characterization of small particles by micro x-ray ... › ~gcas › esp2 › papers ›...

10
Characterization of small particles by micro X-ray fluorescence Thomasin C. Miller a, * , Helen Langley DeWitt b , George J. Havrilla b a X-ray Optical Systems, Inc., East Greenbush, NY 12180, USA b Chemistry Division, Analytical Chemistry Science, Los Alamos National Laboratory, Los Alamos, NM 87545, USA Received 2 February 2005; accepted 9 September 2005 Available online 25 October 2005 Abstract Micro X-ray fluorescence was used to study both homogeneous and heterogeneous particle systems. Specifically, homogeneous glass microspheres and heterogeneous soil particle samples were prepared by both bulk and single particle sample preparation methods for evaluation by micro X-ray fluorescence. Single particle sample preparation methods allow for single particles from a collected sample to be isolated and individually presented to the micro X-ray fluorescence instrument for analysis. Various particle dispersion methods, including immobilization onto Tacky Doti slides, mounting onto double-sided sticky tape affixed to polypropylene film, or attachment to polypropylene film using 3M Artist’s Adhesive, were used to separate the sample particles for single particle analysis. These methods were then compared and evaluated for their ability to disperse the particles into an array of single separated particles for optimal micro X-ray fluorescence characterization with minimal background contribution from the particle mounting surface. Bulk methods of particle sample preparation, which included pellet preparation and aerosol impaction, used a large quantity of collected single particles to make a single homogeneous specimen for presentation to the instrument for analysis. It was found that single particle elemental analysis by micro X-ray fluorescence can be performed if the particles are well separated (minimum separation distance = excitation source beam diameter) down to a particle mass of ¨ 0.04 ng and a mean particle diameter of ¨ 0.06 Am. Homogeneous particulates can be adequately characterized by micro X-ray fluorescence using either bulk or single particle analysis methods, with no loss of analytical information. Heterogeneous samples are much harder to characterize, and both single particle as well as bulk analyses must be performed on the sample to insure full elemental characterization by micro X-ray fluorescence. D 2005 Elsevier B.V. All rights reserved. Keywords: Particle analysis; Micro X-ray fluorescence (MXRF); Cascade impactor; Aerosol; Soil; Glass microsphere 1. Introduction The analysis of particulates is important for many environ- mental, biological, and forensics applications. Precise chemical and structural analysis of collected particles might reveal conditions governing their formation or manufacture, leading to enhanced characterization and attribution of a given specimen. For example, particulate samples obtained from the environment, such as aerosols, fly ashes, soils, and sediments, can help establish sources of pollution and conditions of formation. The biological impact of exposure to various particulates, such as pollutants, irritants or allergens, can be determined and evaluated. In forensics, the analysis of particles found on a suspect or at a crime scene, such as fibers, glass, or gunshot residue, can potentially place the suspect at the scene of a crime. X-ray analytical techniques have been an important means of gathering elemental information about particulate samples. For example proton-induced X-ray emission (PIXE) has been employed for the analysis bulk soil samples (e.g., [1,2]) as well as bulk and individual aerosol particles (e.g., [3–7]). Total reflection X-ray fluorescence (TXRF) analysis has been used for the analysis of soil particulate extracts (e.g., [8,9]) as well as the analysis of aerosol and emission particulates (e.g., [7,10 – 18]). Synchrotron radiation X-ray fluorescence (SRXRF) has been applied to the study of sediment particles [19], municipal solid waste fly ashes [17], and radioactive particles [20]. X-ray fluorescence (XRF) has been used in the study of the bulk elemental properties of aerosols, soils and other particulates (e.g., [9,21 – 29]). This paper demonstrates that micro X-ray fluorescence (MXRF) is a method that shows promise for both bulk and 0584-8547/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.sab.2005.09.003 * Corresponding author. Tel.: +1 518 880 1500x405; fax: +1 518 880 1510. E-mail address: [email protected] (T.C. Miller). Spectrochimica Acta Part B 60 (2005) 1458 – 1467 www.elsevier.com/locate/sab

Upload: others

Post on 07-Jul-2020

5 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Characterization of small particles by micro X-ray ... › ~gcas › esp2 › papers › miller05.pdf · For example proton-induced X-ray emission (PIXE) has been employed for the

w.elsevier.com/locate/sab

Spectrochimica Acta Part B 6

Characterization of small particles by micro X-ray fluorescence

Thomasin C. Miller a,*, Helen Langley DeWitt b, George J. Havrilla b

a X-ray Optical Systems, Inc., East Greenbush, NY 12180, USAb Chemistry Division, Analytical Chemistry Science, Los Alamos National Laboratory, Los Alamos, NM 87545, USA

Received 2 February 2005; accepted 9 September 2005

Available online 25 October 2005

Abstract

Micro X-ray fluorescence was used to study both homogeneous and heterogeneous particle systems. Specifically, homogeneous glass

microspheres and heterogeneous soil particle samples were prepared by both bulk and single particle sample preparation methods for evaluation

by micro X-ray fluorescence. Single particle sample preparation methods allow for single particles from a collected sample to be isolated and

individually presented to the micro X-ray fluorescence instrument for analysis. Various particle dispersion methods, including immobilization onto

Tacky Doti slides, mounting onto double-sided sticky tape affixed to polypropylene film, or attachment to polypropylene film using 3M Artist’s

Adhesive, were used to separate the sample particles for single particle analysis. These methods were then compared and evaluated for their ability

to disperse the particles into an array of single separated particles for optimal micro X-ray fluorescence characterization with minimal background

contribution from the particle mounting surface. Bulk methods of particle sample preparation, which included pellet preparation and aerosol

impaction, used a large quantity of collected single particles to make a single homogeneous specimen for presentation to the instrument for

analysis. It was found that single particle elemental analysis by micro X-ray fluorescence can be performed if the particles are well separated

(minimum separation distance=excitation source beam diameter) down to a particle mass of ¨0.04 ng and a mean particle diameter of ¨0.06 Am.

Homogeneous particulates can be adequately characterized by micro X-ray fluorescence using either bulk or single particle analysis methods, with

no loss of analytical information. Heterogeneous samples are much harder to characterize, and both single particle as well as bulk analyses must be

performed on the sample to insure full elemental characterization by micro X-ray fluorescence.

D 2005 Elsevier B.V. All rights reserved.

Keywords: Particle analysis; Micro X-ray fluorescence (MXRF); Cascade impactor; Aerosol; Soil; Glass microsphere

1. Introduction

The analysis of particulates is important for many environ-

mental, biological, and forensics applications. Precise chemical

and structural analysis of collected particles might reveal

conditions governing their formation or manufacture, leading

to enhanced characterization and attribution of a given

specimen. For example, particulate samples obtained from

the environment, such as aerosols, fly ashes, soils, and

sediments, can help establish sources of pollution and

conditions of formation. The biological impact of exposure to

various particulates, such as pollutants, irritants or allergens,

can be determined and evaluated. In forensics, the analysis of

particles found on a suspect or at a crime scene, such as fibers,

0584-8547/$ - see front matter D 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.sab.2005.09.003

* Corresponding author. Tel.: +1 518 880 1500x405; fax: +1 518 880 1510.

E-mail address: [email protected] (T.C. Miller).

glass, or gunshot residue, can potentially place the suspect at

the scene of a crime.

X-ray analytical techniques have been an important means

of gathering elemental information about particulate samples.

For example proton-induced X-ray emission (PIXE) has been

employed for the analysis bulk soil samples (e.g., [1,2]) as well

as bulk and individual aerosol particles (e.g., [3–7]). Total

reflection X-ray fluorescence (TXRF) analysis has been used

for the analysis of soil particulate extracts (e.g., [8,9]) as well as

the analysis of aerosol and emission particulates (e.g., [7,10–

18]). Synchrotron radiation X-ray fluorescence (SRXRF) has

been applied to the study of sediment particles [19], municipal

solid waste fly ashes [17], and radioactive particles [20]. X-ray

fluorescence (XRF) has been used in the study of the bulk

elemental properties of aerosols, soils and other particulates

(e.g., [9,21–29]).

This paper demonstrates that micro X-ray fluorescence

(MXRF) is a method that shows promise for both bulk and

0 (2005) 1458 – 1467

ww

Page 2: Characterization of small particles by micro X-ray ... › ~gcas › esp2 › papers › miller05.pdf · For example proton-induced X-ray emission (PIXE) has been employed for the

Table 1

Element emission lines and energies monitored in this study

Element Emission line Emission line energy (keV)

Mg K-L2,3 (Ka) 1.254

Al K-L2,3 (Ka) 1.487

Si K-L2,3 (Ka) 1.740

S K-L2,3 (Ka) 2.308

Cl K-L2,3 (Ka) 2.622

K K-L2,3 (Ka) 3.313

Ca K-L2,3 (Ka) 3.691

Ti K-L2,3 (Ka) 4.510

V K-L2,3 (Ka) 4.952

Mn K-L2,3 (Ka) 5.898

Fe K-L2,3 (Ka) 6.403

Cu K-L2,3 (Ka) 8.047

As K-L2,3 (Ka) 10.543

T.C. Miller et al. / Spectrochimica Acta Part B 60 (2005) 1458–1467 1459

single particle analysis. MXRF employs polycapillary optics

that are used in conjunction with the X-ray source to focus

incident X-rays. This results in smaller beam diameters (i.e.,

10–50 Am) and higher X-ray fluxes (¨107 photons/s) at the

sample surface, increasing detection sensitivity and enhancing

performance over conventional X-ray fluorescence instru-

ments. The smaller X-ray focal spot allows for much smaller

sample features to be observed and characterized than larger

spot techniques such as XRF, TXRF, and PIXE. Unlike

techniques such as SRXRF and PIXE which require specialized

instrumentation rarely accessible to the average user, such as a

synchrotron source, MXRF is readily available as a simple

bench-top instrument. It is a nondestructive technique, leaving

the sample intact for other analyses to be performed on it. It has

advantages over more conventional single particle analysis

techniques, such as scanning electron microscopy (SEM), in

that samples can be analyzed in both air and under vacuum,

and a more complete analysis of the volume of a sample

particle can be obtained due to the penetrating nature of X-rays

(tens to hundreds of micrometers). Furthermore, in addition to

single point analyses, elemental imaging can be performed by

MXRF with the ability to scan over large areas to map the

distribution and elemental characteristics of particles deposited

on or impregnated into a given substrate.

In this study, MXRF was used to study both homogeneous

and heterogeneous particle systems through both single point

and imaging analysis. Single particle sample preparation

methods allow for single particles from a collected sample to

be isolated and individually presented to the MXRF instrument

for analysis. Various particle dispersion methods, including

immobilization onto Tacky Doti slides, mounting onto

double-sided sticky tape affixed to polypropylene film, or

attachment to polypropylene film using 3M Artist’s Adhesive,

were used to separate the sample particles for single particle

analysis. These methods were then compared and evaluated for

their ability to disperse the particles into an array of single

separated particles for optimal MXRF characterization with

minimal background contribution from the particle mounting

surface. Bulk methods of particle sample preparation, which

included pellet preparation and aerosol impaction, used a large

quantity of collected single particles to make a single

homogeneous specimen for presentation to the instrument for

analysis. Bulk analyses were compared to results obtained from

single particle analysis for both homogeneous and heteroge-

neous samples. Size and mass method detection limits were

also calculated for observation of a single particle with the

current MXRF instrumentation available at Los Alamos

National Laboratory (LANL).

2. Experimental

2.1. Reagents and materials

All particle samples were used as received, without further

purification. Stearic acid was obtained from Sigma Chemical

Corporation (St. Louis, MO). 5, 15, and 20 Am borosilicate

uniform glass microspheres as well as 30 Am soda lime glass

particles were obtained from SPI Supplies (Westchester, PA).

China Loess CRM CJ-2 simulated Asian mineral dust was

obtained from the National Research Center for Environmental

Analysis and Measurement (China). K-411 Glass Microspheres

(1 Am–40 is Am in diameter) SRM 2066 was obtained from the

National Institute of Standards and Technology (NIST)

(Gathersburg, MD). The following particle mounting supplies

were used as received: Tacky Doti slides and related materials

(SPI Supplies, West Chester, PA), polypropylene film (4 Amthickness) (Spex CertiPrep, Metuchen, NJ), double-sided clear

sticky tape, and 3M Artist’s Adhesive (3M, Minneapolis, MN).

Parafilm and Fisherbrand Superfriendly Air-It were used for

the air dispersion method of particle sample preparation and

were both obtained from Fisher Scientific (Houston, TX).

2.2. Instrumentation and sample preparation

MXRF analysis of individual particles and bulk pellet

samples was performed in vacuum using an EDAX Eagle II

MXRF system equipped with a Rh target excitation source

and a SiLi detector (EDAX, Mahwah, NJ). The X-ray source

was equipped with a polycapillary X-ray focusing optic

having a 60 Am nominal X-ray spot size at Cu Ka, 8.04 keV

(X-ray Optical Systems, East Greenbush, NY). The optic on

the instrument could also be defocused to an X-ray focal spot

size of ¨360 Am. Table 1 lists the element emission lines and

their energies that were monitored in the experiments outlined

in this study. Particle characterization was performed with a

nominal X-ray spot size of 60 Am while bulk analysis of

particles pressed into pellets was performed with an X-ray

spot size of 360 Am.

Microscope observation was performed using a Leica Micro

Star IV stereomicroscope (Leica Microsystems, Bannockburn,

IL). The microscope was equipped with a 21 mm Reticle X-

Scale 5 mm Crossed Micrometer Scale-5 mm/100 Div.

(Edmund Optics, Barrington, NJ) calibrated with a 1 mm/100

divisions stage micrometer to determine or verify particle sizes.

A Carver 4350 Manual Pellet Press (Chemplex Industries,

Inc., Palm City, FL) with a 4-in. platen diameter and a 13 mm

supplied die was used to apply force to samples mixed with

stearic acid to create bulk sample pellets. Bulk particle samples

Page 3: Characterization of small particles by micro X-ray ... › ~gcas › esp2 › papers › miller05.pdf · For example proton-induced X-ray emission (PIXE) has been employed for the

T.C. Miller et al. / Spectrochimica Acta Part B 60 (2005) 1458–14671460

were prepared by forming pellets from a homogenized mixture

of ¨40 mg of mixed particles and equivalent amounts of the

binder stearic acid. A Wig-L-Bug was used for mixing

homogeneous particle/binder samples prior to pellet formation

(Reflex Analytical Corp., Ridgewood, NJ). Approximately

14,000 lbs of force was applied to the pellet during pressing.

A rotating micro-orifice uniform deposit impactor

(MOUDI)i, model number 110, (MSP Corporation, Minnea-

polis, MN) was used to separate and collect different particle

sizes of CRM-CJ-2, a simulated Asian mineral dust sample for

MXRF imaging and analysis. Polypropylene film (4 Amthickness) was used as the impaction substrate.

Dispersed particle samples were prepared using a variety

of methods to both reduce the background contribution of the

mounting surface and to disperse the particles into an array of

individual, separated particles. For effective single particle

detection, the particles must have a minimum separation

distance of the diameter of the X-ray focal spot on the sample

surface (¨60 Am). The different particle mounting methods

used were immobilization onto Tacky Doti slides, mounting

onto double-sided sticky tape affixed to polypropylene film,

or attachment to polypropylene film using 3M Artist’s

Adhesive.

The Tacky Doti array is a unique adhesion system, in

which a 25 Am thick layer of transparent polymer containing a

controlled pattern of ‘‘tacky’’ dots of precisely determined size

and location, that allows the user to produce a highly regular

array of separated particles. For this study, the tacky polymer is

affixed to a normal glass microscope slide by the manufacturer

(SPI Supplies). Immobilization of particles onto the Tacky

Doti slide mount is described in detail in a report generated

by SPI Supplies [30]. Briefly, using the Tacky Doti slide

holder, a few milligrams of particle sample were placed into the

slide holder and the ensemble was agitated from side to side to

distribute the particles across the adhesive dots.

Particles were mounted onto double-sided sticky tape using

two different methods; direct deposition and air dispersion. For

both processes, a piece of double-sided tape was first affixed to

the surface of polypropylene film which had been previously

affixed to a 35 mm slide mount. Direct deposition was

achieved by using a small spatula to transport a few milligrams

of particle sample onto the tape surface. The sample was then

shaken upside down to remove any unfixed particles. For the

air dispersion method, the tape/film mount was placed into the

bottom of a 500 mL beaker and a few milligrams of particles

were placed beside it. The beaker top was sealed with Parafilm

and then punctured with a small hole. The nozzle of a

Superfriendly Air-It air spray can was fitted into the small

hole and used to disperse the particles in the beaker. The

separated particles were then allowed to settle onto the tape-

covered slide at the bottom of the beaker [31].

3M Artist’s Adhesive was also used to prepare a sample of

dispersed particles. 3M adhesive was sprayed onto polypro-

pylene film attached to a 35 mm slide mount. The particles

were then affixed to the adhesive prepared slide using the same

air dispersion method described above for the double-sided

sticky tape.

3. Results and discussion

3.1. Effect of sample mounting on single particle dispersion

and MXRF detection

Fig. 1 shows white light images of 20 Am glass microspheres

prepared with each of the different mounting methods described

above. In terms of particle separation, the Tacky Dot, aerosol

method with tape, and aerosol method with 3M adhesive are all

adequate for particle dispersion. The direct deposition method,

pictured in Fig. 1b, does not adequately separate the glass

spheres from one another. The particles are clumped very close

together on the tape surface. The Tacky Dot array allows for the

particles to be separated; however, much depends on the particle

size. According to the SPI sample preparation report [30], the

‘‘tacky’’ dot size needed to hold a smooth, roughly spherical

particle should have a diameter at least 25% that of the longest

dimensions of the particles present for highest array formation

efficiency. The diameter of the glass microspheres is 20 Am.

According to the specifications given above, a 5 Am dot array is

optimal for the 20 Am spheres. At this time, the smallest Tacky

dot size available is 15 Am. In Fig. 1a, one can see that even with

the non-optimal 15 Am dot size array, the particles can be

separated from one another. However, if one looks closely, one

can see that the particles are immobilized on the tacky surface,

but not necessarily in the dot wells. As a comparison, the inlaid

picture in Fig. 1a shows 100 Am glass microspheres immobi-

lized onto a Tacky dot array; they are nicely immobilized as a

regular array on the tacky surface. Since the smallest Tacky dot

size currently available is 15 Am, this method is only efficient

for particles of �60 Am in diameter. Using the Tacky Dot array

even becomes more problematic with a particle sample

containing many different particle sizes, as is often the case

with environmental samples of interest, such as soil particulates.

The remaining two methods of sample preparation are those

using the air dispersion method combined with either double-

sided sticky tape or 3M adhesive on polypropylene film. Both

show that the particles can be well separated from one another.

The largest separation is achieved with the 3M adhesive,

although the difference is probably due to the amount of

sample introduced into the beaker prior to dispersion as well as

how the particles settled out on the substrate surface.

Another factor that will affect particle analysis is the

background produced by the mounting substrate. Fig. 2 shows

MXRF spectra of each of the different substrate materials; the

Tacky Dot slide, double-sided sticky tape on polypropylene,

and 3M adhesive on polypropylene. The spectra were normal-

ized by dividing by the highest spectral peak intensity in each of

the respective spectra. The Tacky Dot slide shows the most

complex spectrum containing large elemental peaks from Si and

Ca. Smaller peaks of S, Cl, K, and As are also present.

According to SPI Supplies [30], the S and Cl are from the

polymer Tacky surface and the other elements are present in the

glass slide on which the polymer is affixed. Due to the presence

of so many elements, this background may not be suitable for

MXRF analysis of particles, especially those containing these

background elements. The double-sided sticky tape on poly-

Page 4: Characterization of small particles by micro X-ray ... › ~gcas › esp2 › papers › miller05.pdf · For example proton-induced X-ray emission (PIXE) has been employed for the

Fig. 1. 20 Am glass microspheres mounted on a) a Tacky Doti slide (picture inlay shows 100 Am glass microspheres mounted on a on a Tacky Doti slide for

comparison), b) double-sided sticky tape on polypropylene film using direct depositions, c) double-sided sticky tape on polypropylene film using the dispersion

method, and d) 3M Artist’s adhesive on polypropylene film using the dispersion method.

T.C. Miller et al. / Spectrochimica Acta Part B 60 (2005) 1458–1467 1461

propylene shows significant S and K peaks as well as a higher

Bremsstrahlung background. This is unsuitable for particles

containing either S or K, or traces of any elements found in the

high background region ranging from ¨4 to 12 keV. The 3M

adhesive on polypropylene gives the best background trace. It

only contains minor Si and Ca contaminants.

Due to its low elemental and Bremsstrahlung background as

well as its good particle dispersion characteristics, the 3M

adhesive on polypropylene used in conjunction with the air

dispersion method was used as the sample mounting method

for the remaining experiments.

0

0.2

0.4

0.6

0.8

1

0 5

Ener

No

rmal

ized

Inte

nsi

ty (

cou

nts

)

Ca

K

Si

S

Rh

Cl

Fig. 2. Spectra of double sided sticky tape on polypropylene film, 3M Artist’s adhes

kV, 1000 AA for sticky tape and 3M adhesive, 40 kV, 330 AA for Tacky Doti sli

3.2. Particle size detection limit

To determine the smallest relative particle size that can be

detected with our MXRF instrument, a series of isolated

uniform borosilicate glass microspheres of different diameters

smaller than the X-ray beam size (5 Am, 15 Am, and 20 Am;

SPI Supplies, West Chester, PA) were imaged with MXRF. The

Si intensity for each particle size is listed in Table 2. From this

calibration of particle diameter versus elemental intensity, the

particle size limit of detection for MXRF is calculated to be

¨0.06 Am. The intensity is directly proportional to particle

10 15 20

gy (KeV)

Sticky Tape

3M Adhesive

Tacky Dot

As

Scatter

ive on polypropylene film, and a Tacky Doti slide. X-ray tube conditions: 40

de, Spectral dwell time=100 live seconds.

Page 5: Characterization of small particles by micro X-ray ... › ~gcas › esp2 › papers › miller05.pdf · For example proton-induced X-ray emission (PIXE) has been employed for the

Table 2

Particle diameter versus Si intensity for uniform glass microspheres (n =3)

Particle diameter (Am) Si intensity (cps) Particle mass (ng)

4.9T0.5 43T3 0.157T0.047

14.5T1.0 111T11 4.45T0.83

19.9T1.4 185T5 10.5T2.2Calculated limit

of detection

Particle diameter

(Am) 0.06

Particle mass

(ng) 0.04

T.C. Miller et al. / Spectrochimica Acta Part B 60 (2005) 1458–14671462

diameter. Similarly, the mass for each particle was calculated

assuming a spherical shape and known particle density of 2.50

g/cm3. By plotting a calibration curve of Si intensity versus

particle mass, the particle mass limit of detection was

calculated to be ¨0.04 ng. Conversely, the largest diameter

Si particle that can still be analyzed with minimal absorption-

enhancement effects (i.e., primary X-rays penetrate and analyte

line X-rays emerge substantially unabsorbed) can be estimated

by calculating the critical depth of penetration for Si. The

critical depth was calculated to be ¨50 Am by Eq. (1) [32]:

d ¼ 46; 000

lqsinb ð1Þ

where d is the critical depth in Am, l is the mass attenuation

coefficient in cm2/g (lSi=3.192�102 cm2/g [33]), q is the

density of the sample in g/cm3 (2.50 g/cm3), and b is the

takeoff angle of the spectrometer (for our instrument, b =60-).Therefore the largest Si particle can be ¨50 Am in diameter

and would correspond to a mass of ¨90 ng (assuming a

density of 2.50 g/cm3).

3.3. MXRF analysis of homogeneous particulates

MXRF point spectra and imaging analysis can be used to

detect and differentiate particulates of homogeneous composi-

tion, such as glass fibers or particles, by the elemental

0

0.2

0.4

0.6

0.8

1

1.2

0 2 4

Energ

No

rmal

ized

Inte

nsi

ty (

cou

nts

) CaSi

Al

Mg

Ti

Rh scatter

Fig. 3. Point spectra of a SRM 2040 K-411 glass microsphere, a 30 Am soda-lime gl

1000 AA, Spectral dwell time=100 live seconds.

‘‘fingerprint’’ of the material of interest by both single point

spectra as well as elemental imaging. For example, Fig. 3

shows the spectra of three different kinds of glass particles; K-

411 glass spheres, 30 Am soda lime glass, and 20 Amborosilicate glass. The spectra have all been normalized to

their largest intensity for ease of comparison. Each has a

specific MXRF elemental signature. The K-411 glass is

primarily made up of elemental Mg, Si, Ca, and Fe and has

relative intensities in the order of Fe>Ca>Si>Mg. The soda

lime glass spheres primarily contain Si, Ca, and Fe in the

intensity order Si>Ca>Fe (the opposite trend compared to the

K-411 particles). Borosilicate glass consists of Si and Ca with

traces of Al and Ti. The Ca relative intensity is much greater

than that of Si.

Fig. 4a shows MXRF elemental images of a sample

prepared from a mixture of the three different types of glass

particles. By knowing their different elemental signatures, the

different glass particles can easily be identified in the mixture

sample. By examining the Mg, Si, Ca, and Fe elemental maps,

particle 1 shows very high Si, Ca, and Fe intensities and also

contains Mg. This elemental signature corresponds to the K-

411 glass. Particle 2 displays high Si and Ca intensity with

small Fe and minimal Mg, corresponding to the fingerprint of

soda-lime glass. Particles 3, 4, and 5 all contain Si and Ca, but

lack any Mg or Fe intensity. They are all borosilicate glass

particles. Fig. 4b is a cartoon identifying each of the different

types of glass spheres based on their elemental image

signatures.

One advantage of analyzing particles of homogeneous

composition is that each particle should be elementally the

same, both qualitatively and quantitatively, as other particles of

the same types as well as the bulk material. For example, Fig. 5

shows a spectrum of an individual 20 Am borosilicate glass

microsphere as well as a spectrum of a pellet made from 40 mg

of bulk borosilicate glass microspheres. Notice that the spectra

6 8 10

y (KeV)

K-411

Soda Lime

Borosilicate

Fe

ass sphere, and a 20 Am borosilicate glass sphere. X-ray tube conditions: 40 kV,

Page 6: Characterization of small particles by micro X-ray ... › ~gcas › esp2 › papers › miller05.pdf · For example proton-induced X-ray emission (PIXE) has been employed for the

Fig. 4. a) MXRFMg, Si, Ca, and Fe elemental images of a mixture of K-411, soda lime, and borosilicate glass microspheres mounted on polypropylene film with 3M

adhesive using the air dispersion method. X-ray tube conditions: 40 kV, 1000 AA; Imaging conditions: 64�50 pixel matrix, 200 ms dwell time per pixel. b) A

cartoon identifying each of the different types of glass spheres based on their elemental image signatures. Particle 1 is K-411 glass, particle 2 is soda-lime, and

particles 3–5 are all borosilicate glass.

T.C. Miller et al. / Spectrochimica Acta Part B 60 (2005) 1458–1467 1463

show the same elements in the same approximate ratios. No

information is lost by examining one particle or a bulk quantity

of particles.

3.4. MXRF analysis of heterogeneous particulates

The analysis of heterogeneous particulate systems, such as

soils and environmental air samples, is much more complex

than for homogeneous systems. Specifically, the samples are

made up of more than one type of particle and cannot be

characterized by one particle alone. For example Fig. 6 shows

MXRF elemental images of a simulated Asian mineral dust

sample, CRM-CJ-2. Notice that not all of the particles in the

sample have the same elemental composition. For example,

particle 1 and particle 2 contain Al, Si, K, Ca, and Fe.

0

500

1000

1500

2000

2500

3000

3500

4000

4500

0 2 4

Ene

Inte

nsi

ty (

cou

nts

)

Si

Ca

Rh scatter

Al

Fig. 5. Point spectra of a bulk pellet and a single particle of 20 Am borosilicate glass

kV, 220 AA for the pellet, Spectral dwell time=100 live seconds.

Additionally, particle 2 contains Cu. Particle 3 only contains

Al, K, Ti, and Fe. One individual particle obviously is not

representative of the bulk mineral dust sample.

Creating a homogeneous bulk pellet out of the heteroge-

neous particles can allow for a more representative sample to

be prepared in order to gain additional information about a

given sample. For example, Fig. 7 shows the spectrum of a

single particle of CRM-CJ-2, as well as a pellet made out of

¨40 mg of the simulated Asian mineral dust. The pellet shows

the presence of Al, Si, K, Ca, Ti, V, Mn, and Fe. The intensities

of these elements in the single particle spectrum are much less

intense, and the trace quantities of V and Mn are almost non-

existent. Interestingly, the element Cu is present in the single

particle, but is absent in the bulk pellet sample. Cu may only be

present in a small amount of particles in the sample, and its

6 8 10

rgy (KeV)

pellet

particle

Fe

microspheres. X-ray tube conditions: 40 kV, 1000 AA for the single particle, 35

Page 7: Characterization of small particles by micro X-ray ... › ~gcas › esp2 › papers › miller05.pdf · For example proton-induced X-ray emission (PIXE) has been employed for the

Fig. 6. MXRF elemental images of CRM CJ-2 mounted on polypropylene film using 3M adhesive and the air dispersion method. The elemental signatures of

particles 1, 2 and 3 in the sample are all different. X-ray tube conditions: 40 kV, 1000 AA, Imaging conditions: 64�50 pixel matrix, 200 ms dwell time per pixel.

T.C. Miller et al. / Spectrochimica Acta Part B 60 (2005) 1458–14671464

intensity may not be enough to overcome background in the

bulk sample. This shows the importance of analyzing both the

bulk and single particle composition of a sample to completely

characterize a given specimen.

In addition to different composition, heterogeneous samples

are often made up of particles of different sizes and morphology.

It is advantageous to separate and elementally characterize

0

1000

2000

3000

4000

5000

6000

0 2 4

Ene

Inte

nsi

ty (

cou

nts

)

Si

Ca

K

T

Al

Rh scatter

Fig. 7. Point spectra of a bulk pellet and a single particle of CRM CJ-2 simulated As

35 kV, 220 AA for the pellet, Spectral dwell time=100 live seconds.

particulate size fractions for determination of chemical compo-

sition and source attribution of a given specimen. Size

fractionated collection of particulates is usually carried out with

an impactor. Cascade impactors segregate aerosols by size based

on their particulate inertial characteristics [34]. The MOUDI,

used in this study, is a 10-stage cascade impactor with stages

having 50% cut-points ranging from 0.056 to 18 Am in

6 8 10

rgy (KeV)

Pellet

Particle

i

Fe

Cu

V Mn

ian mineral dust. X-ray tube conditions: 40 kV, 1000 AA for the single particle,

Page 8: Characterization of small particles by micro X-ray ... › ~gcas › esp2 › papers › miller05.pdf · For example proton-induced X-ray emission (PIXE) has been employed for the

Fig. 8. MXRF Ca elemental images of CRM CJ-2 particulates collected on polypropylene film supports from 8 of the different stages of a MOUDIi cascade

impactor. No Ca particles were observed for stages 9 and 10. X-ray tube conditions: 40 kV, 325 AA, Imaging conditions: 128�100 pixel matrix, 200 ms dwell time

per pixel. Particle sizes at each stage are 1) �10 Am, 2) 5.6–10 Am, 3) 2.5–5.6 Am, 4) 1.8–2.5 Am, 5) 1.0–1.8 Am, 6) 0.56–1.0 Am, 7) 0.32–0.56 Am, 8) 0.18–

0.32 Am, 9) 0.10–0.18 Am, and 10) 0.056–0.10 Am.

T.C. Miller et al. / Spectrochimica Acta Part B 60 (2005) 1458–1467 1465

aerodynamic diameter. The principle of operation is straightfor-

ward and has been discussed in detail in references [35] and [36].

The MOUDI collects the aerosol in 10 discrete size-fractionated

samples so as to provide information on the distribution of

chemical components as a function of particle size.

Fig. 8 shows the Ca elemental images of particulates

collected at each MOUDI stage, 1 through 8. No images are

shown for stages 9 and 10 because no Ca particles were

collected above stage 8. MXRF can be used to both image and

take point spectra of the particulates collected at each of the

stages. Imaging is used to observe the impaction pattern of the

particles at each impactor stage as well as give the relative

intensities (abundance) of the elements in the collected

0 2 4 6 8

Energy (keV)

AlSi

Cl CaTK

Fig. 9. Point spectra of the center impaction spot for CRM CJ-2 particulates separa

Spectral dwell time=100 live seconds.

fractions. Fig. 9 shows point spectra taken at the particle

impaction spot closest to the center of the impaction substrate

for each of the impaction stages. The point spectra show

changes in elemental intensities between the different particle

sizes. For example, the Fe intensity steadily decreases from

stage 1 to stage 10. Other elements, such as Si, show varying

intensity over the different fractions.

4. Conclusion

Micro X-ray fluorescence is a new tool that can be used for

particle sample characterization. Through the incorporation of

polycapillary X-ray focusing optics in readily available

10

0

5000

10000

15000

20000

25000

30000

35000

01

23

45

67

89

10

Stage

Inte

nsi

ty (

cou

nts

)

Fe

i Cr

ted in 10 stages by cascade impaction. X-ray tube conditions: 40 kV, 325 AA,

Page 9: Characterization of small particles by micro X-ray ... › ~gcas › esp2 › papers › miller05.pdf · For example proton-induced X-ray emission (PIXE) has been employed for the

T.C. Miller et al. / Spectrochimica Acta Part B 60 (2005) 1458–14671466

laboratory bench top instrumentation, the technique overcomes

many of the drawbacks of more conventional X-ray methods

for particle analysis. The smaller beam diameter and high X-

ray flux achieved with MXRF allows for very small sample

features to be observed and characterized. Both homogeneous

and heterogeneous particulate systems can be nondestructively

analyzed in both air and under vacuum using both single point

analyses and elemental imaging with the ability to analyze

small features as well as scan over large areas to map the

sample elemental characteristics.

Single particle elemental analysis can be performed if

samples are prepared such that the particles are well separated

with minimum separation distance greater than or equal to the

excitation source beam diameter. The best methods for

preparation of single particles are air dispersion methods where

the particles are separated and immobilized in an array on a low

background support material such as polypropylene film.

Individual particles can be detected with the instrumentation

used in this study down to a particle mass of ¨0.04 ng and a

mean particle diameter of ¨0.06 Am. Bulk particle samples,

which use a large quantity of collected single particles to make

a single homogeneous specimen, were successfully prepared

for MXRF analysis using both pellet preparation and aerosol

impaction. Homogeneous particles, such as glass microspheres,

and heterogeneous systems, such as soil particles, were both

successfully prepared by bulk and single particle methods and

analyzed by both single point spectral analysis and elemental

imaging. The elemental signatures of homogeneous particu-

lates can be determined using MXRF for either bulk and/or

single particle analysis. Heterogeneous samples are much

harder to characterize, and both single particle as well as bulk

analyses must be performed on sample specimens to be fully

described by MXRF.

Future work will focus on integrating MXRF analysis with

other spectroscopy techniques as particle samples will often

contain both inorganic and organic components. MXRF is a

nondestructive technique and leaves samples intact for other

types of analyses, such as IR or Raman spectroscopies or X-ray

diffraction, that will allow for characterization of the molecular

components of such samples. The use of MXRF in conjunction

with these established methods of molecular analysis will allow

for a more complete characterization of the particles.

References

[1] P.E. Cruvinel, R.G. Flocchini, P. Artaxo, S. Crestana, P.S.P. Herrmann Jr.,

Elemental analysis of agricultural soil samples by particle induced X-ray

emission (PIXE) technique, Nucl. Instrum. Methods Phys. Res., Sect. B

150 (1999) 478–483.

[2] M. Abdulla, M.B. Zaman, M. Khaliquzzaman, A.H. Khan, Determination

of total zinc in soils by external-beam proton-induced X-ray emission

spectrometry, Anal. Chim. Acta 118 (1980) 175–178.

[3] H. Van Malderen, S. Hoornaert, J. Injuk, W.J. Przybylowicz, C.A. Pineda,

V.M. Prozesky, R. Van Grieken, Individual particle characterization of

Siberian aerosols by micro-PIXE and backscattering spectrometry, X-ray

Spectrom. 30 (2001) 320–329.

[4] G. Papaspiropoulos, B. Mentes, P. Kristiansson, B.G. Martinsson, A high

sensitivity elemental analysis methodology for upper tropospheric aerosol,

Nucl. Instrum. Methods Phys. Res., Sect. B 150 (1999) 356–362.

[5] B. Treiger, J. Injuk, I. Bondarenko, P. Van Espen, R. Van Grieken, L.

Breitenbach, U. Watjen, Non-linear mapping of microbeam proton-

induced X-ray emission data for source identification of North Sea

aerosols, Spectrochim. Acta Part B 49 (1994) 345–353.

[6] W. Maenhaut, R. Hillamo, T. Makela, J.L. Jaffrezo, M.H. Bergin, C.I.

Davidson, A new cascade impactor for aerosol sampling with subsequent

PIXE analysis, Nucl. Instrum. Methods Phys. Res., Sect. B 109/110

(1996) 482–487.

[7] J. Injuk, R. Van Grieken, R. Klockenkamper, A. von Bohlen, P. Kump,

Performance and characteristics of two total-reflection X-ray fluorescence

and a particle induced X-ray emission setup for aerosol analysis,

Spectrochim. Acta Part B 52 (1997) 977–984.

[8] A. von Bohlen, H. Berink-Kloke, C. Althoff, Elemental determination in

medieval soil samples by total reflection X-ray fluorescence, Anal. Chim.

Acta 480 (2003) 327–335.

[9] S. Chimidza, A. Viksna, E. Lindgren, EDXRF and TXRF analysis of

aerosol particles and the mobile fraction of soil in Botswana, X-ray

Spectrom. 30 (2001) 301–307.

[10] M. Ebert, J. Dahman, P. Hoffman, H.M. Ortner, Examination of clean

room aerosol particle composition by total reflection X-ray analysis

and electron probe microanalysis, Spectrochim. Acta Part B 52 (1997)

967–975.

[11] M. Schmeling, R. Klockenkamper, D. Klockow, Application of total-

reflection X-ray fluorescence spectrometry to the analysis of airborne

particulate matter, Spectrochim. Acta Part B 52 (1997) 985–994.

[12] M. Schmeling, D. Klockow, Sample collection and preparation for

analysis of airborne particulate matter by total reflection X-ray fluores-

cence spectrometry, Anal. Chim. Acta 346 (1997) 121–126.

[13] T. Stahlschmidt, M. Schulz, W. Dannecker, Application of total-

reflection X-ray fluorescence for the determination of lead, calcium

and zinc in size-fractionated marine aerosols, Spectrochim. Acta Part B

52 (1997) 995–1001.

[14] M. Theisen, R. Niessner, Sapphire sample carriers for silicon determina-

tion by total-reflection X-ray fluorescence analysis, Spectrochim. Acta

Part B 54 (1999) 1839–1848.

[15] J. Injuk, R. Van Grieken, Optimization of total-reflection X-ray

fluorescence for aerosol analysis, Spectrochim. Acta Part B 50 (1995)

1787–1803.

[16] A.C. John, T.A.J. Kuhlbusch, H. Fissan, K.G. Schmidt, Size-fractionated

sampling and chemical analysis by total-reflection X-ray fluorescence

spectrometry of PMx in ambient air and emissions, Spectrochim. Acta Part

B 56 (2001) 2137–2146.

[17] M.C. Camerani, A. Somogyi, M. Drakopoulos, B. Steenari, Synchrotron

radiation induced A-X-ray fluorescence spectroscopy on municipal solid

waste fly ashes, Spectrochim. Acta Part B 56 (2001) 1355–1365.

[18] I. Orlic, S.Y. Chiam, J.L. Sanchez, S.M. Tang, Quantitative analysis of

cascade impactor samples—revisited, Nucl. Instrum. Methods Phys. Res.,

Sect. B 150 (1999) 465–469.

[19] K.H.A. Janssens, F.C.V. Adams, A. Rindby (Eds.), Microscopic X-ray

Fluorescence Analysis, John Wiley & Sons, LTD, New York, 2000.

[20] J. Jernstrom, M. Eriksson, J. Osan, G. Tamborini, S. Torok, R. Simon, G.

Falkenberg, A. Alsecz, M. Betti, Non-destructive characterization of low

radioactive particles from Irish Sea sediment by micro X-ray synchrotron

radiation techniques: micro X-ray fluorescence (A-XRF) and micro

absorption near edge structure (A-XANES) spectroscopy, J. Anal. At.

Spectrom. 19 (2004) 1428–1433.

[21] B. Alfoldy, S. Torok, A. Kocsonya, Z. Szodblackefalvi-Nagy, M.I. Balla,

X-ray analysis of aerosol samples from a therapeutic cave, Nucl. Instrum.

Methods Phys. Res., Sect. B 174 (2001) 361–366.

[22] B. Holynska, J. Ptasinski, W. Maenhaut, H.J.J. Annegarn, Energy-

dispersive X-ray fluorescence spectrometer with capillary optics for the

chemical analysis of atmospheric aerosols with high time resolution,

Aerosol Sci. 28 (1997) 1455–1463.

[23] O. Haupt, K. Linnow, R. Harmel, C. Shaefer, W. Dannecjer, Qualitative

X-ray fluorescence analysis of emitted aerosol particles from incineration

plants sampled on quartz fibre filters, X-ray Spectrom. 26 (1997) 79–84.

[24] M.J. Anjos, R.T. Lopes, E.F.O. Jesus, J.T. Assis, R. Cesareo, R.C.

Barroso, C.A.A. Barradas, Elemental concentration analysis in soil

Page 10: Characterization of small particles by micro X-ray ... › ~gcas › esp2 › papers › miller05.pdf · For example proton-induced X-ray emission (PIXE) has been employed for the

T.C. Miller et al. / Spectrochimica Acta Part B 60 (2005) 1458–1467 1467

contaminated with recyclable urban garbage by tube-excited energy-

dispersive X-ray fluorescence, Radiat. Phys. Chem. 65 (2002) 495–500.

[25] K.N. Yu, Z.L.L. Yeung, L.Y.L. Lee, M.J. Stokes, R.C.W. Kwok,

Determination of multi-element profiles of soil using energy dispersive

X-ray fluorescence (EDXRF), Appl. Radiat. Isot. 57 (2002) 279–284.

[26] A.G. Revenko, X-ray fluorescence analysis of rocks, soils, and sediments,

X-ray Spectrom. 31 (2002) 264–273.

[27] Z. Hartyani, E. David, S. Szabo, T. Horvath, T. Toth, A. Hargitai,

Determination of trace elements distribution of polluted soils in Hungary

by X-ray methods, Microchem. J. 67 (2000) 195–200.

[28] M.J. dos Anjos, R.T. Lopes, E.F.O. de Jesus, J.T. Assis, R. Cesareo,

C.A.A. Barradas, Quantitative analysis of metals in soil using X-ray

fluorescence, Spectrochim. Acta Part B 55 (2000) 1187–1192.

[29] A. Johnson, G.C. Lalor, H. Robotham, M.K.J. Vutchkov, Analysis of

Jamaican soils and sediments by energy-dispersive X-ray fluorescence

spectrometry, Radioanal. Nucl. Chem. 209 (1996) 101–111.

[30] SPI Supplies, Division of Structure Probe, INC., West Chester, PA, 1995,

pp. 1–10.

[31] R.E. Van Grieken, A.A. Markowicz (Eds.), Handbook of X-ray Spec-

trometry, Second edition, Marcel Dekker, Inc., New York, 2002, p. 940.

[32] R. Jenkins, X-ray Fluorescence Spectrometry, Second edition, John Wiley

& Sons INC., New York, 1999.

[33] National Institute of Standards, 1998; Vol. (2003).

[34] L.A. De Bock, H. Van Malderen, R. Van Grieken, Individual Aerosol

Particle Composition Variations in Air Masses Crossing the North Sea,

Environ. Sci. Technol. 28 (1994) 1513–1520.

[35] V.A. Marple, K.L. Rubow, S.M. Behm, A Microorifice Uniform Deposit

Impactor (MOUDI): description, calibration, and use, Aerosol Sci.

Technol. 14 (1991) 434–446.

[36] M.D. Keywood, G.P. Ayers, J.L. Gras, R.W. Gillett, D.D. Cohen,

Relationships between size segregated mass concentration data and

ultrafine particle number concentrations in urban areas, Atmos. Environ.

33 (1999) 2907–2913.