Particulate Measurements

Heinz Burtscher, W. Addy Majewski

This is a preview of the paper, limited to some initial content. Full access requires DieselNet subscription.
Please log in to view the complete version of this paper.

• Introduction
• Particle Measurements and Emission Regulations
• Types of Measuring Techniques

Introduction

Diesel particulate matter (PM) is not a well defined substance; rather, it is a complex mixture characterized by widely changing chemical composition and physical properties. The main components are elemental carbon (EC), a variety of organic species of different volatility, sulfur compounds and metal oxides. Depending on the temperature, semivolatile species may be in the gas phase or condensed on particles.

The properties of diesel PM depend on the fuel, engine technology, operating conditions, and exhaust aftertreatment. They also change with time, as the particulates undergo transformation once released in the atmosphere or while in the sampling train or in the measurement apparatus. Examples of such transformation include particle coagulation, evaporation and/or condensation of volatile compounds. PM transformation occurring in the sampling/measurement equipment may result in the formation of artifacts, i.e., compounds that are not present in either the undiluted engine exhaust gas or in the atmosphere around diesel vehicles.

Due to this complexity—contrary to the gaseous emissions—a single absolute measure of diesel PM emissions does not exist. Since the very definition of diesel particulate matter is in fact determined by the PM measuring technique that is used, all measures of diesel PM have a somewhat arbitrary character.

The Significance of Volatile Fraction. A number of issues and challenges in PM measurement are related to the semivolatile fraction present in the exhaust gas. This is especially true in modern, low emission engines.

In older engines without aftertreatment, the solid fraction—composed mainly of elemental carbon (EC)—was dominant. The semivolatile fraction condensed on the solid fraction when the temperature dropped. This condensation led to some increase in particle diameter and mass, but no new particles were formed.

Combustion optimization led to a decrease of the EC fraction and reduced the available surface for gaseous material to condense on. Under these conditions, a decrease in temperature may lead to a sufficient degree of supersaturation to induce nucleation and the formation of new particles. The result is a bimodal size distribution, consisting of a mode of solid particles with some of the semivolatile species condensed on them (accumulation mode, typically 80-100 nm in diameter) and a nucleation mode (around 20 nm). The nucleation mode becomes even more pronounced when PM aftertreatment is used [642][644]. A particulate filter efficiently removes the solid fraction (by more than 99%), while the volatile fraction passes through the filter in the gas phase. Very little surface is available for condensation when the exhaust cools downstream of the filter, which leads to nucleation. Oxidation catalysts or catalytically coated filters can enhance this trend, for example by oxidizing SO₂ to SO₃, which leads to the formation of sulfuric acid droplets. Oxidation of organic material may also increase the tendency to form particles [2354].

Figure 1 shows examples of size distributions with and without a particle filter. The filter reduces the solid fraction in the accumulation mode by about two orders of magnitude. However, a new mode of ultrafine particles—the nucleation mode—is created by volatile species. The fraction of volatile material which is measured as PM strongly depends on the way the measurement is done, in particular on temperature and dilution. While it was not an important issue for older engines with sufficient solid particle surface area available for the condensation of volatiles, it is of crucial importance for modern low emission engines, especially when measuring downstream of a particulate filter where most of the particle mass is volatile.

Condensation starts around the time when the dew point is reached, at a saturation ratio of 1 (for a more precise consideration, particle diameter and material also have to be taken into account). To initiate nucleation, a supersaturation is required. These processes—which occur in the exhaust system and, mainly, in the sampling lines—can be influenced by the way the exhaust is diluted and cooled [644][1052][838][1056]. If dilution is done with preheated air at a high enough dilution ratio, the saturation ratio can be kept low enough to minimize condensation and nucleation and ensure that only solid particles would be measured.

Figure 2 illustrates the range of temperatures and dilution ratios where supersaturation occurs, leading to nucleation/condensation [644]. The chart shows the saturation ratio for two HC species of different boiling temperatures (T_b) at two exhaust temperatures (T_e). The saturation ratio has the highest values for dilution ratios of about 5 to 50 for both hydrocarbons. Dilution ratios >50 are usually sufficient to prevent supersaturation.

The formation of nucleation particles is also influenced by the residence time before dilution or in the primary dilution stage [1052]. In some studies, the residence time in the primary dilution tunnel had a dramatic effect on particle numbers [192]. When changing the residence time from 40 ms to 6 s, the number concentration of nanoparticles increased by four orders of magnitude (see also Diesel Exhaust Particle Size). The highly non-linear nature of these processes is responsible for the sudden strong effects, depending on the saturation ratio and on the time required by the gas diffusion processes that are involved.

A similar effect, in this case not related to sampling, is observed for fuel additive constituents, whenever fuel additives are applied. If enough soot is available, this material is incorporated into soot particles; otherwise it nucleates homogeneously and forms new particles [1057].

Condensation of liquids on agglomerates may also change their structure by capillary forces, leading to more compact structures [1059]

Measurement of ‘Wet’ and ‘Dry’ Particles. Due to the nucleation and condensation phenomena during sampling, the presence of volatile material makes it more difficult to accurately and repeatedly measure such PM parameters as mass, number, or size distributions. Depending on how the volatiles are handled, particle measurement methods can be classified into two categories:

• Dry Measurement: One can choose to eliminate volatile material from the exhaust gas sample before they enter the measuring instrument. Nucleation can also be avoided by choosing a dilution ratio which is high enough to avoid the supersaturation required for nucleation. However, the ‘dried’ sample—as conveniently measured as it may be—does not properly represent the composition of roadside aerosols, which have been shown to include liquid particles [828].
• Wet Measurement: In another approach, in contrast with the ‘dry’ measurement, one could also cool the undiluted exhaust prior to the measurement to enforce nucleation/condensation of all material that can condense. This ‘wet’ measurement would present a worst case measurement, including all material which potentially can form particles.

Most PM measurements provide results somewhere in between these two extremes. The conventional gravimetric analysis, done at a moderate dilution and temperature, includes a significant part of the semivolatile material and can be classified as a relatively ‘wet’ measurement. The European particle number measurement, on the other hand, represents a ‘dry’ measurement—only solid particles are measured and nucleation is prevented by adequate pretreatment.

###