metric studies of ion-molecule clus- “purest” form of transfer of ions from

ters in the gas phase (I)have con- solution to the gas phase; little if any
tributed much to the understanding extra internal energy is imparted to
of ion-solvent molecule interactions the ions.
-
in solution. It is much easier to un- ES affords ion transfer from solu-
d e r s t a n d t h e n a t u r e of t h e ion- tion to t h e gas phase for a large
Paul Kebarle and Liang Tang solvent bonding when the ion inter- range of ion types: the simple singly
Department of Chemistry acts with only one or a few solvent charged electrolytes such as Na’ and
University of Alberta molecules, and it is these initial in- C1- and organic protonated bases
Edmonton, Alberta teractions t h a t dominate ion solva- BH+ and X-;group I1 ions such as
Canada T6G 2G2
tion in solution (1). Ca”, SP’, and Ba2+,as well as dou-
The transfer of ions from solution bly a n d triply charged transition
It is no exaggeration to say that close to the gas phase is a desolvation pro- metal and lanthanide ions and com-
to half of inorganic chemistry, or- cess, and thus it is strongly “unnatu- plexes thereof; bioorganic ions such
ganic chemistry, and biochemistry ral” or endoergic. The free energy as multiply- protonated peptides and
involves ions in solution. A technique required when a mole of Na’ is proteins of molecular mass 100,000
t h a t allows ions to be transferred transferred from aqueous solution to Da; and multiply-deprotonated, neg-
from solution to the gas phase and the gas phase is very large (2) atively charged nucleic acids.
subjected to m a s s spectrometric ESMS was introduced by Yamashi-
analysis should therefore be of enor- Na’ (aq) + Na’ (g) ta and Fenn in 1984 (31, and it took a
mous importance. Electrospray (ES) -AGLl(Na+) = 98.2 kcaYmol few years before the importance of
is such a technique. the method was recognized (4-6). Re-
The transfer of ions from the gas Analytical m a s s spectrometric c e n t advances (7-9)confirm t h e
phase to solution is a natural pro- methods in which ions a r e “trans- widely held view t h a t ES has ush-
cess. In the presence of solvent mole- ferred” from solution to t h e g a s ered in a new era of the mass spec-
cules such as H,O, naked gas-phase phase, such a s fast atom bombard- trometric analysis of biomolecules.
ions such as H30+ or Na+ will spon- ment (FAB), plasma desorption, and The exciting applications of ESMS
taneously form ion-solvent molecule laser desorption, existed before the have also created great interest in
clusters such a s H,O+(H,O), a n d introduction of electrospray mass t h e mechanism by which t h e gas-
Na+(H,O),. If the pressure of the sol- spectrometry (ESMS). The energy phase ions required for t h e mass
vent vapor is somewhat above the required for ion transfer to the gas spectrometric analysis are produced
s a t u r a t i o n vapor pressure, these phase in these earlier methods is by the ES device. Ideally, a n under-
clusters will grow to small droplets. supplied by complex high - energy col- standing of the mechanism should
To visualize this process one needs lision cascades and highly localized not only provide aesthetic satisfac-
only to t h i n k of t h e Wilson cloud heating, resulting in additional pro- tion, it should be of significant prac-
chamber, where such a process oc- cesses such as net ionization (creation tical use. Understanding the mecha-
curs in front of one’s eyes. of ions from neutrals) and fragmenta- nism essentially provides a mental
In the past 30 years mass spectro- tion of ions. ESMS, in contrast, is the map that allows planning of experi-

972 A ANALYTICAL CHEMISTRY, VOL. 65, NO. 22, NOVEMBER 15, 1993 0003- 2700/93/0365-972A/$04.00/0
0 1993 American Chemical Society
 0
-* 0
0

0 .

0
I

0 '
r4
0
- e
0

o0
a0 .
0 0
0 0
0
O 0
ments and making rational choices position to use the full range of their some of which are not well under-
of experimental parameters leading background and imagination to de- stood. Therefore these effects will not
to a desired result. Unfortunately, velop new applications of the ESMS be considered here.
the understanding of the ES mecha- technique.
nism is not yet complete. Neverthe- ES existed long before its applica- Production of charged droplets at
less, t h e whys a n d why nots for tion to MS. I t is a method of consid- the ES capillary tip
many experimental parameters are erable importance for t h e electro- As shown in the schematic represen-
understood. What is still uncertain is s t a t i c dispersion of liquids. The tation of the ES events in air at at-
exactly how the gas-phase ions are interesting history and notable re - mospheric pressure (Figure l), a
produced from the very small and search advances are very well de- voltage V, of 2-3 kV is applied to the
highly charged droplets. Readers of scribed in Bailey's book, Electrostatic metal capillary, which is typically
this REPORT will learn about current Spraying of Liquids (9).This research 0.2 mm 0.d. and 0.1 mm i.d. and lo-
ideas but will not find certainty. also provides the basis for the ESMS cated 1-3 cm from the large planar
However, much useful and interest - mechanistic studies. The new aspect counter electrode. I n ESMS this
ing information is obtained about the in the mass spectrometric applica- counter electrode has an orifice lead-
conditions existing in ESMS when tion concerns the production of gas - ing to the mass spectrometric sam-
one attempts to pin down the mecha- phase ions from the ions in the solu- pling system. Because the capillary
nism involved. Fortunately, for prac- tion; this was of no interest to the tip is very narrow, the electric field E
tical purposes, the predictions of the spray researchers (9). in the air at the capillary tip is very
two presently favored mechanisms high (E = lo6 V/m).
turn out to be quite similar. The ES mechanism: Features of When the capillary of radius r, is
ESMS has a great advantage over importance to ESMS located at a distance d from the pla-
the other methods because the ionic There a r e four major processes in nar counter electrode, the magnitude
solutions used for transfer of ions to ESMS: t h e production of charged of E, for a given potential V , is given
the gas phase are the same solutions droplets from electrolyte dissolved in by (10,11)
used in conventional wet chemistry. a solvent; shrinkage of charged drop- Ec = 2 VChc1n(4dlrc)
No unusual liquid matrices such as lets by solvent evaporation and re-
the glycerol used in FAB and matrix- peated droplet disintegrations (fis - Equation 2 provides the field at the
assisted laser desorption ionization sions), leading ultimately to very capillary tip in the absence of solu-
are required. Now that good ESMS small, highly charged droplets capa- tion. The field E, is proportional to
interfaces are available on commer - ble of producing gas-phase ions; the V,, and the most important geometry
cia1 mass spectrometers, the users mechanism of gas -phase ion produc - parameter is r,. E, is essentially in-
and not the interface developers are tion; and secondary processes, by versely proportional t o r,; it d e -
the important players. These users which gas - phase ions are modified in creases slowly, with a logarithmic
a r e typically workers i n solution the atmospheric and the ion sam- dependence, with the distance d. For
chemistry. Once these individuals pling regions of the spectrometer. Al- example, when V, = 2000 V, rc =
have acquired some understanding though this last topic is important, it 0.2 mm, a n d d = 2 cm, E, = 3.3 x
of the mechanism, they will be in a includes a wide area of phenomena, lo6 V/m.

ANALYTICAL CHEMISTRY, VOL. 65, NO. 22, NOVEMBER 15, 1993 9 973 A
REPOR7

The typical solution used in ESMS

consists of a dipolar solvent in which
electrolytes a r e at least somewhat
soluble. We use methanol as the sol-
vent and NaCl or BHCl (where B is
an organic nitrogen base) as the elec-
trolytes. Essentially, these electro-
lytes are totally dissociated into Na'
and C1- as well as BH' and C1- in
the solvent. Low electrolyte concen-
trations, routinely in the
M range, are required for the opera-
tion of E$.
The imposed field E will also par-
tially penetrate the liquid at the cap-
illary tip. To simplify the discussion,
we will assume that all experiments
are carried out in the positive ion
mode, which has a positive capillary
potential. When the capillary is the
positive electrode, some positive ions
in the liquid will drift toward the liq-
uid surface and some negative ions
drift away from it until the imposed
field inside the liquid is essentially
removed by this charge redistribu-
tion. However, the accumulated pos- Figure 1. Schematic representation of processes in ESMS.
itive charge at the surface leads t o
The very high electric field imposed by the power supply causes an enrichment of positive electrolyte ions
destabilization of the surface because at the meniscus of the solution at the metal capillary tip. This net charge is pulled downfield, expanding
the positive ions are drawn down- the meniscus into a cone that emits a fine mist of positively charged droplets. Solvent evaporation
field but cannot escape from the liq- reduces the volume of the droplets at constant charge, leading to fission of the droplets. Charge balance
uid. The surface is drawn out down- is attained in the ES device by electrochemical oxidation at the positive electrode and reduction at the
negative electrode.
field such t h a t a liquid cone forms.
This is called a Taylor cone, after the
researcher who was one of the first
to investigate the conditions under observed in the spectra were always lower in the purified solvent, and the
which a stable liquid cone can exist the positive and negative ions of the signal fluctuated widely on a time
with the competing forces of an elec- electrolytes present in the solution. scale of seconds (see Figure 3 in Ref-
tric field and the surface tension of Extraneous ions were observed only erence 14).
the liquid (12). at high capillary voltages where elec- If the charge separation is electro-
At a sufficiently high field E, the tric (corona) discharges were occur- phoretic, at a steady state the posi-
cone is not stable and a liquid fila- ring a t the capillary (15, 16). tive droplet emission will continu-
ment with a diameter of a few mi- The ion separation mechanism, o u s l y c a r r y off p o s i t i v e i o n s .
crometers, whose surface is enriched which is called the electrophoretic Considering t h e requirements for
on positive ions, is emitted from the mechanism, is also most plausible on charge balance in such a continuous
Taylor cone tip. At some distance energetic grounds. Electrical double electric current device, and that only
downstream, the liquid filament be- layers a r e already formed at low electrons can flow through the metal
comes unstable and forms separate fields in electrolyte solutions. The wire supplying the electric potential
droplets. The droplets' surfaces are resulting positive-negative ion re- t o t h e electrodes (Figure l ) , one
enriched with positive ions for which distribution reduces or completely comes to the conclusion that the ES
there are no negative counterions in removes the imposed field and there- process should involve a n electro -
t h e droplet (i.e., t h e droplets a r e fore suppresses other forms of ion- chemical conversion of ions to elec-
charged with a n excess of positive ization such as ionization by electron trons. In other words, the ES device
electrolyte ions). The length of the removal from molecules (field ioniza - can be viewed a s a special type of
unbroken liquid filament decreases if tion), which requires very high elec- electrolytic cell in which that part of
t h e field E is increased. At higher tric fields. t h e ion t r a n s p o r t does not occur
fields, a multispray condition is Another way to show t h a t ES re- through uninterrupted solution, but
reached in which the central cone sults from electrophoretic charging is as charged droplets and later as ions
disappears and droplet emission oc- to deionize the solvent. Experiments in the gas phase. Therefore, a con-
curs from a crown of four to six short involving methanol deionized by dis - ventional electrochemical oxidation
liquid tips formed a t the rim of the tillation, which reduced the conduc- reaction should be occurring at the
capillary (13). tivity from 10-6Q-1cm-1 (reagent - liquid-metal interface of the capil-
We came to the conclusion that the grade methanol, - M in lary. This reaction should be supply-
ion separation described above is the i m p u r i t y electrolyte) to io-'^-^ ing positive ions to the solution by
mechanism responsible for droplet cm-', showed t h a t the ES current converting metal atoms to positive
charging, largely on t h e basis of decreased and became intermittent. ions and electrons or by converting
mass spectra observed with electro- The intensity of the observed impu- negative ions from the solution to
lytes in the 10-5-10-3 M range (13, rity ions (NH',, Na') present in re- neutral molecules and a n electron.
14). The positive and negative ions agent -grade methanol became much This conversion process is accom-

974 A ANALYTICAL CHEMISTRY, VOL. 65, NO. 22, NOVEMBER 15,1993

 trons (although less efficiently than
SF,) and has a high electrical break-
down potential relative to gases such
as N,, Ar, Ne, and He. ES in air or
other gases below atmospheric pres-
sure will also be adversely affected
by electrical discharge because the
electrical breakdown potential de-
creases (i.e., a discharge starts a t a
lower voltage when the pressure is
decreased).
plished by oxidation reactions such onset of charged-droplet emission The capillary current I (Figure l ) ,
as (i.e., t h e onset of ES). When h i s which results from charged droplets
M (s)+ M2'(aq) + 2e- equation is combined with Equation leaving the capillary, is of interest
2, which relates the field E, to the because i t provides a quantitative
4 OH-(aq) + O,(g) + 2H20 + 4e- (3) potential V, one obtains measure of the excess positive elec-
assuming that an aqueous solution is trolyte ions produced by the spray.
electrosprayed. The actual lowest ox- VOn= 2 x 105(yrc)y21n(4d/rc) (4) This current is easily measured and
idation potential reaction expected to where y is the surface tension of the is generally i n t h e 1 x A-
occur will depend entirely on the sol- solvent, r, is the capillary radius, 10 x A range. Higher currents
vent and composition of the solution and d is the distance between capil- generally result from an electric dis-
used. Proof for the occurrence of elec- lary tip and counter electrode (see charge (15,16).
trochemical oxidation a t the metal Equation 2). A theoretical derivation giving the
capillary was provided by Blades et The onset voltages V,, in Table I dependence of I on experimental pa-
al. (17). When a Zn capillary tip was were evaluated with Equation 4 and rameters has been proposed by Hen-
used, the release of Zn2+to the solu- illustrate the range of V,, values re- dricks (11).However, the derivation
tion could be detected with ESMS. quired by different solvents for d = 4 is based on some unproven assump-
Furthermore, the amount of Zn2+re- cm and r, = 0.1 mm. The surface of tions. The equations obtained are
leased to the solution corresponded the solvent with the highest surface
to the amount required to carry the t e n s i o n ( H 2 0 ) i s t h e h a r d e s t to I =AHV;E'o" (54
ES current. Similar results were ob- stretch out into a Taylor cone and a I = AHV;E'(hLc)" (5b)
served with stainless steel capillaries liquid filament; this leads to the re-
that release Fe2+to the solution. quirement for the highest Von. where A, is a constant that can be
The ions added to the solution by Experimental verification of Equa - evaluated (11)and depends on the
the electrolytic process are at a very tion 4 has been provided by Smith dielectric constant and surface ten-
low concentration, - 2 x lo-, M (17). (20)and by work from our laboratory sion of the solvent. Equation 5 brings
They can be detected by MS, b u t (15,16). For stable ES operation one out the parameters whose functional
they do not, in general, interfere needs to go a few h u n d r e d volts dependence has been subjected to ex-
with the detection of other ionic ana- higher than Van. Using water as the perimental tests (11,13).These are
lytes present in the solution. The solvent can lead to the initiation of the flow rate (volume/time) V,, the
work reported by Van Berkel et al. an electric discharge from the capil- electric field E a t the cone tip, and
(18)shows some very interesting re- lary tip, particularly when the capil- t h e conductivity of the solution 0 .
sults when dry nonprotic solvents lary is negative (i.e., in the negative Equation 5b expresses the conduc-
are used. ion mode 115,161).The occurrence of tivity CT in terms of the relationship
There is little explicit discussion in discharge is easily recognized in
the pre - MS ES literature concerning ESMS because it leads t o t h e a p - 0 =a:c (6)
the nature of the charge carriers in pearance of discharge -produced ions. which holds a t the low concentra-
the droplets. Although this is a vital Thus, in the positive ion mode, the tions C of electrolytes prevailing in
question to the mass spectrometrist, appearance of protonated solvent ESMS. The equivalent mglar conduc-
it is of limited interest in other appli- clusters such a s H30+(H20), from tivity of the electrolyte h , gives the
cations of ES. Pfeifer and Hendricks water and CH,OH; (CH,OH), from dependence of t h e conductivity on
(11)were probably the first authors methanol is typical (15).These ions the specific nature of the electrolyte
who explicitly proposed and dis- are abundantly produced by ES in ions.
cussed the electrophoretic mecha- the absence of discharge only when The exponents v = 0.57; E = 0.43,
nism. More recently, Hayati, Bailey, the solvent is acidified. The presence and n = 0.43associated with the flow
and Tadros (19)also explicitly en- of discharge degrades t h e perfor- rate, electric field, and the conduc-
dorsed the electrophoretic charging mance of ESMS. ES with pure water tivity are predicted by the Hendricks
mechanism in a comprehensive ex- solvent is still possible when traces e q u a t i o n ( 1I). E x p e r i m e n t a l l y
amination of features of the ES pro- of discharge - suppressing gases such the following values are found: v =
cess. However, n e i t h e r research as SF, or larger quantities of 0, are 0.5, E = 0.5, and n = 0.2-0.3 (11, 14,
group considered the electrochemical added to the air surrounding the ES 21). Because of the small exponents,
nature of the process at the metal- capillary (6, 15-1 7). the changes of I with flow rate, elec-
liquid interface. The high potentials required (Ta- tric field, and concentration of the
In a treatment of the cone instabil- ble I) also illustrate that air a t atmo- electrolyte are small. Experimentally
ity and charged-droplet emission, spheric pressure is not only a conve- determined changes of I with the
Smith (20), who also assumed the n i e n t b u t also a very s u i t a b l e flow rate V, are shown in Figure 2a.
electrophoretic mechanism of charg- ambient g a s for ES, particularly Willoughby e t al. have recently
ing, provided a very useful equation when solvents such as water are to presented (22)a quantitative theory
for the electric field required for the be used. The 0, in air captures elec- based on an analysis of the conver-

ANALYTICAL CHEMISTRY, VOL. 65, NO. 22, NOVEMBER 15,1993 975 A

REPOR7

sion of energy from electrostatic po- Here we consider briefly the effect of small radius is desirable, increasing
tential energy to surface and kinetic flow rate Vf and the concentration C. the current I by increasing the vol-
energy of the droplets that occurs at As shown below, the efficient conver- ume flow rate Vf (see Equation 4 and
t h e liquid cone t i p a n d droplet sion of droplet charge to gas-phase Figure 2) is counterproductive; Vf
stream. Although details were not ions requires the production of very also increases the radius of the drop-
available, the approach appears very small and highly charged droplets. lets, as shown in Equation 7. Low
promising and may provide a reliable Gomez and Tang (23,24) have stud- flow rates in the range of 1-10 mL/
and useful relationship between the ied ES- produced droplets with phase min, which lead to small droplets
import ant parameters involved. Doppler anemometry and flash shad- (23,24), also lead to optimum ESMS
Comparison of the current I with owgraph techniques and were able to conditions. This is evident by t h e
the mass spectrometrically detected deduce the droplet size and charge broad maximum of the analyte BH'
intensity I(BH+)of the analyte (Fig- from measured droplet velocities. On intensity in the flow range 1-10 mL/
ures 2a and 2b) shows that the cur- the basis of such findings, Fernandez min in Figure 2b.
rent I and the gas-phase ion current de la Mora (25)has proposed a rela- The lack of correlation between
I(BH+) are not closely related. The tionship for the radius R of the ES- the capillary current I and the ion
absence of a close correlation be- produced droplets from sprays intensity I(BH+) demonstrated in
tween these two quantities under- Figure 2 represents a case in which
scores t h e need t o understand the the variable V, leads to a change off;
process by which gas-phase ions are the fraction of droplet charge that is
produced from the charged droplets. where p is the density, Vf is the vol- converted to gas-phase charge (i.e.,
A more detailed general examination ume flow rate, and y is the surface gas-phase ions) and a lack of corre-
of this process will be given later. tension of t h e solvent. Because a lation are expected. Variables t h a t
do not change f appreciably lead to
good correlations between the capil-
lary current and the observed total
gas-phase ion current. As we will
show in the final section of this arti-
cle, the electrolyte concentration is
such a variable, but only over a lim-
ited range of concentrations.
Shrinkage of charged ES droplets
The initial charge and size of drop-
lets produced in E S depend on the
spray conditions used, as explained
in the preceding section. Here we
will consider droplets produced a t
low flow rates, typically - 5 pL/min
at total electrolyte concentrations
not exceeding M. These condi-
tions are probably the best for high
gas-phase ion yields and high ESMS
sensitivity. Furthermore, work by
Gomez and Tang (23, 24) and Davis
and co-workers (26) h a s provided
good d a t a for droplets obtained a t
these conditions. The droplets are
small and have a narrow distribu-
tion of sizes, so they can be consid-
ered monodisperse. The size distri-
bution peaks a t radius Ro = 1.5 pm,
and t h e droplets have a charge of
Qo = C, which corresponds to
N = 50,000 singly charged ions (23,
24).
The Rayleigh equation (271, which
gives t h e condition i n which t h e
charge Q becomes just sufficient to
overcome the surface tension y that
holds the droplet together, is

Q: = 1.25x lO-loR; (8b)

Figure 2. Capillary current and analyte ion intensity dependence on solution flow rate. where eo is the permittivity of vac-
uum and y is t h e surface tension.
(a) Dependence of capillary current I on solvent flow rate through the capillary. Solid line is cocaine HCI
at M in methanol; dashed line gives slope predicted by Equation 5. (b) Dependence of mass- Equation 8b gives the condition for
analyzed ion intensity (countds) of protonated cocaine = BH' on flow rate. Trends are different for current methanol, whose y = 0.0226 N/m2;
land intensity of BH' with flow rate. (Reprinted from Reference 14.) the numerical factor is for Q in cou-

976 A ANALYTICAL CHEMISTRY, VOL. 65, NO. 22,NOVEMBER 15,1993

lombs and R in meters.
The initial charge Q , = C
observed by Gomez and Tang corre-
sponds to only 50% of the Rayleigh
limit charge QR for R, = 1.5 mm. Go-
mez and Tang (24), who also studied
the charge-to-volume ratio of larger
ES-produced droplets, found that Q0
comes closer to the Rayleigh limit for
larger droplets.
Droplets within t h e micrometer
range or larger are known to main-
tain their charge; they do not emit
gas-phase ions (24, 26). The droplets
U.UY - -
s h r i n k by evaporation of solvent
molecules until they come close to
the Rayleigh limit where they be-
come unstable and undergo fission
into smaller droplets (Figure 3).
Recent work has shown that drop-
lets with sizes in the l-pm range fis-
sion somewhat before, at - 80% of
the Rayleigh limit (23, 24, 26). The
droplet will shrink by solvent evapo -
ration at constant Q, until the ra- Figure 3. Schematic representation of time history of parent and offspring droplets.
dius R meets t h a t condition, a n d
Droplet at top left is a typical parent droplet created near the ES capillary tip at low flow rates.
then a fission will occur. Another Evaporation of solvent at constant charge leads to uneven fission. The numbers beside the droplets give
very important observation (24, 26) radius R (pm) and number of elementary charges N on droplet; A t corresponds to the time required for
concerns the type of fission that pre- evaporative droplet shrinkage to size where fission occurs. Only the first three successive fissions of a
dominates. The droplet does not split parent droplet are shown. At bottom right, the uneven fission of an offspring droplet to produce offspring
droplets is shown. The time scale is based on Equation 9b for methanol evaporation, which produces only
evenly into two smaller droplets of a rough estimate. Inset: Tracing of photograph by Gomez and Tang of droplet undergoing “uneven”
approximately equal m a s s a n d fission (24). Typical droplet loses 2% of its mass, producing some 20 smaller droplets that carry off 15%
charge. Typically, the droplets are of the parent charge. (Inset reprinted with permission of Gomez and Tang.)
observed to vibrate alternately from
oblate to prolate shapes. These elas-
tic vibrations stimulate disruptions such as methanol, acetonitrile, and mates. Results obtained with
in which the “parent” droplet emits a water are involved and the droplets Equation 26 from Reference 28 are
tail of much smaller offspring drop - are a few micrometers in diameter or shown in Figure 3, which gives the
lets. An illustration of such a fission smaller, the evaporation rate follows time dependence of Q and R for a
is shown in the inset i n Figure 3. t h e surface evaporation limit law methanol droplet of initial R, =
This disruption pattern is similar to (29), which leads to a simple depen- 1.5 pm and Qo= 8 x C, in ambi-
the disruption at the tip of the Tay- dence (26, 27) of the droplet radius ent air of - 35 “C. The time required
lor cone. The emitted stream of off- on time t for the first fission is - 400 ps. The
spring droplets carries off only about change of R due to the fission is very
2% of the mass of the parent droplet small and barely perceptible in Fig-
but about 15%of the parent’s charge. ure 3. Subsequent fissions that occur
The radius of the offspring droplets, when t h e p a r e n t droplet again
which a r e quite monodisperse, is R = Ro - 1.2 x t
(9b) reaches 80% of the Rayleigh limit are
roughly one-tenth of the radius of where ij is the average molecular ve- included in Figure 3. These fissions
the parent (22,24).A simple calcula- locity of the solvent gas, Po is the va- require shorter times, - 60 ps, which
tion (28) shows t h a t a 2% loss of por pressure of the solvent at the decrease progressively from fission
mass leads to - 20 offspring droplets temperature of the droplet, M is the to fission. A typical offspring droplet
of this radius. The offspring are not molar mass of the solvent molecules, has a r a d i u s R - 0.08 pm a n d a
only much smaller than the parent p is the density of the solvent, R, is n u m b e r of e l e m e n t a r y charges
but also have a much higher charge- the gas constant, and T i s the tem- N = 280 at birth. It reaches 80% of
to-mass ratio. perature of the droplet. The droplet the Rayleigh limit in - 40 ps, where
I n the subsequent discussion we temperature will be about 10 “C its charge is the same and its radius
will call a fission in which such a jet lower than the temperature of the has shrunk to R 0.03 pm. It is the
of small offspring droplets is emitted ambient gas (air) due t o cooling of offspring droplets that are expected
“uneven” fission; “even” fission will the droplet by evaporation (28).The to ultimately lead to gas-phase ions.
mean the formation of two droplets condensation coefficient is a,and its The time scale for evolution to gas-
of similar size and charge. value is - 0.04 for both water and phase ion-producing droplets indi-
The time required for the parent ethanol (28, 29). cated by the results in Figure 3 is in
droplet to reach t h e size R , t h a t Equation 9b was obtained by sub- the hundreds of microseconds, not
leads to the first fission can be esti- stituting the corresponding parame - very much shorter than the total res-
mated (28) with use of expressions ters for methanol and assuming that idence time of the charged droplets
providing the rate of solvent evapo- a = 0.04 for methanol also. The pre- in the atmospheric region. The exact
r a t i o n from small droplets (29). dictions of Equation 9b are only in- residence time depends on the type
When relatively volatile solvents tended to provide “ballpark” esti- of ES interface used. It is probably in

ANALYTICAL CHEMISTRY, VOL. 65,NO. 22, NOVEMBER 15, 1993 977 A

REPORT

the one to a few milliseconds range method for evaluating fractions F, solvent evaporation is required. Un-
for typical interfaces that one would see Reference 26. An initial concen- even fission separates the droplets
use in ESMS (13). tration of 5 x lov3 M would lead to a into two classes. The small offspring
The direct dependence of the rate of final bulk concentration C ==: 20 M. are enriched in charge and are the
evaporation on the vapor pressure of The formation of droplets with such future gas-phase ion emitters; the
the solvent (Equation 9a) indicates a high concentration by solvent large parent droplets become t h e
that solvents of relatively low vapor evaporation at near ambient temper- dumps for t h e unwanted charge-
pressure will lead to acceptable ESMS a t u r e is very unlikely because a paired electrolyte ions.
sensitivity only at elevated tempera- slowing down of evaporation will oc-
cur at high concentrations. Further- Nature of the process leading to
tures. The vapor pressure for metha- formation of gas-phase ions
nol used to obtain Equation 9b is more, even if evaporation did occur,
- 100 torr, which is a usefbl indicator f u r t h e r shrinkage of t h e droplet Two different mechanisms have been
for the required vapor pressure. would lead to charged solid particles proposed to account for the forma-
Unfortunately, because of their from which gas - phase ion emission tion of gas-phase ions from the small
small size, the type of fission t h a t would not be expected. charged droplets. The first one de-
offspring droplets undergo has not An analysis of the solute residue pends on the formation of extremely
been observed. The type of fission, problem described above, which small droplets, R = 1nm, which con-
even or uneven, is of great interest leads t o similar conclusions, was tain only one ion. Solvent evapora-
because it determines the ratio of first given by Iribarne. See Figure 6 tion from such a droplet will lead to
charges (excess positive ions near in Reference 30a for additional infor- conversion of the droplet to a gas-
the surface) N,relative to n, which is mation. phase ion. Such a mechanism was
the number of charge-paired ions in On the other hand, as illustrated assumed by Dole (31),the first inves-
the solution inside the droplet. For i n Figure 3, uneven fission t h a t tigator of gas-phase ion production
the case in which even fission occurs by ES. A more detailed consideration
(i.e., a droplet separates into two and support of this mechanism was
droplets of near equal charge and given by Rollgen (32).We shall call
mass), there will be no increase in this model the single ion in droplet
Nln, which corresponds to the ratio of theory (SIDT). This early work did
the number of excess positive ions not provide any details on how such
near the surface to the number of ion droplets should be formed or whether
pairs in the bulk. the process should include selectivity
On the other hand, uneven fission that may favor the formation of gas-
such as that discussed above involv- phase ions A+ relative to B’ as a re-
ing 15%of the charge but only 2% of sult of specific physicochemical dif-
the mass (or volume) in the offspring ferences between these two species of
droplets leads to a n increase of Nln ions.
by a factor of 7. This number is ar- The other mechanism, proposed by
rived at by assuming that when the Iribarne and Thomson (301,assumes
Offspring! droplets a r e formed the ion evaporation (emission) from very
surface charge “slips over” the drop- small and highly charged droplets.
let bulk solution so t h a t the bulk Typically, the droplets from which
electrolyte concentration in the off- ion emission becomes competitive
spring droplet remains about t h e with Rayleigh fission have a radius
same as that in the parent drop (28). of R = 8 nm and N ==: 70 elementary
If uneven fission also occurs for the charges (30).Under these conditions,
offspring droplets, Nln for the second the droplet does not undergo fission
generation offspring is further in- but emits gas-phase ions. As N de-
creased by a factor of - 7, leading to forms a first offspring followed by creases, emission can still be main-
a total factor of - 49. uneven fission of that offspring leads tained as a result of a decrease of R
The increase of Nln by uneven fis- to a droplet of R = 3 nm with N = 2 by solvent evaporation. Thus, t h e
sion is probably an essential process charges. The calculated concentra- Iribarne mechanism does not require
t h a t makes the production of gas- tion increase factor for these droplets the production of very small droplets
phase ions, and thus ESMS, possi- is only F = 140 or 30 times smaller ( R ==: 1 nm) that contain only one ion.
ble. If all droplet fission were even, a than for 10-nm droplets obtained Iribarne emission can occur even
very large electrolyte concentration with even fission. Assuming an ini- when the droplet contains other so-
increase in the droplets could be ex- tial electrolyte concentration of io-* lutes such as charge-paired electro-
pected to occur before the droplets M, one calculates t h e number of lytes.
reached the small radius R < 10 nm charge-paired electrolyte ions n ==: 1 This last difference can be used as
below which gas - phase ion formation using t h e known volume a n d F = the basis of experimental tests. In
is expected to occur. Thus, for the 140. The evaporation of solvent from experiments in which a single solute
case of droplets t h a t measure R = droplets of this type will lead directly such as NaCl is used, one might
1.5 pm and Q = 8 x C consid- to gas - phase ions! This observation expect, particularly a t higher con-
ered in Figure 3, a concentration in- will be elaborated on in the next sec- centrations, t h a t some of the very
crease by a factor F = 4000 due to tion, which deals with the mecha- small SIDT droplets would also con-
the required solvent evaporation can nism of gas-phase ion formation. tain some charge-paired NaCl spe-
be calculated for the even fission I t is useful to restate some of the cies. Solvent evaporation from such
process, leading to droplets that preceding observations. Even fission droplets should t h e n lead to
measure R = 10 nm. For more infor- treats all droplets equally and, be- Na+ (NaCl), ions if SIDT holds. Mass
mation and for a description of the cause charge is not enriched, much spectra obtained with NaCl solutions

978 A ANALYTICAL CHEMISTRY, VOL. 65, NO. 22, NOVEMBER 15,1993

showed the presence of Na' but no cannot be explained by the SIDT. We based on transition state theory. The
Na' (NaCl), were detected. Even will show in the next section of this rate constant k , for emission of ions
ions where n = 1 were not observed article that an expanded SIDT is also from droplets is
for solutions with concentrations of capable of qualitatively predicting kT
NaCl up to lo-" M (14). experimentally observed selectivi- k, = - exp(-AG'/RT) (10)
The absence of Na+ (NaCl), ions ties. To understand the significance h
was t a k e n as decisive evidence of the evidence, one needs to have where k is the Boltzmann constant, T
against the SIDT. However, at the additional information about t h e is the temperature of the droplet,
time, even fission was assumed to Iribarne model and the nature of the and h is the Planck constant. The
occur (14), and even fission leads to experimental evidence. free energy of activation, AG', was
high solute concentrations in the off- The Iribarne t r e a t m e n t (30) i s evaluated on the basis of the model
spring droplets. As shown in the pre-
vious section, uneven fission dra-
matically increases the ratio Nln of
charges N to paired ions. We also
showed (Figure 3) that a second-
generation offspring droplet can
have a radius of - 3 nm, carry N = 2
charges, and for an initial electrolyte
concentration of [NaCl] = M
have only one Na' C1- ion pair
(n = 1) inside the droplet. At [NaCl]
= lo-" M, one expects N = 3 and
n = 10.
Further evaporation and uneven
fission may lead to a third genera-
tion of offspring droplets, and some
of these may contain only one Na'
and no NaC1. Therefore, mass spec-
tra dominated by the Na' are a pos-
sibility.
The previous discussion shows
that uneven fission provides a mech-
anism for single ions in droplet for-
mation. Although the observed com-
plete absence of Na' (NaCl), ions in
t h e m a s s spectra ( 1 4 ) is not e x -
plained, the difference between ex-
perimental observation and predic-
tion is much less drastic. I n other
words, the realization that uneven
fission is dominant has provided
s t r o n g s u p p o r t for t h e SIDT.
Willoughby and co-workers have
presented a theoretical model (22)
that leads to similar predictions.
Details of the lribarne ion
evaporation theory
Iribarne and Thomson (30) based
their theory on a derived equation
that provides detailed predictions for
t h e r a t e of ion emission from t h e
charged droplets. I n particular, it
predicts the dependence of the rates
on the chemical properties of t h e
ions. Observed differences in t h e
gas-phase ion intensities IA and IB of
ions A+ and B' present at equal con-
centrations in the sprayed solutions
were compared with predictions of Figure 4. The "ion evaporation" transition state and its dependence on the ion
the theory by Iribarne and Thomson solvation energy.
(30b) a n d subsequently by other (a) Transition state for ion evaporation (emission) proposed by lribarne and Thomson. In the initial state,
workers as well ( 4 , 14, 33, 34). I n a droplet of radius R has N ionic charges. Typical values are R = 8 nm, N = 70. Distance of ions from
general, qualitative agreement be- surface d is due to solvation shells of ions; see enlarged inset. In the transition state, the escaping
tween experiment and theory was ion-solvent molecule cluster M+(SI), is shown at distance x, outside the droplet; x, is typically 0.6 nm.
(b) Free energy of escaping ion cluster M+(SI), as a function of distance x obtained with the lribarne
obtained, lending support to the the- equation (30).Maximum of A@ corresponds to the activation energy A@. Values of rate constants k,
ory on the assumption that the ex- given in s-' units at T = 298 K droplet temperature. Individual curves are for different solvation energies
perimentally observed selectivity of the ion solvent molecule cluster M+(SI)m

ANALYTICAL CHEMISTRY, VOL. 65, NO. 22, NOVEMBER 15, 1993 979 A
REPORT

shown in Figure 4a. The transition

state selected by the authors resem-
bles the products rather than the ini-
tial state (Le., it is a "late" transition
state). The advantage of this choice
is that the energy of such a state can
be expressed with a closed equation
based on classical electrostatics and
thermodynamics. However, the tran-
sition state could be occurring ear-
lier, for example, as the ion disrupts
the droplet surface. The energy of
such an early transition state would ,
be much more difficult to evaluate. If
a higher free energy barrier did oc- 1.25 1.5 1
cur at that earlier stage, the predic-
tions of the Iribarne model would be
invalidated.
The barrier in the Iribarne transi-
tion state is due to the opposing elec-
trostatic forces: the repulsion of the
escaping ion by the other charges of
the droplet and the attraction be-
tween the escaping ion and the drop-
let because of the polarizability of
the solvent medium of the droplet.
The attraction is larger at short dis-
tances between the ion and the drop-
let surface, but it falls off faster than
the repulsion as the distance is in-
creased.
The equation for AG* was found
(30) to depend on four parameters.
The first two are N , the number of
charges on the droplet, and R, the ra-
dius of the droplet. The rate constant
k , increases with N and decreases
with R. The other two parameters
express the specific properties of the
ions involved. The escaping ion in
Figure 4a is not the naked electro-
lyte ion M' but an ion-solvent mole-
cule cluster M'(Sl), containing m
solvent molecules. AG' is lowest not
for the naked ion M', but when the
ion takes several solvent molecules
with it. Recall that the transfer of a
naked ion such as Na' from aqueous
solution to the gas phase (Equation
1) requires a large amount of energy
(- 98 kcal/mol). On the other hand,
the transfer of Na'(H,O), requires
only - 56 kcal/mol. The transfer free
energies from gas phase to solution
for the naked alkali ions M' and the
clusters M' (H,O), are given in Ta- Figure 5. I ypical dependence of capillary current I and analyte A' ion intensity in
ble 11. These data are based on ex- experiments in which a single analyte is added to the solvent.
perimental thermochemical mea- A constant concentration of impurity M Na') present in reagent-grade methanol provides the
surements available in the literature dominant electrolyte at low analyte concentrations. Results shown are based on morphine H+ as A+.
( I , 2) and were evaluated in Refer- (Reprinted from References 21 and 28.)
ence 28. The strongly solvated ions
Li' and Na' have large transfer en-
ergies (see Table II), and for these for different values of the ion cluster t h e droplet surface, where t h e y
ions the Iribarne equation predicts solvation energy. would minimize their repulsion, be-
large activation barriers (i.e., low The second parameter t h a t ex- cause that would partly disrupt their
rate constants k , ) . The change of k , presses the individuality of the ions solvation shell. Strongly solvated
with the value of the ion cluster sol- is d , the distance of the ion charges ions such as Li' hold on strongly to a
vation energy is illustrated in Figure from the surface of the droplet (see larger number of solvent molecules
4b, which depicts evaluated barriers Figure 4a). The charges cannot be on and have a larger d. Radii obtained

980 A ANALYTICAL CHEMISTRY, VOL. 65, NO. 22, NOVEMBER 15,1993

from ion mobilities in solution tensities at different concentrations The key to understanding the com-
(Stokes radii) provide suitable values of analytes in t h e solution t h a t is plete analyte curve is to realize that
for d (28).Such values are given in electrosprayed are shown in Figures the solution involved is not a single
Table 11. 5-7. The intensity of a single ana- electrolyte system because of the sol-
Rate constants kI evaluated for a lyte observed in mass spectra ob- vent that is used. Unless special
droplet with R = 8 nm with N = 70 tained with a series of solutions with deionization procedures are imple -
charges at a temperature T of 298 K increasing analyte concentration is mented, impurity electrolytes are al-
for the alkali ions are given in Table shown in Figure 5. The logarithmic ways present. The impurities in the
11. These d a t a illustrate that t h e plot u s e d accommodates a wide reagent-grade methanol used (Fig-
t r a n s f e r e n e r g y of t h e c l u s t e r range of ion intensities and concen- ures 5-7) are primarily ammonium
M+(H,O), g i v e n b y -AGosol trations. The analyte plot has a typi- and sodium salts at a total concen-
[M’(H,O),] in Table I1 is the deci- cal shape that is commonly observed tration of - lop5 M. For analyte con-
sive parameter for the value of kI. (14,28, 35). A linear section with a centrations below M, ES is pos-
Li’, which h a s the largest cluster slope of = 1 in the low concentration sible only because of the presence of
transfer energy, has the lowest k,. range up to - M is followed by impurity electrolyte.
The k, for Cs’ is larger by a factor of “saturation” and a small decrease of In the low concentration range of
- 5000 relative to k , for Li’. These intensity at the highest concentra- analyte A, the capillary current I is
results predict that large differences tion M). The linear section, carried by t h e dominant electro-
in the ESMS sensitivities of these where the intensity IA of the analyte lyte-the impurities B, which are at
ions are expected provided that the A is proportional to the concentra- a constant concentration. I is there-
Iribarne theory is valid. tion [A’], is a region useful for the fore constant in this range (see Fig-
Iribarne and Thomson (30b)have quantitation of A+ in the solution. ure 5a). The sum of the mass ana-
presented calculations of the relative Such linearity is generally observed lyzed total ion intensitites Itot= IA +
k , for the alkali cations. Their results (14,28, 35, 36). IB is also constant in this range be-
(Table 11) show a much s m a l l e r
spread of kI values from Li’ to Cs’
and are much closer to the experi-
mental results. However, the calcu-
lated values (28) are probably more
trustworthy because they were ob-
tained with experimental data for
AG”,,, [M’(H,O),] and d. Iribarne
and Thomson used a classical elec-
trostatic expression of questionable
validity for the solvation energy and
estimated the values for d.
Also given in Table I1 are values
for the experimental sensitivity coef-
ficients k obtained from the experi-
m e n t . T h e significance of t h e s e
values and how they can be com-
pared with predictions of the Irib-
arne theory and t h e SIDT are de-
scribed in the next section.
Dependence of ion intensities on
concentration
In this section we will examine how
the analyte ion intensity determined
with the mass spectrometer depends
on the concentration of the analyte
ion in the solution t h a t is electro-
sprayed and how this intensity is af-
fected by the presence of other elec-
trolytes. O t h e r electrolytes a r e
almost always present as impurities
in the solvent used, as other coana-
lytes, and as buffers required in re-
versed-phase LC or CE.
The question concerning the ex-
pected ion intensity for a given ana-
lyte under given concentration con-
ditions is part of the daily practice of Figure 6. Results from a series of experiments in which a given analyte A is at
ESMS. Trying to understand the fac- constant concentration, [A+] = M, and NH,CI is increased at concentrations
tors that determine the observed in- shown.
tensity leads one rapidly to ask ques- (a) Change of capillary current Iwith increase of NH4CI concentration. Essentially the same Iis observed
with any of the analytes A. (b) Change of ion intensity of mass analyzed A+ corrected for mass dependent
tions about t h e mechanism of ES transmission of A+. Only one A+ and NH4CI were present in a given series in which [NH4CI]was
discussed in previous sections. increased. Solid lines are curves predicted by Equation 12. Analytes A+ are identified above and in
Experimentally determined ion in- Table 111. Best fit ratios k,lk, used with Equation 12 are given in Table Ill. (Adapted from Reference 28.)

ANALYTICAL CHEMISTRY, VOL. 65, NO. 22, NOVEMBER 15,1993 981 A

REPOR7

cause it is dominated by the impuri- electrolyte equation. IA+ is the mass uum of the mass analyzer; values of
ties B. spectrometrically detected ion inten- p = 10-4-10-5 may be expected (28).
Above an analyte A concentration sity of ion A+ corrected for the mass I t should be noted that according
of - low5M, the analyte begins to dependent transmission of the mass to Equation 12a, the intensity IA de-
dominate and the total electrolyte analyzer. [A'] and [B'I are the elec- pends only on the ratio kA/kB and not
concentration begins to increase. An trolyte concentrations in the solution on the individual values of kA and kB.
increase of capillary current I occurs to be electrosprayed. The product f p The expression kA/kB is the yield ra-
in this region as expected from is a factor that was assumed to be tio of gas-phase ions A+ and B' rela-
Equation 5. independent of the chemical nature tive to the solution concentrations.
of t h e ions; f is t h e fraction of In earlier work such ratios were gen-
I= ((,C)" n = 0.2-0.3 (11) charges on the droplets that are con- erally obtained (2, 30, 33) by deter-
Because the molar conductivity hk verted to gas-phase ions and # is the mining the ion intensities IA and IB
for different electrolytes M' gener - ion- sampling efficiency or the frac- with solutions where [A'] = [B'].
ally changes by less than a factor of 2 tion of t h e ions detected with t h e These measurements can be consid-
over a wide range of electrolytes (37) mass spectrometer relative to t h e ered as a special case in the use of
and the exponent n is very small, the g a s - p h a s e ions produced by t h e Equation 12, which when applied to
current I is essentially independent droplets at 1 atm. IA and IB leads to
of t h e n a t u r e of t h e electrolyte. Values of f p can be determined
Changes of I with concentration C with Equation 13,
must be considered because C can be
changed over a much wider range. IA+ IB = f p I (13)
However, again as a result of the which follows from Equation 12a, by A
' kA
small exponent n, these changes are measuring the total mass analyzed - = -for[A]=[B] (14)
still relatively small (a factor of 2-4) ion current, IA + IB, and the capillary IB k*
for a change of C by a factor of 100. current I . The use of Equation 12 permits the
The shape of the Itot curve is seen to determinations of kA/kB for a wider
be very similar to that of the capil- set of concentration conditions where
lary current I (Figure 5 ) . An approx- [AI # [Bl.
imate proportionality between the The dependence of IA and IB on the
two currents is generally observed in solution concentrations [A"] and [B']
this concentration range (14,223). is expected to be indirect. Much more
For an analyte concentration above relevant would be t h e number N A
lo-' M, the intensity IB of the impu- and N B of charges due to ions A' and
rity ions B' is observed to decrease B+ at the surface of the droplets and
(Figure 5). This decrease is a conse- the number nA and HB of ions A' and
quence of t h e weak dependence of B+ that a r e paired with negative
the current I on the total concentra- counterions in the droplets. These
tion. Because the current is propor- numbers are not available. However,
tional to the total droplet charge, the Equation 12 can be used for a second
addition of A barely increases t h e purpose. The extent of agreement or
droplet charge. However, i n t h e deviation of Equation 12 with exper-
droplets, the ions A+ compete with iments in which the concentrations
the ions B' in the conversion process [A] and [Bl are changed over wide
to gas-phase ions. A proportionality limits permits deductions to be made
to concentrations of ions A+ and B' on the relationship between the con-
in the droplets may be expected in centrations [AI and [Bl and the rela-
this competition so that an increase tive numbers of ions in the droplets,
of [A+] should lead to a decrease of Recently, Smith and co - workers
production of gas-phase ions B', that (35)have shown that an equation of
is to a decrease of IB. This is exactly the same form as Equation 12 can be
what is observed in Figure 5. usefully applied to ES involving ef-
On t h e basis of such consider- fluent from CE.
ations, Equation 12a has been pro- In experiments in which concen- Experiments in which only a sin-
posed (20, 26) for a two-electrolyte trations and the nature of the elec- gle electrolyte A is added to the sol-
system; Equation 12b is for a three- trolytes were changed, the product f p vent that contains impurity B, as in
electrolyte system. was found to be approximately con- the case for the data used in Figure
stant (28)up to a total concentration 5 , can be successfully fitted by Equa-
of - 5 x M. A rough estimate of tion 13 (see Reference 20) but do not
f = 0.3 has been obtained (28) for a provide an accurate determination of
20-pL/m flow rate with methanol as kA/kB and a sensitive indication of
the solvent. A decrease of f p is gener- whether t h e value kA/kB remains
ally observed above - 5 x lop4 M constant with changes of concentra-
(Figure 5 ) . It has been attributed tion of A.
(14)to a decrease off, which is a re-
sult of an expected increase in drop- Results where two electrolytes
let size at high electrolyte concentra- are added to the solvent
The equations for IB a n d I , are tions (19).The value of p depends on Better tests are obtained when two
analogous to those for IA. For sim- the design of the interface between electrolytes, A and B, are added to
plicity we will discuss t h e two- the atmospheric region and the vac- the solvent. The presence of electro-

982 A . ANALYTICAL CHEMISTRY, VOL. 65, NO. 22, NOVEMBER 15,1993

lyte impurities in the solvent, which More extensive experiments, par- the ratio of surface charges NA/NB
will be called C, makes this a system ticularly under conditions in which above the ratio expected on the basis
of three electrolytes. t h e added electrolytes a r e a t t h e of the concentrations [AI/ [Bl. The
Results from such experiments same concentration [A"] = [B'], have rate of ion emission from an Iribarne
(28) are given in Figure 6. In these shown (28) that in general a fit with droplet is proportional to the number
experiments, the concentration of a Equations 12 a n d 14 cannot be of surface charges and will be higher
given analyte A was kept constant, obtained over the complete concen- for the more surface-active ion even
[A+] = 1 x M, and the concentra- tration range 10-8-10-3 M with a when the Iribarne rate constants are
tion of B', which is always NH",, was single value for the ratio kA/kB. The equal.
increased above 2 x M. The results i n Figure 7 illustrate t h e The observed decrease of kA/kB at
solid curves in Figure 6, which pro- three typical cases observed. low concentrations was attributed to
vided a good fit to the experimental The ratio kA/kB remains constant depletion of the ion A+ (28), which
points (28), were obtained w i t h over t h e complete concentration has a higher evaporation rate. At
Equation 12. The kA/kB values ob- range when kA = kB. This is the case high concentrations, the ions in the
tained (28)from these fits were used for K" and Cs+ (Figure 7a). Note that solution of the droplet dominate over
to acquire some of the data shown in for the logarithmic plots used, a con- the surface charges (i.e., n >> N ) .
Table 111. stant kA/kB corresponds to a constant The faster ion evaporation of A+ from
The plots obtained in Figure 6 il- vertical distance between log I A and t h e surface does not decrease t h e
lustrate a situation of practical im- log IB. The joint decrease of I A and IB surface charge ratio NA/NB because
portance, that is, how the intensity at low [A] = [B] is due to the pres- A+ ions are rapidly supplied from the
of an analyte A+ is suppressed by the ence of electrolyte impurity in the droplet bulk where nA/nB = [A]/[Bl.
presence of a second electrolyte B, solvent and is analogous to the de- At low concentrations, t h e r e a r e
which is used as a buffer. Substan- crease discussed for Figure 5. many fewer ions i n t h e bulk, al-
tial decreases of I A result. When A+ When kA/kB > 1 and approximately though the surface charge number is
is Bu,N+, the ion with the highest constant at a high concentration, a approximately the same; this leads
kA, the loss of intensity is by a factor lower kA/kB is observed at low con- to the condition N > n. Faster evapo-
of 3; this is the smallest decrease ob- centrations. kA/kB = 1 is observed for ration of A+ from the surface leads to
tained. When Cs+, which has t h e t h e tetraalkylammonium cations a relative depletion of A+ over B+ in
lowest kA, is present, intensity de- Pen4N+ and Et,N+ in the low concen- the droplet and the faster emission
creases by a factor of 12 when [NH",] tration range (Figure 7b), whereas of A+ cannot be maintained. There-
is increased from to M. for the pair and n-C,H,,NH,+ and fore, the observed kA/kB decreases at
Equation 12 predicts that the de- Cs+, a decrease to a lower and ap- low concentrations. The situation il-
crease of I A with [B+]will be greater proximately constant value at low lustrated i n Figure 7b may occur
as the value of kB increases. NH", has concentration is observed, but this when t h e difference between t h e
a low value of k , and therefore the value is > 1 (Figure 74. Iribarne rate constants for A+ and B+
decreases of IA observed in Figure 6 The factors responsible for t h e takes precedence over the difference
are not the worst case. When the ion above changes have been explored in surface activities. For the case in
with the highest coefficient, Bu,N+, (28) in some detail only from t h e Figure 7c, the reverse was assumed.
was used instead of NH",, t h e ob- standpoint of the Iribarne theory, This argument was based on the ra-
served decrease of the analyte inten- which was extended to include the tionale that the relative number of
sities for Cs" and MorH' was much effect of surface activity. The high surface charges NA/NB adjusts to the
higher, by a factor of - 200 (28). kA/kB observed at high concentra- relative surface activities already in
Therefore, in the actual practice of tions are due to A+ having a higher the parent droplets (see Figure 3),
ES, buffers that have cations with Iribarne rate constant and/or higher where t h e number of bulk ions is
low coefficients k should be used if surface activity. The effect of differ- large so that this process will be less
possible. ential surface activity is to increase affected by depletion.
Comparison of coefficients with
lribarne theory and SlDT
Comparisons of the experimental co-
efficients with the Iribarne rate con-
stants and expected surface activi-
ties should be made only in the high
concentration range, where depletion
is absent. The experimental coeffi-
cients given in Table I11 are aver-
aged results from experiments shown
in Figure 6 and experiments in
which [A] = [Bl, as in Figure 7. The
range of coefficient values in Table
I11 is not very large. From the stand-
point of ESMS this is a desired re-
sult and means that we can expect
to detect with a fair sensitivity any
analyte ion present in the solution.
The ions with the highest coeffi-
cients i n Table I11 a r e t h e large
tetraalkyl ammonium ions, which
are expected to have low solvation

ANALYTICAL CHEMISTRY, VOL. 65, NO. 22, NOVEMBER 15, 1993 983 A
REPORT

energies and thus high Iribarne rate indirect. D e t e r m i n a t i o n s of t h e cause this was not the case, the ions
constants. They also have relatively ESMS spectra of cytochrome c and must originate from solution t h a t
high surface activities (38).Actually, myoglobin in 0.2% acetic acid and in has experienced minimal evapora-
a general correlation between low 0.2% propionic acid solutions in wa- tion. The most likely source for such
solvation energies and high surface t e r showed t h a t t h e ion intensity ions is the Taylor cone (39).
activities is expected. Because of the profiles resulting from multiple pro - However, the relative vaporization
absence of suitable quantitative in- tonation were almost identical when estimates were made with the as-
formation on solvation energies and acetic or propionic acid was used sumption that liquid-vapor equilib -
surface activities in the literature, it (39). The authors argued t h a t the rium data apply to evaporation from
is difficult to establish the extent to profiles obtained with the two acids very small droplets of a volatile
which each factor contributes to the should be different if the ions origi- liquid such as water. This assump-
value of the experimental coefficient nated from droplets t h a t had been tion is not justified. The rate of va-
(Table 111). subjected to extensive evaporation, porization from such droplets is not
I t appears that ions with no sur- They estimated t h a t the hydrogen thermodynamically controlled-it is
face activity such as the alkali ions ion concentration in the acetic acid kinetically controlled (29).The so-
are best suited for comparison with solution should increase because called surface - controlled evaporation
theory. As discussed earlier, t h e evaporation increases the acetic acid holds. The composition of the vapor
agreement between the calculated kI content in water. For propionic acid, escaping under surface control from
and the experimental values for the a decrease of acidity was estimated. the very small droplets of water and
alkali ions is not very good (see Ta- With these assumptions one can ex- 0.2% acetic or propionic acid cannot
ble 11). The qualitative agreement pect that the polyprotonated ion pro- be predicted with equilibrium data,
observed between the Iribarne pre - file in the ESMS spectra should be and therefore the evidence (39) for
dictions and experiments involving shifted to higher proton numbers for the absence of evaporation is weak.
organic ions ( 4 , 14, 31, 32) might be the acetic acid solution spectra. Be- E S mass spectra obtained with
due to surface differences only, and if ethylene glycol and water as solvents
this is the case, these results need were compared (40).When the liquid
not be considered as support of the in the ES capillary was at room tem-
Iribarne theory. perature, water was found to lead to
The SIDT, as originally stated (31, normal ion intensities; those ob-
32), did not provide criteria for selec- served with the much less volatile
tivity on the basis of physicochemical ethylene glycol were 200 times lower.
properties of the ions. I t is obvious In experiments in which the spray
that the surface activity is the factor probe was at 100 "C, the intensities
that should lead to selectivity for the with ethylene glycol as the solvent
SIDT. Ions that are enriched on the were found to be 40 times lower than
surface will preferentially end up those observed with water. Higher
with the final droplet that contains a intensities from water relative to
single ion. ethylene glycol are expected on the
An analysis examining how well basis of ion production from droplets
the depletion phenomena can be ra- that have to reduce their size by
tionalized on the basis of the SIDT, evaporation. However, the authors
expanded to include surface activity, provide evidence, largely based on
has not been made yet, but some de- computer modeling, that the inten-
pletion could be expected. At present, sity differences expected between
it is not possible to state with cer- water a n d ethylene glycol would
tainty which theory, ion evaporation have been much larger had the ion
or SIDT, fits better with the avail- production depended on solvent
able evidence. However, a shift in fa- evaporation from droplets, because
vor of the SIDT has occurred. Fortu- the expected evaporation of ethylene
nately, from a practical standpoint, glycol is insignificant.
qualitative predictions for high se- Even if the arguments (40)are ac-
lectivity on the basis of low solvation cepted at face value, they do not
energy or high surface activity will prove that the ion emission for water
both be valid, because the two pa- is not from droplets that have expe-
rameters are positively correlated. rienced evaporation. They only show
that the observed very weak ion pro-
Emission of gas-phase ions from duction from glycerol may be from
the Taylor tip of the ES capillary the Taylor cone. Ion emission may in
Very recently Siu and co-workers general be occurring from the Taylor
(39, 40) have proposed that the gas- cone and from droplets. For volatile
phase ions are not formed from the Figure 7. Observed mass analyzed ion solvents, ion production from t h e
charged droplets but a r e emitted intensities /A and /B for experiments in charged droplets should dominate
from the Taylor cone tip. This is a which solution concentrations of over Taylor cone emission. Nonvola-
radical proposal that, if it turns out analytes [A+] = [B'] were increased. tile solvents such as glycerol may
to be true, would require a n exten- Typical cases discussed in text are illustrated in lead only to the weak Taylor cone
sive reinterpretation of experimental (a)-(c). A decrease of IA and le intensities emission. Ion emission from liquid
observed at [A'] = [B'] < lop5 M is a result of
observations described in the preced- competition from electrolyte impurity C' in tips in a vacuum is known to occur
ing discussion. methanol at constant concentration [C'] 2: and is the basis of electrodynamic
The evidence presented (39, 40) is M. (Adapted from Reference 28.) MS ( 4 1 ) . The ion currents obtained

984 A ANALYTICAL CHEMISTRY, VOL. 65, NO. 22, NOVEMBER 15,1993

a r e several orders of magnitude ries a second charged site. The two Wong, S. F.; Whitehouse, C. M. Science
smaller than the capillary current I charges then facilitate the escape of 1989,246, 64.
(5) Covey, T. R.; Bonner, R. F.; Shushan,
in ES. other charged parts and ultimately B. I.; Henion, J. Rapid Commun. Mass
The work of Siu and co-workers is the escape of the whole macroion. Spectrom. 1988,2, 249.
interesting and provocative; how- An alternative mechanism can be (6) Smith, R. D.; Olivares, J. A.; Nguyen,
ever, its significance will have to be proposed on the basis of the observed N. T.; Udseth, H. R. Anal. Chem. 1988,
60, 436.
established in the future. uneven fission and presence of oscil- (7) (a) Mirza, A. U.; Cohen, S. L.; Chait,
lations ( 2 4 ) i n t h e small charged B. T. Anal. Chem. 1993,65,1; (b) Winger,
Mechanisms for formation of droplets. For example, one of the off- B. E.; Light-Wahl, K. J.; Rockwood,
multiply-charged macroions spring droplets in Figure 3 had a ra- A. L.; Smith, R. D. J. A m , Chem. SOC.
1992, 114,5897.
The preceding part of this article ap- dius of 30 nm and some 280 charges. (8) Beu, S. C.; Senko, M. W.; Quinn, J. P.;
plies to conventional ions, that is, Assume that a polyprotonated pro- Wampler, F. M.; McLafferty, F. W. J.
singly- and doubly-charged inor- tein with - 30 protonated sites is Am. SOC. Mass Spectrom. 1993,4,557.
ganic or organic ions of molecular present in the droplet. Some of the (9) Bajley, A. G. Electrostattc Spraying of
Liauzds: J o h n Wilev a n d Sons: New
mass up to a few hundred Daltons. protonated sites may be near t h e Ygrk, 1988.
Although many ESMS applications surface of t h e droplet where they (10) Loeb, L. B.; Kip, A. F.; Hudson,
in chemistry and biochemistry in- would not be charge balanced by G. G.; Bennett, W. H. Phys. Rev. 1941,60,
714.
volve such ions, the ability of ES to (ll)Pfeifer, R. J.; Hendricks, C. D. AL4.A
produce multiply -charged macroions J 1968,6,496.
such as the polyprotonated peptides (12) Taylor, G. I. Proc. R. SOC.London A
and proteins is also an area of great 1964, A280,383.
interest (4-8). The mechanistic as- (13) Ikonomou, M. G.; Blades, A. T.; Ke-
barle, P. Anal. Chem. 1990, 62,957.
pects of this area deserve a separate (14) Ikonomou, M. G.; Blades, A. T.; Ke-
article. Here, we can consider briefly barle, P. Anal. Chem. 1991, 63, 1989.
only two aspects that are related to (15) Ikonomou, M. G.; Blades, A. T.; Ke-
the previous discussion. For simplic - barle, P. J Am. SOC.Mass Spectrom. 1991,
2, 497.
ity we consider only polyprotonated (16) Wampler, F. W.; Blades, A. T.; Ke-
peptides and proteins. barle, P. J. Am. SOC.Mass Spectrom. 1993,
We side with the view (5)that the 4, 289.
ESMS observed degree of protona- (17) Blades, A. T.; Ikonomou, M. G.; Ke-
barle, P. Anal. Chem. 1991, 63, 2109.
tion is related to the number of basic (18) Van Berkel, G. J.; McLuckey, S. A,;
groups that are protonated in the so- Glish, G. L. Anal. Chem. 1992, 64, 1586.
lution. However, the degree of proto- (19) Hayati, I.; Bailey, A. I.; Tadros, T. F.
nation will be modified by changes of J. Colloid Inte$ace Sci. 1987, 117,205 and
222.
pH due to solvent and acid evapora- (20) Smith, D.P.H. IEEE Truns. Ind. Appl.
tion from the charged droplets (42, 1986, LA-22, 527.
43). As pointed out in the previous (21) Tang, L.; Kebarle, P. Anal. Chem.
section, it is not easy to establish 1991, 63,. 2709.
(22) (a) Willoughby, R.; Sheehan, E.; Jar-
how the pH will change on evapora- negative counterions; thus, they can rell, A.; Masecic, T.; Pedder, R.; Penn, S.
tion from small droplets, particularly be part of the 280 surface charges. Presented at the Annual Meeting of the
when volatile acids are used. An elastic oscillation of the droplet American Society for Mass Spectrome-
The Iribarne theory for ion evapo- may expose additional protonated try, San Francisco, CA, 1993; (b) Shee-
han, E. W.; Willoughby, R. Presented at
ration from a charged droplet can be sites, precipitating a special uneven the Annual Meeting of the American
applied to multiply - charged macro - fission in which the polyprotonated Society for Mass Spectrometry, S a n
ions. A model has been suggested by species with only a small amount of Francisco, CA, 1993.
Fenn (44). In the initial state, the solvent becomes the small and highly (23) Gomez, A.; Tang, K. Proceedircgs of the
Fifth International Confirence on Liquid At-
polyprotonated macromolecule is lo - charged offspring of the uneven fis- omization and Spray Systems I (CLASS-91);
cated inside the small charged drop- sion. The distinction between this Smerjians, H. G., Ed.; NIST: Gaithers-
let. All charges of the macromolecule model and that of Fenn-Iribarne is burg, MD; Special publication 813.
are neutralized by negative counter - that in Fenn’s model the activation (24) Gomez, A.; Tang, K. Physics of Fluids,
in press.
ions of t h e solution. Because of energy is thermal, and in the present (25) Fernandez de la Mora, J.; Rossel-
Brownian motion, a given part of the model it is the presence of elastic de- Liompart, J. Proceedings of the 39th ASMS
macromolecule can approach t h e formations t h a t “activates” t h e Conference on Mass Spectrometry and
droplet surface. As this happens one event. Allied Topics; ASMS: Nashville, TN,
1991; p. 441.
or more protonated sites of the mac- (26) (a) Taflin, D. C.; Ward, T. L.; Davis,
romolecule will replace charges on References E. J. Langmuir 1989, 5, 376; (b) Davis,
the surface of the droplet. Thermal E. J . ISA Trans. 1987,26, 1.
activation may provide the energy (1) (a) Kebarle, P. Annu. Rev. Phys. Chemis- (27) Lord Rayleigh Philos. Mag. 1882, 14,
try 1977,28,445; (b) Keesee, R. G.; Cas- 184.
for one of t h e charged sites of the tleman, A. W. J. Phys. Chem. Re5 Data (28) Tang, L.; Kebarle, P., submitted for
macromolecule to move some dis- 1986, 15, 1011; (c) Chandrasekhar, J.; publication in Anal. Chem.
tance ouside the droplet. This pro- Spellmeyer, D. C.; Jorgensen, W. L. J. (29) Davies, C. N. In Fundamentals ofAero-
cess is similar to the activated step Am. Chem. SOC.1984, 106,903. sol Science; Shaw, D. T., Ed.; John Wiley
(2) Desnoyers, J. E.; Joliceur, C. In Mod- and Sons: New York, 1978; p. 154.
of the Iribarne theory involving the ern Aspects of Electrochemistry; Bockris, (30) (a) Iribarne, J. V.; Thomson, B. A. J.
normal ion mechanism. Once such a J.O.M.; Conway, B. E., Eds.; Plenum Chem. Phys. 1976, 64, 2287; (b) Thom-
charge is past the activation barrier, Press: New York 1969; Vol. 5, p. 20. son, B. A.; Iribarne, J. V. J. Chem. Phys.
the separation between it and the (3) (a) Yamashita, M.; Fenn, J. B. Phys. 1979, 71, 4451.
Chem. 1984, 88, 4451; (b) Yamashita, (31) Dole, M.; Mack, L. L.; Hines, R. L.;
charges on the droplets leads to re- M.; Fenn, J. B. Phys. Chem. 1984, 88, Mobley, R. C.; Ferguson, L. D.; Alice,
pulsion, which facilitates the escape 4671. M. B. J Chem. Phys. 1968,49,2240.
of a portion of the molecule that car- (4) Fenn, J. B.; Mann, M.; Meng, C. K.; (32) (a) Rollgen, F. W.; Bramer-Wegner,

ANALYTICAL CHEMISTRY, VOL. 65, NO. 22, NOVEMBER 15,1993 985 A

REPORT

E.; Buttering, L. J. Phys. Colloq. 1984,45, tained by any other means. He is the recip-
Supplement 12, C9-297; (b) Schmelzeis- ient of the Medal of the Canadian
en-Redeker, G.; Buttering, L; Rollgen, Chemical Society (1986) and the ACS
F. W. Int. J. Mass Spectrom. Ion Processes
1989,90, 139. Frank H. Field and Joe L. Franklin
(33) (a) Sakairi, M.; Yergey, A. L.; Siu, Award for Outstanding Achievement in
K.W.M.; LeBlanc, J.C.Y.; Guevremont, Mass Spectrometry (1994).
R., Berman, R. S. Anal. Sci. 1991,7,199;
(b) Rafaelli, A.; Bruins, A. P. Rapid Com- I
I 1__
mun. Mass Spectrom. 1991,5,269.
(34) Hiraoka, K. Rapid Commun. Mass
Spectrom. 1992,6,463.
(35) Smith, R. D.; Wahl, J. H.; Goodlett,
D. R.; Hofstadler, J. A. Anal. Chem. Paul Kebarle received a Dipl. Eng. Chem-
1993,65,574 A.
(36) Rafaeli, A.; Bruins, A. P. Rapid Com- istry (1952) fiom ETH Zurich (Switzer-
mun. Mass Spectrum. 1991,5,269. land) and his Ph.D. in physical chemistry
(37) Landolt, B. Zahlenwerle und Func- (1956)f’rom the University of British Co-
tionen; Springer Verlag: Berlin, 1960;
Vol. 2, pp. 366, 533, 651. lumbia (Canada). After postdoctoral
(38) Tamaki, K. Bull. Chem. SOC. Jpn. 1967, work at NRC in Ottawa, he joined the De-
40, 38. partment of Chemistry at the University of Liang Tang was born in Shanghai (Chi-
(39) Siu, K.W.M.; Guevremont, R.; Le- Alberta. His early research in ion-mole- na) and received both his B.Sc. degree in
Blanc, J.C.Y.; O’Brien, R. T.; Berman, cule reactions at near-atmospheric pres- chemistry and his M.Sc. degree in envi-
S. S. Org. Mass Spectrom. 1993,28, 579.
(40) Guevremont, R.; LeBlanc, J.C.Y.; sures led to the discovery that ion-mole- ronmental engineering at East China
Siu, K.W.M.Org. Mass Spectrom., i n cule equilibria can be determined by MS University of Chemical Technology, Re-
press. techniques. The ion thermochemistry re- cently he received his Ph.D. in analytical
(41) Cook, K. D. Mass Spectrom. Rev. 1986, sulting f’rom equilibrium measurements chemistry at the University of Alberta,
5, 467.
(42) Kelly, M. A.; Vestling, M. M.; Fense- has become a main source of data on the where he worked under the direction of
lau, C. C.; Smith, P. B. Org. Mass Spec- energetics of ions in the gas phase. These Paul Kebarle. His research interests range
trom. 1992,27,1143. data find application in many areas, in- from environmental chemistry to MS,
(43) Ashton, D. S.; Bedell, C. R.; Cooper, cluding ions in solution. His interest in with emphasis on analytical techniques
D. J.; Green, B. N.; Oliver, R.W.A. Org. including chemical ionization, atmo-
Mass Spectrom. 1993,28,721. ESMS was based on the recognition that
(44) Fenn, J. B. J. Am. Soc. Mass Spectrom. it affords the production of many ion spe- spheric pressure ionization, fast atom
1993,4, 524. cies in the gas phase that cannot be ob- bombardment, and electrospray.

Ournew2000puts
electmchemistry to woo& for you.

Whether your application is materials characterization,synthesis,

corrosion, batteries,fuel cells or fast mechanism studies-
take advantage of the new 2000’s power with speed.
It’s the best potentiostaVgalvanostaton the market.
Use the 2000’s built-in techniques or let us customize
a system for you. We’ll work with you until you’re satisfied.
Find out how the new Cypress 2000 and electrochemistry
can work for you. Call us at 800 235-2436. Fax 913 832-0406.

~ESSSSEMS mq
lz
Yowe/ecimchemiicalresource C y S v

CIRCLE 18 ON READER SERVICE CARD

986 A ANALYTICAL CHEMISTRY, VOL. 65, NO. 22,NOVEMBER 15,1993