A Comparative Investigation of Ultrafine Particle ... - ACS Publications

Ultrafine Particle Number and Mass. Emissions from a Fleet of On-Road. Diesel and CNG Buses. E. R. JAYARATNE, C. HE,. Z. D. RISTOVSKI, L. MORAWSKA,* ...
0 downloads 0 Views 516KB Size
Environ. Sci. Technol. 2008, 42, 6736–6742

A Comparative Investigation of Ultrafine Particle Number and Mass Emissions from a Fleet of On-Road Diesel and CNG Buses E. R. JAYARATNE, C. HE, Z. D. RISTOVSKI, L. MORAWSKA,* AND G. R. JOHNSON International Laboratory for Air Quality and Health, Queensland University of Technology, GPO Box 2434, Brisbane, QLD 4001, Australia

Received February 10, 2008. Accepted May 27, 2008.

Particle number, particle mass, and CO2 concentrations were measured on the curb of a busy urban busway used entirely by a mix of diesel and CNG operated buses. With the passage of each bus, the ratio of particle number concentration and particle mass concentration to CO2 concentration in the diluted exhaust plume were used as measures of the particle number and mass emission factors, respectively. With all buses accelerating past the monitoring point, the results showed that the median particle mass emission from CNG buses was less than 9% of that from diesel buses. However, the median particle number emission from CNG buses was 6 times higher than the diesel buses, and the particles from the CNG buses were mainly in the nanoparticle size range. Using a thermodenuder to remove the volatile material from the sampled emissions showed that the majority of particles from the CNG buses, but not from the diesel buses, were volatile. Approximately, 82% of the particles from the CNG buses and 38% from the diesel buses were removed by heating the emissions to 300 °C.

1. Introduction Vehicle emissions comprise the main source of ultrafine particles (particles smaller than 100 nm) in urban air (1, 2). These particles have received wide attention as they have been linked to a range of harmful health effects (3, 4). Particle emission standards worldwide are generally based on particulate matter (PM) mass. In recent years, emission control strategies based on engine design and after-treatment devices have proved to be largely successful in reducing the average particle mass emissions. Fuel type and composition are other parameters that have been investigated in this respect. For example, the sulfur content in commercial diesel fuel is being reduced with the aim of reducing particle emissions. However, these measures have had limited success in the reduction of Ultrafine particle numbers. Some studies have reported increased ultrafine particle numbers with the introduction of after treatment devices (5). In most countries, particle number emissions from motor vehicles are not controlled by current regulatory standards. In recent years, compressed natural gas (CNG) has emerged as a cleaner alternative vehicular fuel to diesel. CNG* Corresponding author phone: (617) 3138 2616; fax: (617) 3138 9079; e-mail: [email protected]. 6736

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 17, 2008

powered vehicles emit virtually no visible particulate matter (PM) or black soot at the tailpipe and consistently show significantly less PM emissions than diesel vehicles. Associated reductions are often over 90% (6–8). However, there is still a lot of uncertainty about the particle number emissions from vehicles using CNG in relation to diesel, with many studies showing apparently contradictory results. Although some studies have shown a marked reduction in total particle number emissions with CNG over diesel, others have indicated that the number of nanoparticles (particles smaller than 50 nm) may be higher from vehicles operating on CNG (9, 10). Holmen and Ayala (9) monitored the particle number emissions from two diesel buses and one CNG bus at idle and at a steady speed of 55 mph on a chassis dynamometer. They showed that, from the CNG bus, the particle number concentration in the accumulation mode, centered around 100 nm, was 10-100 times lower than from a diesel bus. However, they also noted that the emissions from the CNG bus often displayed large nuclei modes and consistently showed higher particle number concentrations in the size range below 20 nm. Therefore, although CNG emissions contain less PM than diesel, we do not have sufficient knowledge about the relative particle number concentrations of the two types of emissions that may prove to have some implications for health effects. In a recent publication, we have described a method for the rapid identification of high particle number emitting motor vehicles without taking them off the road (11). In this paper, we present the results of an application of this method to include both diesel and CNG buses and both particle number and particle mass emissions. Thus, the main aim of the present study was to use the rapid remote detection method described previously to monitor both particle number and mass emissions from a large number of individual on-road diesel and CNG buses in order to compare the particle mass and number emission factors of buses powered by the two types of fuel.

2. Materials and Methods 2.1. Experimental Methods. The use of the method in identifying high particle number emitting diesel buses on the road has been described in detail in a previous paper (11). The method is based on the measurement of particle number and CO2 concentrations in a small sample of the exhaust plume extracted from behind each passing bus. The CO2 concentration is used to determine the dilution ratio of the exhaust gas in the environment. Hence, the particle number concentration in the exhaust of a bus was estimated. This technique has been shown to be viable for the estimation of particle number emission factors from vehicles (12, 13). These, and other studies, have shown that, at a given speed and load, the CO2 emission factor of a vehicle is relatively constant (14). This is to be expected as the CO2 emission factor is directly proportional to the fuel consumption rate of the vehicle. Since, after emission, both the particle number (or mass) and CO2 concentrations are diluted by ambient air in the same manner, the ratio of particle number (or mass) concentration to CO2 concentration, Z, will not vary with dilution factor and will have the same value everywhere within the exhaust plume (13). The value of Z measured by sampling at any point within the exhaust plume may then be equated to the value at the tailpipe and will give a good measure of the particle number (or mass) emission factor of the vehicle. Hence, the measured Z ratios provide a means for comparing emissions from different vehicles under similar running conditions. 10.1021/es800394x CCC: $40.75

 2008 American Chemical Society

Published on Web 07/22/2008

TABLE 1. Composition of the Bus Fleet at the Time of the Study bus type

in-service date

number in service

diesel Pre-Euro diesel Euro I diesel Euro II CNG Scania

1982-1990 1991-1995 1997-2005 2000-2005

237 178 90 267

The measurements were carried out at a bus station located on a dedicated busway used by a city transport fleet consisting of around 750 buses of which roughly 500 operate on diesel and 250 on CNG. Since 2000, the older diesel buses in the fleet have been gradually replaced by CNG fuelled buses. During the time of the study, the entire diesel fleet was operating on ultralow sulfur (ULS, 50 ppm) diesel. The Euro standard diesel and CNG buses were routinely provided with a full service every 30 000 km, whereas the pre-Euro diesel buses were serviced more regularly: every 20 000 km. Table 1 shows the composition of the bus fleet according to bus type. Of these buses, approximately, 190 diesel and 120 CNG buses operated through the station each day, with each bus doing roughly 4-6 trips in each direction. Approximately 90 buses passed the station in each direction every hour during the peak hours, dropping to about 60 per hour at other times of the day. Monitoring was carried out over 7 days. Peak hours were avoided as most of the buses passed too close together to allow discrimination between individual exhaust plumes. 2.2. Technique. The instruments were placed on a trolley close to the curb of the departure end of the outbound platform. All buses passed in one direction close to the curb. The air at the edge of the busway was continuously sampled as the buses traveled past. The passage times and registration numbers of buses were available on video footage from the permanent overhead cameras on the platform. These were also recorded manually. The unique registration number on each bus enabled the identification of diesel and CNG powered buses. Particle number and mass, and CO2 concentrations were measured with fast-response instrumentation and recorded in real time. The concentrations contributed by each bus were estimated from the difference between the peak concentration observed soon after the passage of the bus and the background concentration just before the bus passed. The ratio of the step height-particle number to CO2 concentration contributed by the vehicle, Z, was calculated in units of millions of particles per mg of CO2. The corresponding step height Z for the particle mass to CO2 concentration contributed by the vehicle was calculated in units of mg of particle mass per mg of CO2 and was therefore expressed with no units. During the measurements, almost all of the buses stopped at the bus station platform. As the monitoring was carried out at the departure end of the platform, all buses were steadily accelerating past the sampling point. Thus, all buses passed under approximately the same load and had very similar CO2 emission factors and, therefore, the particle number and mass emission factors of a bus were assumed to be directly proportional to the respective values of the ratio Z. 2.3. Instrumentation. A schematic diagram of the measurement system is shown in Figure 1. The sample was extracted from a point 0.6 m above the ground at the curb through a short length of conductive tubing. The particle number concentration was measured with a TSI 3025A condensation particle counter (CPC). This is a fast-response continuous flow instrument that can detect airborne particles down to a size of 3 nm in total concentrations up to 105 cm-3.

The response time is less than 1 s for 95% response to concentration step change when sampling at a flow rate of 1.5 L min-1. As the particle number concentrations on the busway often exceeded the maximum detectable level of the instrument (105 cm-3), a HEPA filter was inserted in parallel with the inlet of the CPC to dilute the sample below the upper detectable limit. The flow rate through the filter was monitored with a TSI flowmeter and controlled with a pinch valve to provide a steady dilution factor of 10. Laboratory tests showed that for combustion aerosols in the size range 10-400 nm, at the flow rate used, particle losses in the valve and sampling tubing was negligible. The equivalent particle mass concentrations were monitored directly with a TSI Dustrak aerosol monitor, model 8520. This is a fast response instrument that uses optical scattering to determine equivalent mass concentration in real time. Although, it has an excellent signal-to-noise ratio and time resolution, the readings are not gravimetric and generally need to be corrected for aerosol type and density (15). However, Dustrak PM measurements show good linear correlation with aerosol mass measurements using filters and other standard instruments such as the tapered element oscillation monitor (TEOM) (15) and, in the present study, it served as a good indicator of equivalent mass emissions. An inlet impactor was used to limit the mass to PM2.5 (particles smaller than 2.5 µm). The CO2 concentrations were measured with a Sable CA-10A analyzer. This instrument uses a dual wavelength infrared beam and has a response time of less than 1 s at a sampling rate of 1 L min-1. Data from all three instruments were recorded at 1 s intervals in real time. Wind speed and direction were also monitored in real time using an automatic portable weather station. 2.4. Dilution Effects. The highly volatile gaseous products present in vehicle emissions often produce secondary particles by nucleation and, subsequently, these particles can evaporate rapidly when the gases are diluted by subsaturated environmental air (16, 17). Could this process occur during the enforced dilution of the sample with the air introduced through the HEPA filter? As shown in Figure 1, the air used for the enforced dilution was drawn from near the same point from which the plume sample was extracted and passed through the HEPA filter to remove all particles. However, the filter did not remove the gaseous components in the air. Hence, the volatile and semivolatile gas concentrations in the sample air and the dilution air were identical, that is they had the same vapor pressure. As such, the dilution process was unlikely to result in particle volatilization. Furthermore, the possibility of any increase in the number concentration of particles due to nucleation during the enforced dilution process was unlikely for the following reasons. The primary dilution process generally occurs soon after emission from the tail pipe. During this time, the rapid cooling of the exhaust gases give rise to the production of secondary particles by nucleation (16–18). Although the concentration of these volatile particles is very sensitive to the conditions imposed by the primary dilution process, they are not significantly affected by subsequent dilution (19–21). That is, once they are formed, further dilution has very little effect on their concentration. Kasper (19) explains in detail the influence of the dilution conditions on ultrafine particle formation. On page 318 of the paper, he states that “even high secondary dilution is usually insufficient to get rid of the nucleation mode”. Chase experiments have shown that exhaust emissions dilute very rapidly behind moving vehicles. The dilution ratios at distances of 1 and 2 m behind a diesel passenger vehicle have been measured to be about 50 and 100, respectively (22, 23). Ronkko et al. (21) monitored exhaust particle size distributions behind an on-road heavy-duty vehicle using a VOL. 42, NO. 17, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6737

FIGURE 1. Schematic diagram of the measurement system. chaser vehicle and showed that the nucleation mode had already formed after 0.7 s residence time in the atmosphere and no further changes were observed for longer residence times. At a vehicle speed of 40 kmph, nucleation modes observed at a distance of 5 m behind the vehicle did not change significantly when the chase distance was doubled. In the present study, time-resolved measurements showed that the plume took at least 0.75 s to reach the sampling tube inlet after emission. In this time, the exhaust plume had cooled to almost ambient temperature and it was extremely unlikely that any further particle formation occurred during the subsequent enforced dilution process. 2.5. Heating Experiments. Studies have shown that the majority of ultrafine particles in diesel emissions consist of volatile and semivolatile hydrocarbons and sulfur compounds (18, 24, 25). In order to test this hypothesis, in another set of experiments, the exhaust plume sample was heated to 300 °C by passing through a thermodenuder before entering the CPC (Figure 1). This temperature of 300 °C was chosen because it was above the volatilization temperatures of components such as sulfuric acid, ammonium sulfate, and most volatile organic compounds emitted by vehicles. At this temperature, only the carbonaceous core and metallic ash would remain (24, 26, 27). A two-way switch valve allowed the flow to be switched to and from the thermodenuder as, and when, required. Prior to use on the busway, the thermodenuder was tested in the laboratory and calibrated for normal tube and valve losses and for thermophoretic losses which occur when a hot aerosol flow cools to ambient temperature while sampling. Polydisperse aerosols were produced by atomizing a salt solution with compressed nitrogen gas. The concentration of the solution was varied according to the most probable size of aerosol required. The aerosols were size-classified by passing them through a TSI 3085 electrostatic classifier set to a known voltage in order to produce a monodisperse aerosol size distribution of the required size. Next, the monodisperse aerosols were passed through the thermodenuder into the CPC. The total number concentrations of particles before and after the thermodenuder were alternatively measured with the CPC. The thermophoretic losses of particles were calculated as a function of temperature, particle size, and air flow speed.

3. Results and Discussion All three instruments recorded distinct emission signals as each bus passed the sampling point. However, it was not always possible to obtain reliable data from every bus that 6738

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 17, 2008

passed. This was due to various reasons such as the wind blowing the exhaust plume away from the instruments and the difficulty in separating emissions from two or more buses passing too close to each other. These difficulties and the methods adapted to overcome them are described in detail in Jayaratne et al. (11). Despite these shortcomings, useful data were obtained from about 30% of the buses that passed. This amounted to 317 diesel and 264 CNG bus passes corresponding to 164 diesel and 98 CNG individual buses during the period of measurement. The Dustrak was used to obtain particle mass data from 123 diesel and 99 CNG bus passes, corresponding to 89 diesel and 60 CNG individual buses. The thermodenuder was used in line with the CPC to collect particle number data from 226 diesel and 130 CNG bus passes. The Dustrak was not used with the thermodenuder to obtain particle mass volatility. 3.1. Emission Signals. Figure 2 shows a set of typical emission signals of particle number concentration, particle matter mass concentration (PM2.5), and CO2 concentration recorded from three successive buses passing the measurement point. The first two buses were diesel buses and the last was a CNG bus. Note the relatively higher particle masses from the diesel buses and the higher particle numbers from the CNG bus. Also, shown are the bus identification numbers and the calculated particle number Z ratios for each of the three buses. The Z ratio values are shown in units of millions of particles per mg of CO2 and were found to be 1005 and 906 for the two diesel buses and 9184 for the CNG bus. This was a typically consistent observation; on the average, the particle number Z ratio’s for the CNG buses were significantly higher than that for the diesel buses. The median Z values for the 317 diesel bus passes and 264 CNG bus passes were (1150 ( 67) and (7465 ( 339) million mg-1, respectively, where the uncertainties indicate the standard errors. After averaging over multiple passes of the same buses, the corresponding median Z values for the 164 diesel buses and 98 CNG buses tested were (1265 ( 64) and (7584 ( 258) million mg-1, respectively. This shows that the particle number emissions from CNG buses were about 6 times higher than that from diesel buses. The difference between the mean emissions from the two types of buses was tested using a Student’s t test analysis and found to be statistically significant at a confidence level of 95% (p < 0.001). In general, although the particle number concentrations from the CNG buses were over 6 times higher than the diesel buses, the corresponding PM2.5 values were much lower. This is apparent in the example shown in Figure 2. The median particle mass (PM2.5) Z value in units of mg of particle mass per mg of CO2 for the 89 diesel

FIGURE 2. Typical emission signals obtained from three buses; the first two buses are diesel and the last bus is CNG. The triangles (2) indicate the times at which the buses passed. Figure (a) shows the equivalent particle mass PM2.5 (thin line, above) and particle number (PN) concentrations (thick line, below) and Figure (b) shows the CO2 concentration. The PN Z ratios for the three buses (left to right) were 1005, 906, and 9184 million mg-1, respectively. buses and 60 CNG buses that were tested with the Dustrak were (1.7 ( 0.1) × 10-3 and (1.5 ( 0.4) × 10-4 respectively, where the errors indicate the standard errors. This shows that the PM2.5 emissions from diesel buses were over 10 times higher than that from CNG buses. The difference between the mean emissions from the two types of buses was tested using a Student’s t test analysis and found to be statistically significant at a confidence level of 95% (p < 0.001). 3.2. Fleet Distribution. Figure 3(a) is a line plot of the particle number (PN) Z ratio for all the bus passes against the unique identification number assigned to each bus in the fleet. Each point corresponds to a bus pass. Buses numbered between 600 and 800 are CNG and the rest are diesel powered. The Z values for the CNG buses are clearly higher than the diesel values. The respective mean values are also shown in the figure. Figure 3(b) is the equivalent plot for the particle mass (PM2.5) Z ratio. Here, the CNG bus Z values are clearly less than the diesel values. Figure 4 shows the distributions of (a) particle number and (b) particle mass Z ratios for the diesel buses (dark shade) and the CNG buses (light shade). Where data were available for more than one pass of a given bus, average values of Z have been used. Both the PN and PM graphs show that the diesel and CNG bus Z value distributions were significantly different from each other. Both graphs are plotted as lognormal distributions. However, a detailed statistical analysis showed that each of these distributions more closely approximated a gamma distribution, with a long tail at the high emission end, not apparent in the graph due to the log axis. This is a common feature of vehicle emissions, where a small

percentage of vehicles give rise to a disproportionately large percentage of the emissions. It has been shown that emissions of carbon monoxide and hydrocarbons from a large mix of motor vehicles are well represented by gamma distributions (28). As described in our previous paper (11), the Z ratio does not change with dilution as the exhaust plume mixes with the ambient air and, therefore, the measured value of Z is the same as it is at the tailpipe. In this paper, although we have not attempted to use Z to estimate emission factors, we can assume that the two parameters are directly proportional. Thereby, we can rank the buses according to their respective Z ratios and use the cumulative values to estimate the percentage of buses that are responsible for a given proportion of the emissions. Thus, it was found that 10% of the diesel buses were responsible for approximately 28 and 25% of the total particle number and mass emissions, respectively, from the diesel bus fleet. The corresponding figures for the CNG fleet were 20 and 40%, respectively. These results show that a small number of buses are responsible for a disproportionately large amount of particle number and mass emissions. 3.3. Particle Size Distributions. It is well-known that the particle number distributions in diesel emissions exhibit a characteristic accumulation mode at around 80-100 nm with a smaller nucleation mode at around 10 nm (14). Most of the particle mass is contributed by the accumulation mode particles. However, CNG emissions have very little particle mass. Although, it was not possible to use an SMPS to measure particle size distributions owing to the rapidly changing VOL. 42, NO. 17, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6739

FIGURE 3. Line plots of the (a) particle number and (b) equivalent particle mass Z ratios against the individual bus identification numbers. Bus numbers between 600 and 800 are CNG buses and the rest are diesel buses. conditions and the short times during which the vehicle plumes passed over the instruments, the relatively large number concentration and very small mass concentration of particles in the CNG bus emissions strongly suggested that the majority of the particles were in the nucleation mode, that is, in the nanometer size range. The interpretation of the observation of higher particle number emissions from CNG buses over diesel buses requires some caution. It must be reiterated that the measurements were carried out at the departure end of the bus platform where all buses were in an accelerating state. Particle emissions from heavy-duty vehicles are highly variable and depend on many factors, especially engine load (7, 14, 29). It has been observed that the ultrafine particle emissions from a CNG bus were significantly affected by its driving condition such as acceleration and deceleration (30). The present results show that, during acceleration, a CNG bus emits more particle numbers than a diesel bus. However, it does not provide a comparison during other driving modes such as cruise and deceleration. 3.4. Heating Experiments. Passing the sample through the thermodenuder electrically heated to 300 °C before entering the CPC reduced the measured mean particle number concentrations significantly for both types of buses. From the results of the calibration experiments in the 6740

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 17, 2008

laboratory, it was estimated that, for combustion aerosols in the size range 10-400 nm, at the sample air flow rate present in the field study, particle losses in the valve was negligible. The corresponding thermophoretic losses of particle number in the thermodenuder were approximately 8%. All data were adjusted accordingly. For the diesel and CNG bus passes, the corresponding median Z values with the heater on were 775 and 1387, translating to particle number reductions of approximately 38 and 82%, respectively. Each of these two differences was statistically significant at a confidence level of 95% (p < 0.001). This observation clearly shows that the majority of particle numbers in the CNG emissions and about a third of the numbers in the diesel emissions were composed of volatile material. After volatilization, the particle number concentration from the CNG buses was still higher than that from the diesel buses. However, the ratio between the two reduced from 6.0 with no heating to just 1.8 after heating. It follows that, after volatilization, CNG emissions had only 80% more particle numbers than diesel. A large majority of buses that passed through the station were monitored both with and without the thermodenuder during two or more separate passes. Figure 5 shows a comparison of the respective particle number Z ratio values that were obtained with and without heating the emissions

FIGURE 4. Distributions of the (a) particle number and (b) particle mass Z ratios for the diesel and CNG bus fleets shown as separate log-normal plots for each parameter. The total number of each type of bus has been normalized and represented as percentages. from individual buses. Note that heating the emissions decreased the Z ratios for all CNG buses and for the large majority of diesel buses. The points above the line of equality indicate an increase of particle number with heating. This is unrealistic and may be explained as an artifact of the different running conditions of the buses during different passes, indicating that, for some of the diesel buses, the reduction in particle number concentration due to heating was less than the difference due to engine load changes. This was not observed with the CNG buses owing to the relatively large decrease in particle number concentration due to heating. It should also be noted that heating the emissions resulted in relatively narrow Z ratio distributions with both types of buses, with the values being more tightly grouped for CNG buses than for diesel buses (Figure 5). This suggests that, once the volatile material is stripped off, the concentration of the resulting solid particles does not vary significantly between buses. It is interesting to speculate why the heated emissions from the CNG buses showed any particles at all. Although it is generally accepted that the majority of the particles in CNG emissions are volatile, these results confirm that they include a substantial amount of Ultrafine particles in the form of solid ash and metals (31). 3.5. Summary of Results. Table 2 gives a concise summary of the Z ratio values determined in this study. In agreement with previous studies (6–8), it is confirmed that the particle mass emission from CNG buses is significantly

FIGURE 5. Particle number Z ratio values for individual buses with the emissions unheated (abscissa) and heated (ordinate). The upper and lower graphs show the diesel and CNG buses, respectively. The straight lines represent equality of the two plotted parameters.

TABLE 2. Summary of the Z Values Found in This Study diesel

CNG

median Z

Particle Number Z Ratio (millions mg-1) 1265 7584

median Z

Equivalent Particle Mass (PM2.5) Z Ratio 1.7 × 10-3 1.5 × 10-4

Particle Number Z Ratio (millions Thermodenuder at 300 °C median Z 775 1387 TD on/TD off 0.62 0.18

CNG/diesel

6.0

0.088 mg-1) 1.8

less than from diesel buses (8.8%). CNG buses emit large concentrations of particle numbers when in the accelerating mode. The results showed that the median particle number emission from CNG buses, when accelerating, was 6 times higher than that from the diesel buses and that the particles from the CNG buses were mainly in the nanoparticle size range. Using a thermodenuder to remove the volatile material from the sampled emissions showed that the majority of particles from the CNG buses, but not from the diesel buses, were volatile. Approximately, 82% of the particles from the CNG buses and 38% from the diesel buses were removed by heating the emissions to 300 °C. The study also showed that the method of detecting high-emitting vehicles is versatile and can be applied to many situations including different types of vehicles and fuel types. VOL. 42, NO. 17, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6741

Acknowledgments This project was supported by the Australian Research Council and Queensland Transport through Linkage Grant LP0454296. We would like to thank Jurgen Pasieczny and Ray Donato of Queensland Transport for their invaluable help and advice, and the staff of the Busway Operations Centre, Brisbane, for their assistance during the measurements.

(16)

(17)

Literature Cited (1) Shi, J. P.; Evans, D. E.; Khan, A. A.; Harrison, R. M. Sources and concentration of nanoparticles (