(551c) Treatment of Reverse Osmosis Concentrate to Improve Overall Recovery: The Effect of Prior Antiscalant Oxidation on Particle Characteristics and the Extent of Precipitation

In many
locations, fresh water resources are insufficient for local needs, and
alternative sources with lesser water quality are being considered as drinking
water supplies. In particular, the United States has many inland regions with
untapped brackish water (500?10,000 mg/L total dissolved solids (TDS))
resources.  Interest has increased for using these brackish water resources to
produce drinking water.  The two primary process choices for water desalination
are thermal distillation or membrane filtration.  Reverse osmosis (RO)
desalination has become the primary choice for brackish water desalination due
to lower energy requirements and a smaller footprint.  However, from brackish
water, the product recovery (volume of product water per volume of feed water)
range is only 75 ? 90%; i.e., at least 10% of the feed water becomes the
RO waste stream, or concentrate. The costs and technical feasibility of
concentrate disposal are the key limitations to widespread application of
inland RO.  This research focuses on the development of a novel process to
reduce the volume of brackish water RO concentrate.

Several critical
differences between brackish and sea waters are the relative metal ion
concentrations and the typical RO recovery ranges.  While a typical brackish
water (1,500 mg/L TDS) has a chloride ion concentration that is 30 times
smaller than sea water and a sodium ion concentration that is 90 times smaller,
the calcium concentration is only 2 ? 5 times smaller and the carbonate
concentration is equal to or larger than that of sea water.  In addition,
brackish water RO recovery is higher than sea water RO recovery (40-60%). 
These factors cause the recovery in brackish water RO systems to be limited by
salt precipitation. 

Chemicals
called antiscalants are often used to complex with problematic salts (CaCO₃,
CaSO₄, BaSO₄, SrSO₄, silica), delaying
precipitation. However, salt concentration increases with recovery, and
antiscalants work successfully within a limited concentration range. 
Therefore, eventually precipitation control is overcome. To increase system
recovery and decrease the concentrate volume, a new approach is required.

Previous
research using precipitation and separation to treat concentrate has shown that
significant increases in total system recovery are possible [1]. However, the presence and influence of antiscalants and natural organic matter (NOM) during RO concentrate treatment have not been investigated.

This paper
presents the development of a novel three-stage process to treat the
concentrate from a brackish water RO system. The process achieves problematic
salt removal through (I) antiscalant deactivation, (II) precipitation, and
(III) solid/liquid separation. Antiscalant deactivation is performed using
ozone (O₃) and hydrogen peroxide (H₂O₂).  pH
elevation is used to precipitate salts, and solid/liquid separation is achieved
through sedimentation and filtration. While technologies for solid/liquid
separation are well-established, the combination of antiscalant oxidation and
precipitation represents a new system; research on antiscalant oxidation has
been limited [2], and the effect of ozonation on precipitation has not been investigated.

Specifically,
this study focused on changes to the precipitate particles and the extent of
precipitation caused by prior antiscalant oxidation.  Experiments were
performed on a series of increasingly complex waters.  First a simplified
concentrate (~ 8,000 mg/L TDS), containing only sodium chloride (NaCl), sodium
bicarbonate (NaHCO₃), and calcium chloride (CaCl₂*2H₂O)
was used.  Subsequently, magnesium and sulfate were added, and finally, real
water samples were tested.  Four different antiscalants, including several
phosphonates and one acrylic polymer blend were used during experiments.

Several
oxidation parameters, such as ozonation time and antiscalant concentration,
were varied.  Ozonation times of 1, 10, and 30 minutes were tested.  For the
first antiscalant tested (amino tri(methylene phosphonic acid), or AMPA) and
the simplified concentrate, results show increased calcium precipitation for
all ozonation times and all antiscalant concentrations (4 ? 85 mg/L).  The
simplified concentrate with 85 mg/L AMPA, treated with only 1 minute of
ozonation and subsequent precipitation at pH 10.5 for 1 hour, had a final
dissolved calcium concentration of 5.2 mg/L Ca²⁺.  In comparison,
the same precipitated solution with no antiscalant resulted in a final dissolved
calcium concentration of 3.7 mg/L, while the same precipitated solution with 85
mg/L AMPA and no prior ozonation resulted in a final dissolved calcium
concentration of 77 mg/L.

A laser granulometer Mastersizer S (Malvern Instruments) and a laser
particle counter (Met One) were used to evaluate the effect of the ozonation
step on precipitate particle size and number.  Previous results have
shown that certain antiscalants, such as AMPA (40 - 85 mg/L), can change the
particle size range and the modality of the size distribution for calcium
carbonate (CaCO₃) precipitation.  A solution of precipitated
simplified concentrate with 40 ? 85 mg/L AMPA is bimodal and has particle size
ranges of 0.2 ? 2 microns and 2 - 50 microns, whereas the same solution without
antiscalant has a single particle size range between 10 and 100 microns.  Results
from combined ozonation-precipitation experiments show that ozonation causes
the particle size distribution to shift towards that of a solution containing
no antiscalant.  Light microscope photos (25x) are consistent with the
granulometer results and show particles that resemble a precipitated solution
with no antiscalant present.

The fouling
potential of precipitated solutions was evaluated through dead end filtration
experiments (MWCO = 0.1 micrometers, Millipore nitrocellulose membranes).  For
antiscalant AMPA (85 mg/L), ozonation times of 1 and 10 minutes increase the
permeate flux.  An ozonation time of 30 minutes causes a higher initial
permeate flux but the flux decline is greater overall and results in a lower
final permeate flux.  These differences in flux decline indicate different
membrane fouling mechanisms.  Moreover, this ozonation duration is not in
agreement with an industrial development.

Future work
includes manipulation of flux data to determine the type of membrane fouling
occurring and continuing experiments with more complex waters. 

[1] Rahardianto, A.; Gao, J.; Gabelich, C.J.; Williams, M.D.; Cohen, Y., High recovery
membrane desalting of low-salinity brackish water: Integration of accelerated
precipitation softening with membrane RO. Journal of Membrane Science 2007,
289, 123-137.

[2] Yang, Q.; Ma, Z.; Hasson, D.;
Semiat, R., Destruction of Anti-Scalants in RO Concentrates by Electrochemical
Oxidation. Journal of Chemical Industry and Engineering (China) 2004,
55(2), 339-340.