In science and engineering, the partsper notation is a set of pseudounits to describe small values of miscellaneous dimensionless quantities, e.g. mole fraction or mass fraction. Since these fractions are quantityperquantity measures, they are pure numbers with no associated units of measurement. Commonly used are ppm (partspermillion, ), ppb (partsperbillion, ), ppt (partspertrillion, ) and ppq (partsperquadrillion, ). This notation is not part of the SI system and its meaning is not unambiguous.
Partsper notation is often used describing dilute solutions in chemistry, for instance, the relative abundance of dissolved minerals or pollutants in water. The unit "1 ppm" can be used for a mass fraction if a waterborne pollutant is present at onemillionth of a gram per gram of sample solution. When working with aqueous solutions, it is common to assume that the density of water is 1.00 g/mL. Therefore, it is common to equate 1 kilogram of water with 1 L of water. Consequently, 1 ppm corresponds to 1 mg/L and 1 ppb corresponds to 1 ?g/L.
Similarly, partsper notation is used also in physics and engineering to express the value of various proportional phenomena. For instance, a special metal alloy might expand 1.2 micrometers per meter of length for every degree Celsius and this would be expressed as "? = 1.2 ppm/°C." Partsper notation is also employed to denote the change, stability, or uncertainty in measurements. For instance, the accuracy of landsurvey distance measurements when using a laser rangefinder might be 1 millimeter per kilometer of distance; this could be expressed as "Accuracy = 1 ppm."^{[1]}
Partsper notations are all dimensionless quantities: in mathematical expressions, the units of measurement always cancel. In fractions like "2 nanometers per meter" (2 nm/m = 2 nano = 2 × 10^{9} = 2 ppb = 2 × 0.000000001) so the quotients are purenumber coefficients with positive values less than 1. When partsper notations, including the percent symbol (%), are used in regular prose (as opposed to mathematical expressions), they are still purenumber dimensionless quantities. However, they generally take the literal "parts per" meaning of a comparative ratio (e.g., "2 ppb" would generally be interpreted as "two parts in a billion parts").^{[2]}
Partsper notations may be expressed in terms of any unit of the same measure. For instance, the coefficient of thermal expansion of a certain brass alloy, ? = 18.7 ppm/°C, may be expressed as 18.7 (µm/m)/°C, or as 18.7 (µin/in)/°C; the numeric value representing a relative proportion does not change with the adoption of a different unit of measure.^{[3]} Similarly, a metering pump that injects a trace chemical into the main process line at the proportional flow rate Q_{p} = 125 ppm, is doing so at a rate that may be expressed in a variety of volumetric units, including 125 µL/L, 125 µgal/gal, 125 cm^{3}/m^{3}, etc.
In nuclear magnetic resonance spectroscopy (NMR), chemical shift is usually expressed in ppm. It represents the difference of a measured frequency in parts per million from the reference frequency. The reference frequency depends on the instrument's magnetic field and the element being measured. It is usually expressed in MHz. Typical chemical shifts are rarely more than a few hundred Hz from the reference frequency, so chemical shifts are conveniently expressed in ppm (Hz/MHz). Partsper notation gives a dimensionless quantity that does not depend on the instrument's field strength.
Although the International Bureau of Weights and Measures (an international standards organization known also by its Frenchlanguage initials BIPM) recognizes the use of partsper notation, it is not formally part of the International System of Units (SI).^{[2]} Note that although "percent" (%) is not formally part of the SI, both the BIPM and the ISO take the position that "in mathematical expressions, the internationally recognized symbol % (percent) may be used with the SI to represent the number 0.01" for dimensionless quantities.^{[2]}^{[5]} According to IUPAP, "a continued source of annoyance to unit purists has been the continued use of percent, ppm, ppb, and ppt."^{[6]} Although SIcompliant expressions should be used as an alternative, the partsper notation remains nevertheless widely used in technical disciplines. The main problems with the partsper notation are the following:
Because the named numbers starting with a "billion" have different values in different countries, the BIPM suggests avoiding the use of "ppb" and "ppt" to prevent misunderstanding. In the English language, named numbers have a consistent meaning only up to "million". Starting with "billion", there are two numbering conventions: the "long" and "short" scales, and "billion" can mean either 10^{12} or 10^{9}. The U.S. National Institute of Standards and Technology (NIST) takes the stringent position, stating that "the languagedependent terms [ . . . ] are not acceptable for use with the SI to express the values of quantities."^{[7]}
Although "ppt" usually means "parts per trillion", it occasionally means "parts per thousand". Unless the meaning of "ppt" is defined explicitly, it has to be guessed from the context.
Another problem of the partsper notation is that it may refer to mass fraction, mole fraction or volume fraction. Since it is usually not stated which quantity is used, it is better to write the unit as kg/kg, mol/mol or m^{3}/m^{3} (even though they are all dimensionless).^{[8]} The difference is quite significant when dealing with gases and it is very important to specify which quantity is being used. For example, the conversion factor between a mass fraction of 1 ppb and a mole fraction of 1 ppb is about 4.7 for the greenhouse gas CFC11 in air. For volume fraction, the suffix "V" or "v" is sometimes appended to the partsper notation (e.g., ppmV, ppbv, pptv).^{[9]}^{[10]} Unfortunately, ppbv and pptv are also often used for mole fractions (which is identical to volume fraction only for ideal gases).
The usage is generally quite fixed inside most specific branches of science, leading some researchers to draw the conclusion that their own usage (mass/mass, mol/mol, volume/volume, or others) is the only correct one. This, in turn, leads them to not specify their usage in their publications, and others may therefore misinterpret their results. For example, electrochemists often use volume/volume, while chemical engineers may use mass/mass as well as volume/volume. Many academic papers of otherwise excellent level fail to specify their usage of the partsper notation.
SIcompliant units that can be used as alternatives are shown in the chart below. Expressions that the BIPM explicitly does not recognize as being suitable for denoting dimensionless quantities with the SI are shown in red text.
Measure  SI units 
Named partsper ratio (short scale) 
Partsper abbreviation or symbol 
Value in scientific notation 

A strain of...  2 cm/m  2 parts per hundred  2%^{[11]}  2 × 10^{2} 
A sensitivity of...  2 mV/V  2 parts per thousand  2 ?  2 × 10^{3} 
A sensitivity of...  0.2 mV/V  2 parts per ten thousand  2 ?  2 × 10^{4} 
A sensitivity of...  2 µV/V  2 parts per million  2 ppm  2 × 10^{6} 
A sensitivity of...  2 nV/V  2 parts per billion  2 ppb  2 × 10^{9} 
A sensitivity of...  2 pV/V  2 parts per trillion  2 ppt  2 × 10^{12} 
A mass fraction of...  2 mg/kg  2 parts per million  2 ppm  2 × 10^{6} 
A mass fraction of...  2 µg/kg  2 parts per billion  2 ppb  2 × 10^{9} 
A mass fraction of...  2 ng/kg  2 parts per trillion  2 ppt  2 × 10^{12} 
A mass fraction of...  2 pg/kg  2 parts per quadrillion  2 ppq  2 × 10^{15} 
A volume fraction of...  5.2 µL/L  5.2 parts per million  5.2 ppm  5.2 × 10^{6} 
A mole fraction of...  5.24 µmol/mol  5.24 parts per million  5.24 ppm  5.24 × 10^{6} 
A mole fraction of...  5.24 nmol/mol  5.24 parts per billion  5.24 ppb  5.24 × 10^{9} 
A mole fraction of...  5.24 pmol/mol  5.24 parts per trillion  5.24 ppt  5.24 × 10^{12} 
A stability of...  1 (µA/A)/min.  1 part per million per min.  1 ppm/min.  1 × 10^{6}/min. 
A change of...  5 n?/?  5 parts per billion  5 ppb  5 × 10^{9} 
An uncertainty of...  9 µg/kg  9 parts per billion  9 ppb  9 × 10^{9} 
A shift of...  1 nm/m  1 part per billion  1 ppb  1 × 10^{9} 
A strain of...  1 µm/m  1 part per million  1 ppm  1 × 10^{6} 
A temperature coefficient of...  0.3 (µHz/Hz)/°C  0.3 part per million per °C  0.3 ppm/°C  0.3 × 10^{6}/°C 
A frequency change of...  0.35 × 10^{9} ?  0.35 part per billion  0.35 ppb  0.35 × 10^{9} 
Note that the notations in the "SI units" column above are all dimensionless quantities; that is, the units of measurement factor out in expressions like "1 nm/m" (1 nm/m = 1 nano = 1 × 10^{9}) so the quotients are purenumber coefficients with values less than 1.
Because of the cumbersome nature of expressing certain dimensionless quantities per SI guidelines, the International Union of Pure and Applied Physics (IUPAP) in 1999 proposed the adoption of the special name "uno" (symbol: U) to represent the number 1 in dimensionless quantities.^{[6]} This symbol is not to be confused with the alwaysitalicized symbol for the variable "uncertainty" (symbol: U). This unit name "uno" and its symbol could be used in combination with the SI prefixes to express the values of dimensionless quantities that are much lessor even greaterthan one.^{[12]}
Common partsper notations in terms of the uno are given in the table below.
Coefficient  Partsper example  Uno equiv.  Symbol form  Value of quantity 

10^{2}  2%  2 centiuno  2 cU  2 × 10^{2} 
10^{3}  2 ?  2 milliuno  2 mU  2 × 10^{3} 
10^{4}  2 ?  0.2 milliuno  0.2 mU  2 × 10^{4} 
10^{6}  2 ppm  2 microuno  2 µU  2 × 10^{6} 
10^{9}  2 ppb  2 nanouno  2 nU  2 × 10^{9} 
10^{12}  2 ppt  2 picouno  2 pU  2 × 10^{12} 
In 2004, a report to the International Committee for Weights and Measures (known also by its Frenchlanguage initials CIPM) stated that response to the proposal of the uno "had been almost entirely negative" and the principal proponent "recommended dropping the idea".^{[13]} To date, the uno has not been adopted by any standards organization and it appears unlikely it will ever become an officially sanctioned way to express lowvalue (highratio) dimensionless quantities. The proposal was instructive, however, as to the perceived shortcomings of the current options for denoting dimensionless quantities.
Partsper notation may properly be used only to express true dimensionless quantities; that is, the units of measurement must cancel in expressions like "1 mg/kg" so that the quotients are pure numbers with values less than 1. Mixedunit quantities such as "a radon concentration of 15 pCi/L" are not dimensionless quantities and may not be expressed using any form of partsper notation, such as "15 ppt". Other examples of measures that are not dimensionless quantities are as follows:
Note however, that it is not uncommon to express aqueous concentrationsparticularly in drinkingwater reports intended for the general publicusing partsper notation (2.1 ppm, 0.8 ppb, etc.) and further, for those reports to state that the notations denote milligrams per liter or micrograms per liter. Although "2.1 mg/L" is not a dimensionless quantity, it is assumed in scientific circles that "2.1 mg/kg" (2.1 ppm) is the true measure because one liter of water has a mass of about one kilogram. The goal in all technical writing (including drinkingwater reports for the general public) is to clearly communicate to the intended audience with minimal confusion. Drinking water is intuitively a volumetric quantity in the public's mind so measures of contamination expressed on a perliter basis are considered to be easier to grasp. Still, it is technically possible, for example, to "dissolve" more than one liter of a very hydrophilic chemical in 1 liter of water; partsper notation would be confusing when describing its solubility in water (greater than a million parts per million), so one would simply state the volume (or mass) that will dissolve into a liter, instead.
When reporting airborne rather than waterborne densities, a slightly different convention is used since air is approximately 1000 times less dense than water. In water, 1 µg/m^{3} is roughly equivalent to partspertrillion whereas in air, it is roughly equivalent to partsperbillion. Note also, that in the case of air, this convention is much less accurate. Whereas one liter of water is almost exactly 1 kg, one cubic meter of air is often taken as 1.143 kgmuch less accurate, but still close enough for many practical uses.^{[]}
