement speed is more important. That is to say, to guarantee an excellent system performance, a good compromise between the measurement speed and the low power consumption should be considered adequately depending on the practical applications.
5. Conclusions In this paper, a new rapid online turbidity measuring system based on SPDT has been developed for turbidity measurement. The excellent performances of the system, such as linearity, stability, speedability and high precision, have been well demonstrated by series of repeated experiments. The experimental results show that a turbidity, which is as low as 0.1 NTU, can be measured steadily within 100 ms. It also shows a good linearity and stability over a dynamic range of 0.1–400 NTU and the absolute error can be controlled within 5% full scale. Although these values are not particularly impressive, the solution we propose is in satisfactory agreement with the requirements of many industrial applications, especially for monitoring of drinking water. Based on the theoretical analysis, which combined the Mie theory with the principle of SPDT, a turbidity measurement model has been proposed to explain the experimental results. On the basis of the proposed model, the performance of the system has been discussed and it has been proved that the performance of the system can be improved by increasing the sampling time properly, when the low power consumption is more important than the measurement speed. On the other hand, the performance of the system can also be improved by increasing the incident light intensity properly, when the measurement speed is more important. That is to say, to guarantee an excellent system performance, a good compromise between the measurement speed and the low power consumption should be considered adequately depending on the practical applications.
Acknowledgments This work was financially supported by the National Natural Science Foundation of China (Grants 61201401 and 61176105) and National Science and Technology Major Project (No. 2011ZX05051004). The authors wish to thank Mr. Zhe Huang (Dept. of Automation, University of Science and Technology of China) and Dr. Deyong He
49
(Key Laboratory of Quantum Information, CAS, at University of Science and Technology of China) for their valuable contributions and discussions.
References [1] 〈http://www.lenntech.com/turbidity.htm#ixzz2ihAkFioK〉. [2] Sadar MJ. Understanding turbidity science. 〈http://www.hach.com/1720e-lowrange-process-turbidimeter-turbidity-sensor-only/product-downloads? id ¼7640457219&callback¼ bc〉; 2013. [3] Kramer A, Paul T. Fiber-optic probes as sensors for diffuse backscattering. In: Advanced Photonics & Renewable Energy, OSA Technical Digest (CD) (Optical Society of America, 2010), paper SThD2. [4] H. Wang, Y. Yang, Z. Huang, H. Gui. Instrument for real-time measurement of low turbidity by using time-correlated single photon counting technique. Instrum Meas IEEE Trans 2015;64(4):1075–83. [5] International Organization for Standardization, ISO 7027: water quality – determination of turbidity. Available online at: 〈http://www.iso.org/iso/ catalogue_detail?csnumber ¼ 30123〉. [6] Burlingame GA, Pickel MJ, Roman JT. Practical applications of turbidity monitoring. J AWWA 1998;90(8):57–69. [7] Becker Wolfgang. Overview of photon counting techniques. 1st ed.. Heidelberg, Germany: Advanced Time-Correlated Single Photon Counting Techniques; 2005. p. 20–1. [8] Degnan JJ. Photon-counting multikilohertz microlaser altimeters for airborne and spaceborne topographic measurements. J Geodyn 2002;34:503–49. [9] Hach Company. 1720E Turbidimeter user manual. Available online at: 〈http:// www.hach.com/1720e-low-range-process-turbidimeter-turbidity-sensor-only/ product-downloads?id ¼ 7640457219&callback¼ bc〉; 2013.
Yixin Yang is pursuing a master's degree in engineering at the University of Science and Technology of China. He received his B.S. degrees in measuring and controlling technology and instrument from the China Jiliang University in 2012. His current research interests mainly focus on optical sensing, including single photon detection, water quality monitoring and laser ranging finder.
Huanqin Wang was born in Yongzhou, Hunan, China, in 1982. He received his B.S. degree in applied physics in 2004 and the Ph.D. degree in microelectronics and solid state electronics in 2009, both from University of Science and Technology of China. He is currently an Associate Professor with State Key Laboratory of Transducer Technology, Institute of Intelligent Machine, Chinese Academy of Sciences. His research interests mainly focus on optical sensing, including particle monitoring, laser ranging finder, 3D imaging, and single photon detection. He was in charge of several projects from National Natural Science Fund of China, Strategic Priority Research Program of the Chinese Academy of Sciences, etc. He has published more than 30 papers and applied 27 patents.
Contents lists available at ScienceDirect
Optics & Laser Technology journal homepage: www.elsevier.com/locate/optlastec
The design of rapid turbidity measurement system based on single photon detection techniques Yixin Yang a,d, Huanqin Wang a,n, Yangyang Cao a,d, Huaqiao Gui b, Jianguo Liu b, Liang Lu c, Huibin Cao b, Tongzhu Yu b, Hui You a a State Key Laboratory of Transducer Technology, Institute of Intelligent Machines, Chinese Academy of Sciences, 350 Shu Shang Hu Road, Hefei, Anhui 230031, China b Key Laboratory of Environmental Optics and Technology, Anhui Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, 350 Shu Shang Hu Road, Hefei, Anhui 230031, China c Key Laboratory of Opto-Electronic Information Acquisition and Manipulation of Ministry of Education, Anhui University, Jiulong Road 111#, Hefei 230601, China d Department of Automation, University of Science and Technology of China, 96 Jin Zhai Road, Hefei 230026, China
art ic l e i nf o
a b s t r a c t
Article history: Received 8 January 2015 Received in revised form 30 March 2015 Accepted 12 April 2015 Available online 14 May 2015
A new rapid turbidity measurement system has been developed to measure the turbidity of drinking water. To determinate the turbidity quantitatively, the total intensity of scattering light has been measured and quantified as number of photons by adopting the single photon detection techniques (SPDT) which has the advantage of high sensitivity. On the basis of SPDT, the measurement system has been built and series of experiments have been carried out. Combining then the 90° Mie scattering theory with the principle of SPDT, a turbidity measurement model has been proposed to explain the experimental results. The experimental results show that a turbidity, which is as low as 0.1 NTU (Nephelometric Turbidity Units), can be measured steadily within 100 ms. It also shows a good linearity and stability over the range of 0.1–400 NTU and the precision can be controlled within 5% full scale. In order to improve its precision and stability, some key parameters, including the sampling time and incident light intensity, have been discussed. It has been proved that, to guarantee an excellent system performance, a good compromise between the measurement speed and the low power consumption should be considered adequately depending on the practical applications. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Turbidity Scattering Photodetection
1. Introduction Turbidity is a measure of the degree to which the water loses its transparency due to the presence of suspended particulates [1]. These suspended particulates, which induce the turbidity in water, are the potential insecurity factors for human health, because of their adsorption of bacteria, viruses, parasites and many other toxic organic compounds and pesticides. The World Health Organization (WHO) establishes that the turbidity of drinking water should not be more than 5 NTU (Nephelometric Turbidity Units), and should ideally be below 1 NTU. Therefore, the demand for low turbidity of water has increased and the method of fast measurement draws more and more attention in the world. It is well-known that the Mie scattering has been widely employed in turbidity measurement. The scattering intensities will be very low for media with low turbidity according to the principle of Mie n
Corresponding author. Tel.: þ 86 551 65591137; fax: þ 86 551 65592420. E-mail address: [email protected] (H. Wang).
http://dx.doi.org/10.1016/j.optlastec.2015.04.005 0030-3992/& 2015 Elsevier Ltd. All rights reserved.
scattering. In the traditional turbidity measurement system, there are four types of detectors usually adopted: photomultiplier tubes, vacuum photodiodes, silicon photodiodes, and cadmium sulfide photoconductors [2]. Since their disadvantages of low sensitivity, lots of time has to be spent in waiting for the response of system, accumulating weak signal and averaging multi-measurements to ensure accurate measurements. For example, the precision of 1720E (Hach) is 75% over the range of 40–100 NTU, while the response time is 60 s and the minimum measurement time is 6 s [9]. Recently, some new techniques have also been introduced in the turbidity measurement. Fiber-optic probes, for instance, are commonly used for in-situ measurement. A. Kramer and Th.A. Paul have analyzed the performance of various fiber-optic probe designs, in which the light exits under an angle of 10–20° [3] and its precision is 70.1 NTU over the range of 0.1–60 NTU. The application of fiber optic has presented a flexible measurement, allowing measurements to be made online. However, when turbidity is higher than 60 NTU, the probe sensitivity decreases gradually which will result in nonlinearity. Although a bi-exponential calibration fit function is well suited to describe the relation of
Y. Yang et al. / Optics & Laser Technology 73 (2015) 44–49
backscattered signal versus turbidity, the number of fit parameters has to be reduced to facilitate calibration. In our previous works [4], a real-time turbidimeter based on time-correlated single photon counting (TCSPC) was developed to measure low level turbidity for drinking water. The drawback of the design is its nonlinearity at the range of 0.1–400 NTU. In addition, in order to obtain the turbidity, a statistical histogram should be obtained by sorting the echo photons according to their arrival time, and the peak of the histogram should be calculated. The complex signal processing method is not convenient for engineering applications, especially for the fast on-line monitoring. Moreover, the electronics for sorting photons by arrival time is costly. In this paper, a rapidly straightforward turbidity measurement method for drinking water was proposed on the basis of the single photon detection techniques (SPDT). Since the Geiger mode APD has advantages of high sensitivity and short response time compared with the normal detectors, a high signal to noise ratio (SNR) can be acquired in a short measurement time. Therefore, high precision can be guaranteed while maintaining rapid measurement. Owing to the quantum statistical properties of SPDT, the fluctuation of scattering intensity can be reduced in the process of measurement, which can finally improve the stability of turbidity measurement. The experimental results show that a good linearity can be guaranteed over the range of 0.1–400 NTU, which significantly reduces calibration effort (enabling 2 points calibration). Moreover, the signal processing method is concise. The turbidity can be calculated by quantifying the total intensity of 90° Mie scattering light as the number of photons and utilizing a linear conversion model. In a word, the measurement system has the advantages of rapidity, high precision, high stability and linearity.
2. Principle and setup 2.1. The principle of SPDT based turbidity measurement system The turbidity measurement principle consists of the 90° Mie scattering and the SPDT. The 90° Mie scattering, which has been identified as the turbidity quantitative determination method by the international water quality standard ISO7027 [5], has been adopted in our turbidity measurement system. In this paper, an 850 nm low power semiconductor laser, which is used as the light source, has been modulated and its modulated light has been projected into measured medium. According to Mie scattering theory, when the distance between the particle and the test point is r, the total intensity of the scattering light Ir can be expressed as [6]
Ir = T
I0 r
2
σ=T
I0λ2 ⎡ ⎣i1(k, m, θ ) + i2(k, m, θ )⎤⎦ 8π r 2
Ir = KS TI0
light has been measured and quantified as the number of photons by avalanche photodiode (APD) in the Geiger model. Moreover, by utilizing the high avalanche gain inside device, the Geiger mode APD can directly generate large amplitude signal without multistage amplification, which has advantages of shot-noise-limited detection of single-photon events [8]. In addition, the lack of A/D conversions and multistage amplification can remove quantization errors and the usual non-idealities associated with these components. As shown in Fig. 1, the simplest photon counter consists of a detector, followed by a discriminator and a counter [7]. In this paper, APD is used as a photoelectric detector. A series of singlephoton pulses have been produced by the detector when a faint ray of light illuminates to the detector and the avalanche effect occurred. A pre-amplifier can, but need not, be used behind the detector to obtain pulses of sufficient amplitude at the discriminator input. The discriminator with an adjustable threshold, which is set to discriminate the single-photon pulses against the background noise, has been adopted in the photon counter. The discriminator threshold is set well above the noise level, but below the peak amplitude of the photon pulses delivered by the detector. When a single-photon pulse exceeds the selected threshold, the discriminator delivers a pulse of a defined duration and a defined logic level. The discriminator output pulses are counted by the subsequent counter. The photons are acquired for a given time interval, after which the result is read from the counter. And then, a faint ray of 90° Mie scattering light has been quantified as the number of photons. At last, the turbidity of the sample T can be calculated by Eq. (2). 2.2. Experiment setup To implement the turbidity measurement principle, the experimental setup has been built and its block diagram is shown in Fig. 2. In the turbidity measurement system, an 850 nm laser diode (L850P010, Thorlabs) has been modulated (fs ¼1 MHz, square waves, duty ratio¼20%) by a pulse driving circuit, and then pulse laser has been generated. The synchronous signal, which is so called as start-signal, has been generated by the oscillator at the same time. The pulse laser has been projected into sample cell
(1)
where Ir is the intensity of scattering light, I0 is the intensity of incident light, λ is the wavelength of the incident wave, σ is the scattering coefficient of a single particle, i1(k,m,θ) and i2(k,m,θ) are scattering light intensity functions, T is the turbidity of the sample, θ is the angle between the scattering light and the incident light, r is the distance between the particle and the test point of the scattering intensity, that is the optical path of the scattering light. When the angle between the scattering light and the incident light θ is 90°, the Eq. (1) can be simplified as
Fig. 1. The block diagram of SPDT based signal processing [7].
(2)
where Ks is the proportionality coefficient. Thanks to the SPDT, which is rarely reported in the field of turbidity measurement, the total intensity of 90° Mie scattering
45
Fig. 2. The block diagram of SPDT based turbidity measurement system.
46
Y. Yang et al. / Optics & Laser Technology 73 (2015) 44–49
PS = k‵ΩIr = k‵ΩKS I0⋅T = k ‵‵ T
(4)
where k ‵‵ = k‵ΩKS I0 . Therefore, the photon counting of the scattering light NS can be obtained by inserting Eq. (3) into Eq. (4), and be stated as
NS =
ηλk ‵‵ tS T = k ‵‵‵⋅T ⋅tS mhc
(5)
The symbol k ‵‵‵ used in the equation is defined k ‵‵‵ = ηλk ‵‵/m⋅hc = ηλk‵ΩKS I0/m⋅hc . Assuming that the photon counting of noise in unit time Nb is a constant, and then the background noise photon counting NB can be expressed as
NB = Nb tS
Fig. 3. The experimental setup.
with turbidity solution by a special aspherical lens, and been scattered by turbidity solution. The 90° Mie scattering light which together with background light has been detected by an APDbased single-photon detector (SPCM-AQRH-14, PerkinElmer) with low dark count (no more than 30 counts). The width of the output signal from the detector, which is so called as stop-signal, is 15 ns. In order to improve the SNR, before the detector, an 850 nm narrowband pass filter with half bandwidth of 10 nm has been used to filter the background light and an occluder with 1 mm aperture diameter has been adopted to limit the field of view. Besides, all light paths are in the dark box, although the near-infrared light (850 nm) just occupies a small proportion in the ambient light. A FPGA chip (EP2C8Q208N, Alter) has been adopted as a counter and the count time can be adjusted suitably via programming. By counting the rise edge of the stop-signal, the light detected by an APD-based single-photon detector has been quantified as the number of photons. And then, the photon count value has been sent to the PC for display by the RS232 interface. The picture of experimental setup is shown in Fig. 3. In this paper, we calibrate the system by using the standard formazin turbidity solution (particle size is 0.6 μm) and different turbidity samples can be obtained by diluting 400 NTU standard formazin turbidity solution.
3. Model and experiment results 3.1. Turbidity measurement model based on SPDT
(6)
The total photon counting of the scattering light can be expressed as
N = NS + NB
(7)
It can be simplified using Eqs. (5) and (6):
N = (k ‵‵‵⋅T + Nb ) tS
Since the photon counting of noise in unit time Nb is a constant, the background noise photon counting NB (equal to Nb tS ) will be a constant when different turbidity of samples have the same sampling time (tS ). For the different turbidity of sample Ti , Eq. (8) can be simplified as
mhc
[Ti] = ηλk‵ΩK I t [Ni] + N0 = K ·[Ni] + N0 S 0 s
(mNS )⋅hc /λ ηtS
3.2. Experiment result 3.2.1. Linearity On the basis of the above setup and model, the experiment (I) has been done to show its linearity over the range of 40– 400 NTU when the measurement interval is 40 NTU and the sampling time is 100 ms. To testify the stability and repeatability of the proposed system, 1000 repeated tests have been done. The results are shown in Fig. 4.
(3)
where tS is the sampling time when the photon count value is NS , and η is the efficiency of detector. Moreover, the power of scattered light PS is related to the scattered light intensity Ir , the solid angle Ω and the conversion factor between luminous flux and light power k‵. When the incident light is I0, and the turbidity of the sample is T , the scattering light power can be stated as
(9)
where Ni is the total photon counting of the scattering light for the sample i , and Ti is the turbidity of sample i , [Ni ] and [Ti ] are the matrixes of Ni and Ti respectively. The symbol K used in the equation is defined as K = mhc /ηλk‵ΩKSI0ts , and the symbol N0 used in the equation is defined as N0 = − Nb ⋅mhc /ηλk‵ΩKS I0 .
A turbidity measurement model, which combines the 90° Mie scattering theory with the theory of SPDT, is established. The SPDT considers the light composed of many photons. The relationship between the energy of a single photon ε and the frequency of light υ can be expressed as ε = hυ = hc/λ . According to our previous works [4], assuming that the energy of m photons can generate one photon count pulse by the detector, then the optical power corresponding to NS received photon counting pulses can be calculated by
PS =
(8)
Fig. 4. The result of experiment (I) and its linear fitting.
Y. Yang et al. / Optics & Laser Technology 73 (2015) 44–49
47
Fig. 5. The result of experiment (II) and its linear fitting: (a) the linear fitting of experiment (II), and (b) the absolute error of experiment (II).
As shown in Fig. 4, the results of the 1000 repeated tests have been shown by red circles. A linear fitting between the turbidity of sample and the photon counting has been done. As shown in Fig. 4, the photon counting is increasing constantly with the increase of turbidity. Moreover, the functional form after fitting is y = a + bx , the value of b is negative, the Adj. R2 is 0.99016 and Pearson's correlation coefficient is 0.99507. It shows an excellent fitting result and good linearity, which demonstrates the correctness of the model we proposed. In addition, the stability and repeatability of the proposed system have been testified by the 1000 repeated tests, since the absolute error can be controlled within 5% full scale over the range of 40–400 NTU. 3.2.2. Precision Since we are mainly focusing on the turbidity of drinking water, which should ideally be below 1 NTU, the precision is tested in the regime of low turbidity (0.1–1 NTU). In experiment (II), the measurement interval is 0.1 NTU and the sampling time is also 100 ms. It is worth mentioning that the system precision has been defined as the absolute error. The results are shown in Fig. 5. As shown in Fig. 5(a), the results of 1000 repeated tests have been shown by red circles. A linear fitting between the turbidity of sample and the photon counting has been done. In Fig. 5(b), the xaxis and the y-axis represent the turbidity measurement points and the absolute error, respectively. As shown in Fig. 5, the Adj. R2 is 0.98355 and Pearson's correlation coefficient is 0.99174. It shows an excellent fitting result and good linearity. Additionally, the stability and repeatability of the proposed system have been testified by the 1000 repeated tests and the maximum mean error is less than 0.04 NTU. It is indicated that a turbidity, which is as low as 0.1 NTU, can be measured steadily within 100 ms.
4. The system performance The performance of the turbidity measurement system is reflected in several factors, among which we mainly focus on the precision and the stability in this paper. According to Eq. (9), the absolute error of turbidity sample ΔT, which represents the measurement precision, can be expressed as
ΔT = K ⋅ΔN =
m⋅hc ⋅ΔN ηλk‵ΩKS I0 ts
(10)
where ΔN is the absolute error of the photon counting. The uncertainty of the turbidity sample μT , which is a representative of the system’s measurement stability, can be expressed as:
μT =
K ⋅μN =
m⋅hc ⋅μN ηλk‵ΩKS I0 ts
(11)
where μN is the uncertainty of the photon counting. According to Eqs. (10) and (11), to improve the precision and stability of the system, the value of K should be decreased. As is shown in Eq. (9), the value of K can be decreased by increasing the sampling time ts and incident light intensity I0. 4.1. The sampling time Experiment (III) has been added in the regime of low turbidity (0.1–1 NTU) to demonstrate the benefits of increasing sampling time on system performance. In this experiment, the sampling time ts is 2 ms, 10 ms, 20 ms, 40 ms, 60 ms, 80 ms, 100 ms, 120 ms, 140 ms, 160 ms, 180 ms, and 200 ms, while maintaining the same light intensity. The results have been shown in Fig. 6. As shown in Fig. 6(a), the x-axis and the y-axis represent the photon counting and the turbidity measurement points, respectively. The linear fitting has been done on the 1000 repeated tests. As shown in Fig. 6(b), the x-axis and the y-axis represent the sampling time and the value of K, respectively. A nonlinear fitting has been done on the value of K according to Eq. (9). The Adj. R2 is 1 and the value of b is 0.9929, which show an excellent fitting result, demonstrate the correctness of the proposed model. Hence, it shows the benefits of increasing sampling time on decrease the value of K. However, the measurement speed is an important parameter on the practical applications. On one hand, faster measurement speed can be achieved when lower measurement precision and poor stability can be accepted on some practical applications. On the other hand, a better performance of the system can be obtained when more sampling time consumed. Hence, a trade-off between the measurement speed and the system performance should be considered adequately depending on the practical applications.
48
Y. Yang et al. / Optics & Laser Technology 73 (2015) 44–49
4.2. The intensity of incident light Experiment (IV) has been added in the regime of low turbidity (0.1–1 NTU) to demonstrate the benefits of increasing the incident light intensity on system performance. In experiment (IV), eight different intensities of the incident light I0 have been generated by adopting a circular variable neutral density filter which attenuate light as the filter is rotated. The eight kinds of relative incident light intensity are 0.07, 0.12, 0.17, 0.24, 0.37, 0.53, 0.89 and 1, while maintaining the same sampling time (ts ¼100 ms). The results have been shown in Fig. 7. As shown in Fig. 7(a), the linear fitting has been done on the 1000 repeated tests. As shown in Fig. 7(b), a nonlinear fitting has been done on the value of K. The Adj. R2 is 0.99877 and the value of b is 0.9745, which show an excellent fitting result, demonstrate the correctness of the proposed model. Hence, the value of K can be decreased by increasing the incident light intensity I0. However, to obtain a better performance, we cannot increase the light intensity infinitely. As we all know, a very important parameter of the APD-based single-photon detector is dead-time which means once triggered APD must be reset to enable the detection of another event. Assuming that more than two photons successively arrive to the APD-based single-photon detector
within the reset-time a.k.a. dead-time, there is only one photon that can be recorded according to the nature of APD-based singlephoton detector itself. This phenomenon will appear when the intensity of 90° Mie scattering light is too strong which is caused by high turbidity. In this case, the system performance cannot be improved and end up in poor performance. That is to say, a better performance can be obtained by setting a stronger intensity of incident light in the regime of low turbidity. Meanwhile, the strong intensity of incident light will result in a poor performance in the regime of high turbidity because of dead-time. Therefore, we can conclude that the dynamic range of turbidity measurement will be degraded with the increase of the incident light intensity. Moreover, although a better performance can be obtained when increasing the incident light intensity properly, it is quite uneconomic for us to improve the system performance by increasing the intensity of the incident light I0. Nowadays, the low power design is very popular in some special applications, such as hand-held turbiditymeter or remote online monitoring under a harsh environment. Hence, a trade-off between the low power consumption and the performance of the system should also be considered adequately depending on the practical applications. All in all, when the low power consumption is more important than the measurement speed, the performance of the system can
Fig. 6. The result of experiment (III). (a) The results of 1000 repeated tests while the sampling time ts ¼ 2 ms, 10 ms, 20 ms, 40 ms, 60 ms, 80 ms, 100 ms, 120 ms, 140 ms, 160 ms, 180 ms, and 200 ms, and (b) the value of K versus the sampling time.
Fig. 7. The result of experiment (IV). (a) The results of 1000 repeated tests while the relative intensity of incident light I0 ¼0.07, 0.12, 0.17, 0.24, 0.37, 0.53, 0.89 and 1, and (b) the value of K versus the relative intensity of the incident light I0.
Y. Yang et al. / Optics & Laser Technology 73 (2015) 44–49
be improved by increasing the sampling time properly. On the other hand, the performance of the system can be improved by increasing the incident light intensity properly, when the measurement speed is more important. That is to say, to guarantee an excellent system performance, a good compromise between the measurement speed and the low power consumption should be considered adequately depending on the practical applications.
5. Conclusions In this paper, a new rapid online turbidity measuring system based on SPDT has been developed for turbidity measurement. The excellent performances of the system, such as linearity, stability, speedability and high precision, have been well demonstrated by series of repeated experiments. The experimental results show that a turbidity, which is as low as 0.1 NTU, can be measured steadily within 100 ms. It also shows a good linearity and stability over a dynamic range of 0.1–400 NTU and the absolute error can be controlled within 5% full scale. Although these values are not particularly impressive, the solution we propose is in satisfactory agreement with the requirements of many industrial applications, especially for monitoring of drinking water. Based on the theoretical analysis, which combined the Mie theory with the principle of SPDT, a turbidity measurement model has been proposed to explain the experimental results. On the basis of the proposed model, the performance of the system has been discussed and it has been proved that the performance of the system can be improved by increasing the sampling time properly, when the low power consumption is more important than the measurement speed. On the other hand, the performance of the system can also be improved by increasing the incident light intensity properly, when the measurement speed is more important. That is to say, to guarantee an excellent system performance, a good compromise between the measurement speed and the low power consumption should be considered adequately depending on the practical applications.
Acknowledgments This work was financially supported by the National Natural Science Foundation of China (Grants 61201401 and 61176105) and National Science and Technology Major Project (No. 2011ZX05051004). The authors wish to thank Mr. Zhe Huang (Dept. of Automation, University of Science and Technology of China) and Dr. Deyong He
49
(Key Laboratory of Quantum Information, CAS, at University of Science and Technology of China) for their valuable contributions and discussions.
References [1] 〈http://www.lenntech.com/turbidity.htm#ixzz2ihAkFioK〉. [2] Sadar MJ. Understanding turbidity science. 〈http://www.hach.com/1720e-lowrange-process-turbidimeter-turbidity-sensor-only/product-downloads? id ¼7640457219&callback¼ bc〉; 2013. [3] Kramer A, Paul T. Fiber-optic probes as sensors for diffuse backscattering. In: Advanced Photonics & Renewable Energy, OSA Technical Digest (CD) (Optical Society of America, 2010), paper SThD2. [4] H. Wang, Y. Yang, Z. Huang, H. Gui. Instrument for real-time measurement of low turbidity by using time-correlated single photon counting technique. Instrum Meas IEEE Trans 2015;64(4):1075–83. [5] International Organization for Standardization, ISO 7027: water quality – determination of turbidity. Available online at: 〈http://www.iso.org/iso/ catalogue_detail?csnumber ¼ 30123〉. [6] Burlingame GA, Pickel MJ, Roman JT. Practical applications of turbidity monitoring. J AWWA 1998;90(8):57–69. [7] Becker Wolfgang. Overview of photon counting techniques. 1st ed.. Heidelberg, Germany: Advanced Time-Correlated Single Photon Counting Techniques; 2005. p. 20–1. [8] Degnan JJ. Photon-counting multikilohertz microlaser altimeters for airborne and spaceborne topographic measurements. J Geodyn 2002;34:503–49. [9] Hach Company. 1720E Turbidimeter user manual. Available online at: 〈http:// www.hach.com/1720e-low-range-process-turbidimeter-turbidity-sensor-only/ product-downloads?id ¼ 7640457219&callback¼ bc〉; 2013.
Yixin Yang is pursuing a master's degree in engineering at the University of Science and Technology of China. He received his B.S. degrees in measuring and controlling technology and instrument from the China Jiliang University in 2012. His current research interests mainly focus on optical sensing, including single photon detection, water quality monitoring and laser ranging finder.
Huanqin Wang was born in Yongzhou, Hunan, China, in 1982. He received his B.S. degree in applied physics in 2004 and the Ph.D. degree in microelectronics and solid state electronics in 2009, both from University of Science and Technology of China. He is currently an Associate Professor with State Key Laboratory of Transducer Technology, Institute of Intelligent Machine, Chinese Academy of Sciences. His research interests mainly focus on optical sensing, including particle monitoring, laser ranging finder, 3D imaging, and single photon detection. He was in charge of several projects from National Natural Science Fund of China, Strategic Priority Research Program of the Chinese Academy of Sciences, etc. He has published more than 30 papers and applied 27 patents.