Camera level (CL) and detection threshold (DT) optimisation
At first, the NPs' detection with the NS300 using different camera level (CL) and detection threshold (DT) settings was optimised. To do so, an aqueous suspension of 102 nm Nanospheres with a particle concentration of 4x108 particles/mL was prepared, analysed with varying CL settings, and processed with three different DT settings (5, 10, 15). The influence of both parameters on the particle concentration is illustrated in Fig. 1.
Figure 1 shows no visible correlation between CL or DT and the particle concentration. A slight underestimation of the particle concentration is generally seen with CL levels 10 and 11. Notably, from CL 11 and up, the particle concentration increases with camera level. Based on the particle 'suspensions' preparation, it was expected that the particle concentration should be within the range of 80–110% of the prepared value, and that is indeed the case for all CL and DT levels, except for the settings CL 13 and 14 with DT 5 where the particle concentration is slightly overestimated. Therefore CL 12 with DT of 5 or 10 were selected as the most optimum settings giving the closest to the prepared particle concentration value with the lower standard deviation (SD) in the determined particle concentration.
The influence of CL and DT on both the mean and mode particle size was also determined using the 102 nm Nanospheres. The reference value for the particle size of this particle was 102 ± 3 nm. The mean particle size provides information about the average particle size, while the mode particle size provides information about the most common particle size. The results for the different CL and DT settings are presented in Figs. 2 and 3.
Unlike concentration, particle size is affected by CL settings. Both Figs. 2 and 3 show a decreasing particle size with increasing CL. Higher CL leads to more exposure to the light scattering of the particles, making it more difficult for the processing software to locate the particle centre and track the particle motion. In practice, the determined particle size was expected to be within 90–110% of the reference value of the particle, which means that all particle sizes between 92 and 112 nm would be acceptable. While CL 9 best agrees with the reference value, CL 10 gives a lower SD for both mean and mode particle size. In Figs. 2 and 3, one can also observe that a higher DT gives a larger particle size; however, this impact is insignificant since the impact is generally more negligible than the SD. Based on the particle concentration and size measurements, the optimal settings for further experiments were chosen as CL 11 in combination with DT 15.
Capturing time (CT) optimisation
The capturing time (CT) may also be of influence the results. Therefore the 102 nm Nanosphere suspension with a particle concentration of 4x108 particle/mL was also analysed with capturing times of 30, 60 and 120 seconds to optimise this parameter. The effect of the capturing time on the determined particle concentration is presented in Fig. 4.
Figure 4 shows an increase in accuracy and precision with increasing capturing time. The increase in precision results from longer capturing time due to a higher analysis volume. Figure 5 overviews the influence of increasing capturing time on both the mean and the mode particle size. The precision of the mean particle size decreases with increasing capturing time with no significant impact on the accuracy. The precision of the mode particle size was highest at a capturing time of 60 seconds. Somewhat surprising, the accuracy of the mode particle size increased with a longer capture time. Regarding both particle size and concentration, a capturing time of 60 seconds was selected for further experiments as a compromise between precision and accuracy.
Sample flow optimisation
As the last parameter, the influence of the sample flow in the flow cell was determined with the flow speed varying between 25, 50 and 75 a.u. Figure 6 shows the influence of the flow speed on the determined particle concentration. Results show that the most accurate concentration is found at a flow speed of 50 a.u. At the time, the precision is best at a flow speed of 75 a.u. This is expected since a higher flow speed increases the total analysed sample volume; however, a too high flow speed makes it harder for the processing software to track the particle motion, which could decrease the number of tracked particles.
An overview of the influence of flow speed on particle size is given in Fig. 7. The impact of flow speed is relatively small and shows no correlation to the particle size in both mean and mode size. For the mean size, the highest accuracy is found at a flow speed of 25 a.u., while for mode size, the highest accuracy was found at a flow speed of 50 a.u. The precision was highest at a flow speed of 75 a.u. for mean particle size and 50 a.u. for mode particle size.
The tested flow speeds have only a minor influence on particle concentration and size. There is no clear correlation between accuracy and precision. A flow speed of 50 a.u. was chosen as optimal because of sample volume reduction compared to a flow of 75 a.u.
Analytical method validation characteristics
Now that the different parameters are optimised, the performance characteristics of the entire method were determined at particle concentrations of 4.0x108 and 2.0x107 particles/mL using the same 102 nm Nanospheres. The accuracy, repeatability (RSDr) and within-laboratory reproducibility (RSDRL) were determined using the low flow-cell top-plate and the O-ring top-plate (no flow). To determine repeatability, measurements were carried out in seven-fold in one day. To determine within-laboratory reproducibility, these measurements were repeated on two more days, totalling three days. One-way ANOVA was performed to determine the accuracy, RSDr and RSDRL, and the results are presented in Table 1.
Table 1
Accuracy, repeatability and within-laboratory reproducibility results with optimised analyses parameters at particle concentrations of 4.0x108 and 2.0x107 particles/mL.
Cell type | Evaluated parameter | Particle concentration [particles/mL] |
4.0x108 | 2.0x107 |
Accuracy [%] | RSDr [%] | RSDRL [%] | Accuracy [%] | RSDr [%] | RSDRL [%] |
Flow-cell 50 [a.u.] | Particle concentration | 84 | 8.4 | 15 | 101 | 34 | 37 |
Mean size | 98 | 0.8 | 1.4 | 102 | 3.4 | 5.7 |
Mode size | 95 | 1.1 | 1.2 | 93 | 4.6 | 4.7 |
O-ring (no flow) | Particle concentration | 96 | 8.6 | 10.8 | 134 | 47 | 62 |
Mean size | 97 | 1.2 | 1.3 | 97 | 4.5 | 4.7 |
Mode size | 95 | 1.2 | 1.3 | 96 | 6.0 | 6.4 |
At the higher particle concentration, the NS300 NTA instrument shows reliable results for the analysis of 102 nm Nanospheres. The accuracy of the particle concentration is well within the expected 80–110% range, while the results for particle size are well within the 90–110% range. Except for the accuracy, the low flow-cell and the O-ring top-plate results are comparable, with those using the O-ring top-plate slightly better. At the lower particle concentration, it is clear that the low particle count influences repeatability and within-laboratory reproducibility. Except for the accuracy of particle concentration measured with the O-ring top-plate, the results for the accuracy are within the expected range.. In this case the configuration with the O-ring top-plate is overestimating the particle concentrations. It is also visible that the repeatability and within-laboratory reproducibility are higher with the O-ring top-plate than with the low flow-cell top-plate. While the differences between the low flow-cell and O-ring top-plate were minimal at the higher particle concentration, the results at the lower particle concentration seem to suggest that using the low flow-cell is beneficial at low particle concentrations. Therefore the low flow-cell configuration was used in most experiments.
In order to determine the working range for particle size, a calibration curve with particle sizes ranging from 30 nm to 350 nm was established. The calibration curves of both, the mean and mode particle size are presented in Fig. 8. This figure represents a five-fold analysis of each standard with on the x-axis the reference value and on the y-axis the determined particle size. Both calibration curves met the criterion of R2 > 0.99, indicating linearity. The accuracy and repeatability for the mean and mode particle size of each calibration point are also given in Table 2. If the accuracy of 90–110% is used as the criterium for the working range for particle size of the NS300, then the lower limit of particle size detection is 46 nm. The upper limit of particle size is > 350 nm, and with the technique of NTA, the particle size determination is limited to the nano-range, i.e. <1000 nm. Recently, Caputo et al. (2021)have determined this for various techniques and gave an upper limit of 800 nm for NTA.
Table 2
Accuracy and repeatability for mean and mode particle size of the calibration standards within a size range of 30–350 nm.
Particle size of particle standard [nm] | Mean particle size | Mode particle size |
Accuracy [%] | RSDr [%] | Accuracy [%] | RSDr [%] |
30 | 137 | 5.3 | 125 | 7.4 |
40 | 127 | 5.8 | 119 | 2.0 |
46 | 103 | 3.2 | 95 | 1.3 |
102 | 95 | 0.6 | 94 | 0.8 |
203 | 97 | 1.0 | 96 | 1.4 |
350 | 96 | 0.4 | 97 | 1.4 |
In order to determine the working range for particle concentration, a calibration curve with particle concentrations ranging from 5.0x106 to 2.0x109 particles/mL was prepared. The calibration curve is presented in Fig. 9, again as a five-fold analysis for each standard. The x-axis shows the reference value, and on the y-axis, particle concentration is determined. The calibration curve meets the criterion of R2 > 0.99, meaning the response is linear in the calibration range. The accuracy and repeatability of the particle concentration are given in Table 3. The working range criterion for particle concentration is an accuracy of 80–110%, and a maximum repeatability according to the Horwitz equation (RSDr max = 2C− 0.15 where C is expressed as an exponent of 10). From this it can be concluded that the working range for the particle concentration ranges from 1x107 to 2x109 part./mL.
Table 3
Accuracy and repeatability for particle concentrations in the range of 5x106 and 2.0x109 part./mL.
Particle concentration [part./mL] | Accuracy [%] | RSDr [%] | RSDcriteron.max [%] |
2x109 | 86 | 3.2 | 16 |
1x109 | 89 | 2.8 | 18 |
5x108 | 88 | 9.1 | 20 |
2x108 | 96 | 5.2 | 22 |
1x108 | 97 | 9.8 | 25 |
5x107 | 107 | 14 | 28 |
2x107 | 91 | 16 | 32 |
1x107 | 106 | 24 | 35 |
5X106 | 133 | 36 | 39 |
To test if NTA can distinguish between different particle sizes in the same sample, i.e. analyse polydisperse samples, three different mixtures with two particle sizes were analysed: a mixture of 46 and 102 nm particles, a mixture of 102 and 203 nm particles, and a mixture of 46 and 203 nm particles. In all mixtures, the particle number concentration of the two types of particles was equal and about 1x108 particles/mL. All mixtures were analysed five-fold, and the results are expressed as accuracy and relative standard deviation in Table 4.
Table 4
Accuracy and RSDr results for particle number concentration and mode particle size of three polydisperse suspensions. Fraction 1 describes the smaller-sized particles, while fraction 2 describes the larger-sized particles.
Mixture | Fraction 1 | Fraction 2 |
Concentration | Size | Concentration | Size |
Acc. [%] | RSDr [%] | Acc. [%] | RSDr [%] | Acc. [%] | RSDr [%] | Acc. [%] | RSDr [%] |
46 : 102 nm | 1.1 | 105 | 99 | 9.7 | 107 | 8.4 | 97 | 2.5 |
102 : 203 nm | 52 | 10 | 103 | 3.1 | 106 | 7.1 | 95 | 1.9 |
46 : 203 nm | < 1 | n.a. | n.a. | n.a. | 104 | 8.4 | 96 | 0.7 |
While the accuracy and particle size of the larger particles in a mixture are as expected, the particle number concentrations of the smaller particles in a mixture are largely underestimated. The particle size of the smaller particles in a mixture is as expected, at least if these particles are observed. The reason for these differences is that large particles scatter light much more strongly than small particles. The intensity of the scattered light relates to the particle size to the power of six. In the mixture of 46 and 203 nm particles, the 203 nm particle scatters light 4.46 (= 7400) more strongly than the 46 nm particle. Consequently, in the mixture the 46 nm particle, a particle size already on the limit of detection, is virtually invisible and not detected. In the mixture of 46 and 102 nm particles, we can just observe the 46 nm particles. In the case of the mixture of the 102 and 203 nm particles, both particles are well observed, but the concentration of the smaller-sized particle is still underestimated by a factor of 2. In real samples, this is a serious problem since particle numbers of small particles may be strongly underestimated when large particles are present.
Analysis of bottled water samples
Finally, the validated method was used to determine the presence of NP in bottled water. It was expected that the concentration of NP in bottled water was low. Therefore it was decided to use the maximum video capturing time of 180 s to count more particles and thus lower the linear range of the particle number concentration. In addition, it was expected that there might be large differences between the NP content in individual bottles of the same brand, and therefore ten individual bottles were analysed from each brand. In total, eight brands of mineral bottled water samples (A-H) were analysed. After opening each bottle, a subsample was collected and measured with the NTA instrument directly without any sample processing. The results of the particle number concentration and average particle size (mean and mode) are given in Table 5. Generally, the particle number concentration is between 1x106 and 4x106 particles/mL, which actually is on the verge of the lower limit of the linear concentration range taking into account the longer video capturing time. Two samples did show higher particle number concentrations of 1.5x107 (A) and 2.2x107 (E) particles/mL, and are clearly within the linear concentration range of the NS300. Considering particle sizes, the mass-based particle concentrations are in the range of 0.01 to 0.06 mg/L (ppm). The relative standard deviation (RSD in table 6) reflects not only the reproducibility (in the validation 37% at a particle number concentration of 2x107) but also the difference in NP content in the ten different bottles from one brand. This is particularly clear from brand A where we found an RSDRL of 165%, indicating large differences between individual bottles. This is not unusual and often encountered in the analysis of MP in bottled water, for instance Oßmann et al. (2018) reported 2689 ± 4371 MP (163%) in bottled water, while Schymanski et al. (2018) reported 118 ± 88 MP (75%) in mineral water of reusable bottles. Comparing the results of this study with other observations is difficult since all other studies were limited to MP with minimum particle sizes of 1 µm. However, a higher number of particles can be expected when smaller particle sizes are included because of the degradation of larger plastic particles to a higher number of smaller particles4. This trend is also clear from MP studies that were reported. While Kosuth et al. (2018) report 3.57 particles/L with a smaller particle size of 100 µm, Kankanige et al. (2020) report 140 particles/L with a smaller particle size of 6.5 µm, and Oßmann et al. (2018) report up to 8339 particles/L with a smaller particle size of 1 µm. Zuccarello et al. (2019) even report up to 1.1x108 particles/L using a particle size range of 1.28–4.2 µm. So, finding an even higher number of particles in this study may not be a surprising.
Table 5
Particle number concentrations and particle size (mean and mode) in bottled water samples analysed in ten-fold and with a video capturing time of 180 s.
Bottled water samples | Particle number concentration | Particle size |
Concentration [particles/mL] | RSDRL [%] | Mean [nm] | Mode [nm] |
A | 1.5x107 | 165 | 121 | 75 |
B | 4.0x106 | 45 | 113 | 81 |
C | 2.9x106 | 34 | 163 | 59 |
D | 1.0x106 | 50 | 114 | 85 |
E | 2.2x107 | 13 | 114 | 73 |
F | 3.1x106 | 62 | 119 | 76 |
G | 2.9x106 | 62 | 132 | 91 |
H | 1.3x106 | 72 | 142 | 84 |
The mean particle size in all bottled water samples is in the range of 110 to 170 nm, while the mode size is between 60 and 90 nm. This means that the most abundant particle size is < 100 nm in all bottled water samples. The higher value for the mean particle size compared to the mode particle size indicates the presence of larger particles. The NS300 can also produce a particle size distribution of an analysed sample. The particle size distribution of the samples A and E are shown in pictures 10 and 11, respectively.