United States
Environmental Protection
Agency
Environmental Monitoring and
Support Laboratory
Cincinnati OH 45268
Research and Development
Dome
I^EPA Project Summary
EPA-600/S4-84-045 June 1984
EPA Method Study 12, Cyanide
in Water
.Paul Britton, John
A/inter, and Robert C. Kroner
EPA Method Study 12, Cyanide in
Water, reports the results of a study by
EMSL-Cincinnati for the parameters
total cyanide and cyanides amenable to
chlorination which are present in water
at microgram per liter levels. Four
methods, pyridine-pyrazolone, pyridine-
barbituric acid, electrode, and Roberts-
Jackson were used by 60 laboratories in
Federal and State agencies, municipali-
ties, universities, and the private/indus-
trial sector in the method validation
study.
Sample concentrates were prepared
in pairs with similar- concentrations at
each of three levels. Analysts diluted
the samples to volume with distilled and
natural waters for analysis. Precision,
accuracy, bias and I the natural water
interference were evaluated for each
analytical method
were made between
and comparisons
the four methods.
This Project Sumrr, ary was developed
by EPA's Environn
ental Monitoring
and Support Laboratory, Cincinnati,
OH, to announce key findings of the
research project thai is fully document-
ed in a separate report of the same title
(see Project Report
tion at back).
ordering informa-
Introduction
Cyanides are comman constituents in a
variety of industria wastes such as
quench waters from coke plants and
blasts furnaces, and the rinse water from
heat treating and metal finishing opera-
tions. In addition to these metal industries,
other industrial operations, particularly
chemical manufact
jring, utilize and
discharge cyanide conpounds. Becuase
of the extreme toxicit
to aquatic life and tc
treatment of waste1
cyanide is required.
of the cyanide ion
humans, rigorous
waters containing
The measurement of cyanides in the
analytical laboratory is complicated by
the ease with which the cyanide ion
complexes with other metal ions. Com-
plexing occurs in several ways with
metals. For example, cyanide combines
with iron to form ferrous cyanide,
FefCNJa ferric cyanide, Fe(CN)3; ferrous
ferrocyanide FeafFefCNJete ferric ferricya-
nide Fe[Fe(CN)e]; ferrous ferroycyanide,
Fe2[Fe(CN)e] and ferriferrocyanide, Fe[FE
(CN)e]3. Similarly, copper may form a
series of copper-iron-cyanide complexes.
Other metals such as cadmium, silver,
tin, zinc and lead form simple cyanides
(Cd{CN)2, Ag{CN)a, and so on, as well as
the metal-iron-cyanide complexes.
At the normal pH'sandtemperaturesof
streams, the dissociated cyanide ion (CN~)
is toxic to most species of fish at a level of
0.1 mg/liter. Complexed cyanides are far
less toxic and discharge of relatively large
concentrations of these compounds to
receiving streams is not immediately
harmful. Consequently, it has been
suggested that cyanide be complexed
with metals such as iron, prior to
discharge, to reduce toxicity. However,
enforcement authorities are reluctant to
authorize cyanide discharge, even com-'
plex cyanides, because such complexes
may revert to simpler, more toxic forms
under the influence of stream pH,
temperature, and ultra-violet radiation.
Cyanide Measurements
Cyanide is usually measured as the
following parameters: free cyanide,
cyanides amenable to chlorination and
total cyanide.
Free or simple cyanide, such as NaCN,
KCN, or HCN, is directly measurable by
volumetric titration or colorimetry.
Cyanides amenable to chlorination
measures common metal cyanide com-
pounds and most complexes except for
the iron cyanides. The sample is divided
-------
into two parts and total cyanide is
determined before and after an alkaline
chlorination step. Cyanides amenable to
chlorination are the difference between
th two total cyanide analyses. The
method is equivalent to ASTM, Annual
Book of ASTM Standards; 11.02 Water,
D2036-82, Method B-Cyanide Amenable
to Chlorination by Difference, p. 113,
1983.
Total cyanide is a measure of all
cyanides including iron cyanide complexes
after conversion to HCN by acidification,
distillation and absorption in an NaOH
scrubber. The cyanide is titrated as HCN
against a silver nitrate solution or is
converted to cyanogen chloride and read
colorimetncally using a pyridine-pyrazo-
lone or pyridine-barbituric acid reagent.
Analytical Methods for
Cyanide
The methods for the measurement of
cyanide have been a subject of debate
and dissatisfaction among environmental
chemists for many years. Unfortunately,
the complexing properties of the cyanide
radical which make it useful in metal
plating operations are the same properties
which make definition and measurement
difficult. Current analytical methods can
directly measure cyanide-metal complex-
es and cannot uniformly breakdown the
complexes for measurement as simple
cyanide.
The analytical methods attempt to
isolate the cyanide as sodium cyanide,
using a distillation or stripping action,
with absorption of (HCN) as a basic
solution followed by a colorimetric
measurement. The methods of distillation
stripping and the color forming reagents
used vary. Cyanide ions may also be
measured electrometrically using an ion
selective electrode, but despite its speed
and simplicity the method has not been
widely accepted. So called free cyanide,
such as sodium or potassium cyanide, is
also frequently determined volumetrically
by titration with silver nitrate, but the
procedure is not useful for levels of
cyanide below one (1) microgram found in
may wastewaters. Cyanides amenable to
chlorination also employs the usual
colorimetric methods, but measures total
cyanide before and after oxidation of the
sample by chlorination with Chloramine-T.
With the variety of methods available
and the lack of consensus among analysts
regarding the most reliable method, it
was agreed that a collaborative study was
required to ascertain, if possible, which
method should be adopted for general
use. The following methods were subse-
quently chosen for collaborative testing:
1) Serfass distillation/pyridine-barbituric
acid colorimetric method.
2} Serfass distillation/pyridine-pyrazo-
lone colorimetric method.
3} Serfass distillation/ion-selective
electrode.
Participants were requested to analyze
for both total cyanide and cyanides
amenable to chlorination, using Serfass
distillation and their choice of detection
method listed above. However, partici-
pants were encouraged to use the
barbituric acid method if possible.
Upon receiving the invitation to partici-
pate in the study, a number of industrial
labs indicated they would also like to
provide data using the Roberts-Jackson
method of measuring simple cyanides.
This method uses the conventional
colorimetric procedures, but modifies the
distillation so that only simple cyanides
are measured.
Description of Study
Design of Study
The study design is based on Youden's
nonreplicate technique for the collabora-
tive study of analytical methods. Using
this design, sample pairs were developed
with slightly different concentrations of
the constituents, at each of several levels.
The analyst is directed to do a single
analysis and report one value for each
sample, as in normal routine.
In this study, six samples were prepared
as concentrates in sealed glass ampuls
and presented to the analyst as unknowns.
Three levels of cyanide concentration in
three pairs of samples were tested at
levels typical of those observed in
wastewaters.
The analyst was directed to dilute a 5.0-
mL aliquot of each concentrate to one-
liter volume with distilled water and a
second 5.0-mL aliquot to one-liter
volume with a natural or effluent water.
Natural or effluent water samples were
analyzed with and without incremental
aliquots and the recovery determined by
difference. Each sample" was analyzed
only once. Analysis in distilled water
evaluated the proficiency of the analyst in
using the method on a sample free of
interferences, while recovery of the
increment from a natural or effluent
water, such as river, lake, or an estuary,
indicated whether the method was
affected by interferences in these waters.
Data were recorded on standard forms
and returned to EMSL-Cincinnati for
statistical evaluation and preparation of
the report.
Preparation of Samples
Sample concentrates were prepared by
dissolving precisely weighed amounts of
reagent grade chemicals in high purity
water obtained bypassing distilled water
through a four cartridge Millipore Super-
Q System, to produce accurate levels of
simple and complex cyanides. Each
sample contained simple and complex
cyanides, present as potassium cyanide
and potassium ferrocyanide, respectively.
The concentrates were preserved with
sodium hydroxide and checked by repeated
analyses over a period of three months to
confirm the calculated concentrations an
the stability of the samples. Analyses of
the samples by an outside laboratory
confirmed the data of the Quality Assur-
ance Branch, EMSL-Cincinnati.
When diluted to volume according to
the instructions, the samples contained
concentrations of cyanide as shown in
Table 1.
Conduct of the Study
An invitational memorandum announced
the study to the ten EPA Regions and to
the ASTM D-19 committee members in
October, 1974. A separate invitational
letter was sent to industrial laboratories
known to be routinely analyzing wastesfor
cyanides. Although it was estimated to
require a minimum of one workweek of
analytical effort, 112 laboratories from
EPA, other Federal, State and local
agencies, Canadian groups, universities
and private industry agreed to participate.
Each participant received a set of six
ampuls, instructions for sample prepara-
tion, duplicate report sheets, and a copy
of the analytical procedures to be used.
The participating laboratories were
required to analyze samples using
methods from EPA's Methods of Chemical
Table 1. True Values for Cyanide Concentrations*
Sample
Total Cyanide
Uff/L
Cyanide Amenable to
Chlorination, fjg/L
1
2
3
4
5
6
25
372
35
106
106
352
13
149
18
64
64
141
*The concentrations were the actual levels calculated and added. Analyses were used for
verification only.
-------
Analysis of Water and Wastes, 1974,
EMSL-Cincinnati. However, a number of
industrial laboratories participating in the
study asked if it would be agreeable with
EPA if they performed the cyanide
analyses by a modified Roberts-Jackson
method as well as by the pyridine
methods. EPA agreed. Appendix B provides
description of EPA methodologies, the
original Roberts-Jackson paper, and the
Woods River modification (Shell Oil
Company) which the industrial laboratories
used in this study. Participants were
allowed 50 days to complete the analyses
and report the data. Data reported later
than the cut-off date were omitted. Fifty-
six laboratories returned data in time to
be included in this report.
Summary for C\ATC in Natural
Waters I _
The mean recbvery (X), overall standard
deviation (S), and single-analyst standard
deviation (Sr) results for natural water
analyses by each of the methods, within
the concentration rehge 13-150//g/liter,
and pyridine-pyrazolone methods were
very similar. The electrode method
showed significantly greater data varia-
bility. However, all three methods seem
capable of producing valid data in the
hands of a skilled analyst.
are as follows.
Method
P-B*
P-P**
Electrode
1.19
1.61
1.35
X
(cone.) -
(cone.) -
(cone.) -
10.0
15.0
11.9
0
0
0
540
686
852
S
X +
X-
x +
15.5
0.5
33.2
0.232
0.225
0.352
Sr
X +
X-
X +
16.1
4.0
22.1
Results and Discussion
Basic statistical results were computed
for each combination of sample method
and water matrix using the Collaborative
Study (COLST) computer system developed
by the USEPA. Final relationships for
recovery from the natural water matrix
are given below.
Summary for Total Cyanide in
Natural Waters
The mean recoveries (X), overall
standard deviations (S), and single
analyst standard deviations (Sr) for
natural water analyses by each of the
methods within the concentration range
25-400 jug/liter, are as follows.
At the extremes of the applicable
range, these equations lead to the
following statistical estimate in //g/liter.
* ' At 13 /ig/liter"
At"150 Tig/liter
Method
X
Sr
X
Sr
P-B* 5.47 22.5
19.1
P-P** 5.93 9.4 -1.1a
Electrode $.65 44.3
26.7
168.5
226.5
190.6
96.5
103.4
161.0
50.9
29.8
74.9
* Pyridine-barbitunic acid
** Pyridine-pyrazolone
a The variance-estimating
concentration.
equations for this method are not valid at this low
Although these tables show slightly
better statistics forthe pyridine-barbituric
acid method, the cyanides amenable to
chlorination statistics for all methods look
very unsatisfactory. This suggests an
inherent problem within the cyanides
Method
P-B*
P_P**
Electrode
X
0.916 (cone.)
0.965 (cone.)
1.00 (cone.)
- 2.0
- 3.6
-0.8
S
0.259 X+ 9.0
0.107 X + 16.3
0.213 X + 40.6
Sr
0.104 X+ 9.2
0.018 X + 12.3
0.246 X + 0.2
At the extremes of the applicable
range, these equations lead to the
following statistical estimates inyt/g/liter.
Method
P-B*
P_P**
Electrode
X
20.
20.
24.
25
9
5
2
yug/liter
S
15.5
19.0
45.9
Level
S
11
14
6
r
8
3
.3
40Qyug/liter
X
364.4
382.4
399.2
S
Level
112.6
59.1
125.8
S
50
44
98
r
.8
.7
6
* Pyridine-barbituric Acid
** Pyridine-pyrazolone
These tables make the larger variability
of the electrode method obvious and,
although fortuitous averaging makes its
mean recovery look better, recall that its
mean recoveries were much more
variable. Between the colorimetric meth-
ods, the apparent statistical advantage of
the pyridine-pyrazolone method is also
quite clear.
amenable to chldrination definition
rather than a problem with measurement
technique applied.
Conclusions
Total Cyanide
For the total
statistics for the py
cyanide parameter,
•idine-barbituric acid
Cyanides Amenable to
Chlorination
Although the pyridine-barbituric acid
method showed the least bias and the
smallest standard deviation, none of the
three methods tested provided satisfac-
tory data for the cyanides amenable to
chlorination parameter. As a corollary,
cyanides amenable to chlorination did not
provide a reliable means for distinguishing
between free and complexed cyanides.
Roberts-Jackson
The generation of data by the Roberts-
Jackson method was encouraged, but-
only three labs submitted results and one
of these sets was incomplete.
On the basis of the limited data
obtained in this study, the Roberts-
Jackson method shows promise of
improved accuracy and precision, while
providing greater safety by avoiding the
open generation of toxic cyanogen
chloride. However, because of limited
data this study did not establish its real
value.
General Conclusion
The compelling reason for preferring the
pyridine-barbituric acid method over the
pyridine-pyrazolone method is conveni-
ence rather than statistical improvement
in the data produced.
-------
The EPA authors John Winter (EPA contact, see below), PaulBritton, and Robert
Kroner are with the Environmental Monitoring and Support Laboratory,
Cincinnati, OH 45268.
The complete report, entitled "EPA Method Study 12, Cyanide in Water," (Order
No. PB 84-196 674; Cost: $ 13.00, subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
John Winter can be contacted at:
Environmental Monitoring and Support Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
'•ff U.S. GOVERNMENT PRINTING OFFICE; 1984—759-015/7736
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use $300
EMSL0159000
MARY SULLIVAN
BRANCH
CINCINNATI*™ 45268
------- |