Effect of dwell time on single particle inductively
coupled plasma mass spectrometry data
acquisition quality
Aaron Hineman and Chady Stephan
The characterization, sizing, and quantification of metal-based nanoparticles (NP) in a variety of matrices
using single particle-inductively coupled plasma-mass spectrometry (SP-ICP-MS) is becoming
increasingly popular due to the sensitive nature of the technique. Nanoparticle events in the plasma are
less than 0.5 ms in duration; however current quadrupole-based ICP-MS instruments are limited to
instrument dwell times in the millisecond range and have data acquisition overhead that adversely affects
data quality. Novel instrument settings and data processing techniques can be used to explore the
benefits of continuous data acquisition rates as fast as 105
Hz (or 10 ms dwell times). This paper provides
data on the different effects data acquisition rate has on the quality of data that can be obtained by SPICP-MS.
The effect of varying the dwell time and its influence on particle integration, particle counting,
particle sizing, and background signal is discussed. This paper provides data on identifying the significant
instrument settings and their implications on nanoparticle characterization.
Introduction
The characterization and quantication of metal-based nanoparticle
(NP) suspensions utilizing ICP-MS is gaining
momentum. The main advantage of using ICP-MS as a detector
is its high sensitivity when compared to other particle characterization
techniques, such as dynamic light scattering, differential
centrifugal sedimentation, and eld ow fractionation.1
Avoiding dilution of environmental samples reduces the
possibility of changing the sample chemistry affecting any NP
determination due to transformation of the particles. In addition
exposure concentrations of NPs in aqueous environments
are speculated to be in the range of 103
to 105
particles per mL.2,3
One of the limitations of many quadrupole-based ICP-MS
instruments is that they have a minimum dwell times in the
millisecond range and are designed to settle between each
measurement, thereby missing parts of the transient signal.
However, short transient signals generated by individual NP
events in the ICP are less than about 0.5 ms in length.4
Because
the duration of transient signals generated by individual NPs
are in the range of hundreds of microseconds, their detection
involves the use of instruments capable of fast data acquisition
at frequencies around 105
Hz (or 10 ms).5
The objective of this paper is to present experimental data in
evaluating ICP-MS data acquisition speed on particle integration,
particle counting, particle sizing, and the background
signal-ultimately improving the counting and sizing of NPs
using SP-ICP-MS. In that regard, the benets of collecting data
using dwell times as short as 10 ms, continuously without any
settling is demonstrated, with data being provided as proof of
concept.
Theory
Single particle-inductively coupled plasma-mass spectrometry
(SP-ICP-MS) is based on the assumptions that each transient
peak or pulse represents a single particle event, and the difference
in the continuous baseline signal of the blank and nanoparticle
sample represents the dissolved concentration of the
element. The frequency of the events is directly proportional to
the number of particles (or agglomerates) entering the plasma,
while the intensity of the event (i.e. transient event area) is
proportional to the particle diameter or the mass of the element
within the particle. Once the transport efficiency of the introduction
system is determined, the particle concentration in the
delivered suspension of NPs can then be calculated. There may
be a background response on that mass due to contamination,
spectral overlap ions or instrument background. The change in
that continuous response (on top of which are the NP events) is
directly proportional to the dissolved concentration of the
element.
The number of particles entering the plasma is dependent
on the particle concentration in the sample, the sample introduction
system used (i.e. the transport efficiency to the plasma),
the sample uptake rate, and the sample matrix. Although not
specically covered in this paper, Pace et al. present information
and theory on how to establish the system transport
PerkinElmer Inc, Woodbridge, Ontario, Canada. E-mail: Aaron.Hineman@
PerkinElmer.com; Chady.Stephan@PerkinElmer.com
Cite this: J. Anal. At. Spectrom., 2014,
29, 1252
Received 11th March 2014
Accepted 14th April 2014
DOI: 10.1039/c4ja00097h
www.rsc.org/jaas
1252 | J. Anal. At. Spectrom., 2014, 29, 1252–1257 This journal is © The Royal Society of Chemistry 2014
JAAS
PAPER
Publishedon14April2014.DownloadedbyMasarykovaUniverzitaVBrneon15/12/201423:19:41.
View Article Online
View Journal | View Issue
efficiency.6
Following nebulization, the NPs randomly arrive in
the plasma; therefore the number of particles must be low
enough to separate the ion plumes produced from the ionization
of individual NPs. The probability of more than one NP
being measured within a dwell time (coincidence) can be estimated
by Poisson statistics.4
This probability of coincidence
decreases by decreasing the dwell time as well as the concentration
of NPs in the suspension.
Capture of the entire peak event has important implications
on how to precisely and accurately determine particle concentration
and the mass of the element in each particle. It has been
established that the full width of a particle event is less than
0.5 ms;4
therefore the choice of instrument dwell time will affect
the maximum number of particles that can be introduced to the
plasma within a given amount of time without signals from
individual particles overlapping. If the dwell time is not short
enough (i.e. equal to or less than the event width of $0.5 ms),
the instrument response will be the summation of the particles,
resulting in an articially larger event response and a reduced
number of event counts, thus affecting particle mass and
concentration calculations (Fig. 1). In most literature to date,
dwell times between 3 and 10 ms are commonly used. With
such long dwell times anywhere between 6 and 20 particle
events can occur during each dwell time (assuming two particles
never arrive in the plasma at the same time). However, if the
system is limited to millisecond level dwell times or the particle
concentration is too high then the sample can simply be diluted
to reduce the probability of coincidence.
Another important consideration of the ICP-MS data acquisition
is that it is not interrupted for data processing or other
hardware operations. Most commercial instruments invoke a
settling time during data acquisition. During this settling time,
events are missed or partial events are captured (Fig. 1). Missing
events lead to the requirement that the users increase their total
acquisition time so that counting statistics improve. For
example, if the settling time equalled the dwell time it would
take twice as long to obtain the same number of data points (or
events) compared to a system that did not require any settling.
In the case of partial event capture, a low bias in the event
intensities can be created resulting in a low-biased particle size
distribution.
Instrumental
All data in this work was acquired with a PerkinElmer NexION
series ICP-MS (Shelton, CT) operating in single particle mode, a
mode optimized for collection of NP event data, using the
Syngistix Nano Application Module. Based on the challenges
described above relating to single particle analysis, optimal data
acquisition requires extremely fast data acquisition rates. This
is accomplished by eliminating system settling time and
allowing data acquisition rates as fast as 105
Hz (10 ms dwell
times). The elimination of the settling time avoids the issues
due to partial event capture as illustrated in Fig. 1. The fast rate
of data acquisition enables the user to capture up to 6 million
data points per minute, thereby allowing for baseline (or down
to the dissolved background response) resolution of multiple
events. These capabilities are demonstrated in Fig. 2 and 3.
Fig. 2 is a screen capture of a zoomed in portion of the raw data
collected at a dwell time of 50 ms when analysing a suspension
of NIST 8013 60 nm gold NPs with a particle concentration of
approximately 250 000 particles per mL of solution. Despite the
high particle concentration, baseline separation of the events is
Fig. 1 Effect of dwell and settling times on single particle measurements.
(a) 2 particles detected (b) 1 single particle detected (c) the
leading edge of 1 particle detected (d) the trailing edge of 1 particle
detected (e) no particles detected.
Fig. 2 Zoomed in trace of NIST SRM 8013 60 nm gold nanoparticle
standard at a dwell time of 50 ms. Particle concentration of 250 000
particles per mL.
Fig. 3 Zoomed in of four NP events at a dwell time of 50 ms. Particle
concentration of 250 000 particles per mL.
This journal is © The Royal Society of Chemistry 2014 J. Anal. At. Spectrom., 2014, 29, 1252–1257 | 1253
Paper JAAS
Publishedon14April2014.DownloadedbyMasarykovaUniverzitaVBrneon15/12/201423:19:41.
View Article Online
easily achieved, thus reducing the probability of coincidence. In
Fig. 3 the graph is further zoomed in, demonstrating the
resolving power of the single particle mode analysis.
Results & discussion
Effect of dwell time on particle integration
A major benet of the instrument's single particle mode is that
multiple points can be measured to dene a single particle
event, and the area of the peak can be related to the mass of the
element within the NP. The ability to view the full event provides
the user with insight into the quality of the data by displaying a
visual conrmation of NP event separation. Fig. 4–6 show NP
events using dwell times of 50 ms, 30 ms, and 10 ms, respectively.
Each of these gures displays an event of approximately 300–
400 ms in duration indicating consistent system performance
regardless of the data acquisition rate. Since there is a normal
distribution around the nominal size of the particle, the event
response and duration will vary slightly from event to event.
However by reviewing the number of points per peak in Fig. 4–6,
the advantage of shorter dwell times is emphasized in the
resulting single event prole. The number of points per peak is
summarized in Table 1.
Effect of dwell time on particle counting: number of particles
per mL
In theory the upper limit to the number of particles that can be
measured within a given time frame is limited by the duration
of the NP event. Assuming the entire signal from one particle
lasts no more than 0.5 ms (see Fig. 4–6), then the maximum
number of particles per second that could be measured would
be 2000 (1 per 0.0005 s).4
However this assumes that the particles
are not arriving randomly. In actuality, particles arrive
randomly, the result of introducing heterogeneous suspensions
with a pneumatic nebulizer. This must be taken into consideration
when choosing sample introduction parameters.
As an example, two experiments were conducted where the
rate of sample introduction was changed while the number of
particles in the solution was kept constant at 250 000 particles
per mL. Fig. 7 displays the two sets of data where the number of
particle events is plotted vs. the dwell time used for data
acquisition. Note that the transport efficiencies of the two ow
rates were different. As illustrated, increasing the ow rate
increases the event rate (peaks per minÀ1
), which in turn
increases coincidence. When comparing the two sets of data,
the reduced ow rate (300 mL minÀ1
) allows for longer dwell
times to be used without any signicant amount of coincidence
occurring. A closer look at the two datasets shows that a
signicant increase in coincidence can be seen earlier in the
higher ow rate (450 mL minÀ1
) data (around 100 ms) than in the
Fig. 4 Raw data of a NIST SRM 8013 60 nm gold nanoparticle acquired
at a dwell time of 50 ms.
Fig. 5 Raw data of a NIST SRM 8013 60 nm gold nanoparticle acquired
at a dwell time of 30 ms.
Fig. 6 Raw data of a NIST SRM 8013 60 nm gold nanoparticle acquired
at a dwell time of 10 ms.
Table 1 Summary of points per peak and event duration for different
dwell times of single nanoparticle events
Dwell time (ms) Points integrated Event duration (ms)
50 8 400
30 13 390
10 31 310
1254 | J. Anal. At. Spectrom., 2014, 29, 1252–1257 This journal is © The Royal Society of Chemistry 2014
JAAS Paper
Publishedon14April2014.DownloadedbyMasarykovaUniverzitaVBrneon15/12/201423:19:41.
View Article Online
300 mL minÀ1
data (around 200 ms). Another important
consideration is that no peaks will be measured if the dwell
time is increased to the point where the NP events begin to form
a continuum response. Upon review of the data represented by
Fig. 7, this event effectively occurs at 3 ms and 4 ms for the
sample ow rates of 450 mL minÀ1
and 300 mL minÀ1
, respectively.
The most noteworthy nding resulting from this study is
that with fast acquisition (less than 100 ms) the system is more
easily able to resolve a greater number of NP events per second,
reducing the need for excessive dilution and further changes in
the sample matrix.
The effect of dwell time on particle counting was further
investigated by processing each of the size histograms generated
during this study, with the objective of excluding events outside
the primary Gaussian distribution. Fig. 8 below shows an
example of a size histogram from the analysis of NIST SRM 8013
60 nm gold nanoparticles at a concentration of 250 000 particles
per mL acquired at a dwell time of 50 ms. The data derived from
each histogram is the number of particle events, mean particle
counts and the corresponding mean particle diameter. The
particle size histograms were then divided into the primary
distribution of the particles and the remainder of events greater
than and outside the primary distribution. Each section of the
histogram will then have the associated the number of particle
events, mean particle counts and the corresponding mean
particle diameter. It can be then assumed that for the particle
responses greater than the primary distribution are due to
coincidence and/or any agglomerates in the solution.
This treatment of the data allows for further investigation of
the effect of dwell time on particle counting. Fig. 9 shows the
data from the 300 mL minÀ1
ow rate processed to view the total
number of events, events within the primary distribution and
the remaining particle events (coincidence and/or
agglomeration).
As displayed, the number of particle events in the primary
distribution and the total number of particle events decrease as
the dwell time increases, restating our earlier nding about the
increased rate of coincidence with longer dwell time. This effect
can also be represented by ratioing the number of remaining
particle events to the total number of events at each dwell time,
(Fig. 10). This shows that at dwell times above 100 ms, the
number of events outside the primary distribution is greater
than 10% of the total events under the above stated data
acquisition conditions.
To further investigate the effect of dwell time on the rate of
coincidence and its effect on particle counting, we posed that
Fig. 7 Particle events vs. dwell time for 250 000 particles per mL of 60
nm Au NPs nebulized at different flow rates.
Fig. 8 Size histogram of NIST SRM 8013 60 nm gold nanoparticles at a
concentration of 250 000 particles per mL acquired at a dwell time of
50 ms.
Fig. 9 Particle events vs. dwell time for 250 000 particles per mL of
60 nm Au NPs nebulized at 300 mL minÀ1
. Data separated by grouping
events in the primary distribution.
Fig. 10 Percent of particle events outside the primary distribution vs.
dwell time for 250 000 particles per mL of 60 nm Au NPs nebulized at
300 mL minÀ1
.
This journal is © The Royal Society of Chemistry 2014 J. Anal. At. Spectrom., 2014, 29, 1252–1257 | 1255
Paper JAAS
Publishedon14April2014.DownloadedbyMasarykovaUniverzitaVBrneon15/12/201423:19:41.
View Article Online
the number of remaining particle events under the minimum
dwell time of 10 ms was completely due to agglomeration (i.e. no
coincidence at all). Under this assumption, the relative number
of events due to agglomeration was subtracted from the
remaining particle events to give the number of events that had
a greater probability of being due to coincidence. Under this
assumption, the amount of events most likely due to coincidence
is greater than 10% at dwell times above 500 ms in
agreement with what it was demonstrated earlier with regard to
the duration of single particle transient signal. Using a dwell
time that is greater than the individual nanoparticle transient
signal will inevitably increase the rate of particle coincidences
leading to inaccurate particle counting and sizing.
Effect of dwell time on particle sizing
To evaluate the effects of dwell time on particle size calculation
and coincidence counting, a two point particle size calibration
was performed using NIST RM 8012 and NIST RM 8013 gold
nanoparticles with nominal diameters of 30 nm and 60 nm,
respectively. Both Au NPs standards were prepared at a
concentration of approximately 250 000 particles per mL. The
calibration was constructed using 50 ms dwell time at an aspiration
rate of 300 mL minÀ1
. Once the calibration was obtained,
the same RM 8013 standard was continuously aspirated as the
dwell time was varied from 10 ms to 10 000 ms. Fig. 11 shows the
effect of dwell time on the mean particle intensity. As expected,
with longer dwell time more coincidences occur. The rate of
coincidence increases steadily as the dwell time increase where
an exponential increase is noticed once the dwell time reaches
3000 ms. Therefore one can state at 3000 ms, with the sample
introduction conditions used and particle concentration of the
solution, the NP responses begin to form a continuum and
cannot be differentiated from background signal.
As a result, the event response increases which correlates to a
greater mass of the element, resulting in a bias on particle size
(Fig. 12). Note that the NIST certied value7
of the particle size is
approximately 56 nm as determined by TEM which is reected
in the results presented in Fig. 12 up to the 500 ms dwell time.
Using the data reduction techniques (described in the previous
section) one can alleviate the effect of dwell time on particle
sizing. However at longer dwell times, greater than 3 ms, this
data reduction technique reaches its limitation due to the
increased level of coincidence encountered.
Effect of dwell time for various particle concentrations
As discussed in the previous sections, the rate of coincidence
increases with increased particle concentration and increased
dwell times. To further investigate this fact and its implications
on various particle concentrations, another experiment was
conducted where solutions of 50 nm gold NP concentrations
ranging from 100 000 to 500 000 particles per mL were prepared
and analysed at dwell times of 50, 100, 500, 1500, and 3000 ms.
In order to calculate particle concentration for each analysis,
the system transport efficiency was established using the
100 000 particles per mL solution at a dwell time of 50 ms. The
ow rate was held constant at approximately 300 mL minÀ1
. The
effect of dwell time on the particle counting efficiency of this
experiment is summarized in Fig. 13. The particle counting
efficiency was calculated from the events in the primary distribution
by comparing the determined particle concentration to
the target concentration of each prepared solution. The data
obtained supports the data in Fig. 10 that displays increasing
the dwell time increases coincidence and, in addition, provides
Fig. 11 Effect of dwell time on mean particle intensity for 250 000
particles per mL of 60 nm Au NPs nebulized at 300 mL minÀ1
.
Fig. 12 Effect of dwell time on mean particle size calculation for
250 000 particles per mL of 60 nm Au NPs nebulized at 300 mL minÀ1
.
Fig. 13 The effect of dwell time and NP concentration on particle
counting.
1256 | J. Anal. At. Spectrom., 2014, 29, 1252–1257 This journal is © The Royal Society of Chemistry 2014
JAAS Paper
Publishedon14April2014.DownloadedbyMasarykovaUniverzitaVBrneon15/12/201423:19:41.
View Article Online
insight into the fact that increasing the particle concentration
will also increase the rate of coincidence.
The effect of dwell time on particle sizing of this experiment
is shown in Fig. 14. The mean particle size is calculated from the
mean event response from the primary distribution. When the
primary distribution is selected, the error in sizing was not as
large as counting since the mean particle response, even when
including responses due to coincidence, will remain closer to
the true value until the rate of coincidence becomes too high
and biases the mean. The data obtained supports the data in
Fig. 12 that displays increasing the dwell time slightly increases
the particle size of the primary distribution until the mean
response is obtained mostly from coincident events creating a
signicant bias in the calculated size as what is seen at 3000 ms.
Effect of dwell time on the background signal
The signal from the dissolved component of the element in the
solution is directly proportional to the difference between the
continuous background in the blank and nanoparticle sample.
Therefore another consequence of using too long a dwell time is
that it becomes increasingly difficult to differentiate the particles
from the background since the continuous background
signal increases with dwell time while the signal from a single
particle does not. As shown in Fig. 15, the background response
increases proportionally to the dwell time. However, the signal
due to a single NP is independent of the dwell time as long as
only one particle is present within the dwell time window and
the signal resulting from each single particle is completely
within the dwell time.4
This change in the apparent continuous
background signal will adversely affect the minimum size
particle that can be detected since the background response
must be subtracted from the NP response creating greater error
in the NP size determination due to the greater noise amplitude
in the elevated background. Therefore faster data acquisition
(dwell times of 100 ms or less) allows smaller particles to be
measured compared to data obtained with longer dwell times.
The large increase in background signal at 3000 ms in Fig. 15
indicates that more coincidence is occurring (at this particle
concentration) and where the NPs cannot be differentiated from
background signal. This effect was similarly illustrated in Fig. 7
and Fig. 9–14 where the particle counting approached one
because the system was at the point where the NP events begin
to form a continuum response due to the longer dwell times.
Conclusion
This paper summarizes the different effects that data acquisition
rate (dwell time + settling time) have on the quality of data that
can be obtained by SP-ICP-MS. The single particle mode of the
NexION, with no settling time and dwell times as low as 10 ms,
has the ability to provide the user with data on the particle
concentration, particle size, and the dissolved concentration of
the elements in solution. To avoid errors due to partial NP event
integration and coincidence with long dwell times, concentrated
nanoparticle solutions have to be diluted signicantly. Excessive
sample dilution is to be avoided, as it may create greater
uncertainty in NP counting due to subsequent manipulation and
changes to the sample chemistry affecting nanoparticle transformation
(dissolution, agglomeration, etc.).
Notes and references
1 M. Hassellov, J. Readman, J. F. Ranville and K. Tiede,
Ecotoxicology, 2008, 17(6), 344–361.
2 A. Boxall, K. Tiede and Q. Chaudhry, Nanomedicine, 2007,
2(6), 919–927.
3 F. Gottschalk, T. Sonderer, R. W. Scholz and B. Nowack,
Environ. Sci. Technol., 2009, 43(24), 9216–9222.
4 J. W. Olesik and P. J. Gray, J. Anal. At. Spectrom., 2012, 27,
1143–1155.
5 C. Engelhard, Anal. Bioanal. Chem., 2011, 399(1), 213–219.
6 H. Pace, N. Rogers, C. Jarolimek, V. Coleman, C. Higgins and
J. Ranville, Anal. Chem., 2011, 83, 9361–9369.
7 V. Hackley, Report of Investigation - Reference Material®
8013., from National Institute of Standards and Technology,
https://www-s.nist.gov/srmors/reports/8013.pdf, accessed
February 2014.
Fig. 14 The effect of dwell time and NP concentration on particle
sizing.
Fig. 15 Effect of dwell time on background signal.
This journal is © The Royal Society of Chemistry 2014 J. Anal. At. Spectrom., 2014, 29, 1252–1257 | 1257
Paper JAAS
Publishedon14April2014.DownloadedbyMasarykovaUniverzitaVBrneon15/12/201423:19:41.
View Article Online