fix links and double brackets

docs/joss/paper.md

@@ -37,7 +37,7 @@ Accurately predicting the thermodynamic and transport properties of complex elec
 solutions containing many solutes, especially at moderate to high salt concentrations
 commonly encountered in water desalination and resource recovery applications, remains a
 major scientific challenge [@rowland_ProgressAqueousSolution_2019]. This challenge is
-compounded by the fact that the best available models (such as the Pitzer model @May2011b)
+compounded by the fact that the best available models, such as the Pitzer model [@May2011b],
 are difficult to implement on an as-needed basis and require looking up many parameters.
 Researchers and practitioners in fields such as water treatment and desalination, electrochemistry,
 or environmental engineering need accurate information about electrolyte solutions to perform their work,
@@ -49,7 +49,7 @@ specialists. However, they are not highly accessible for routine use by others,
 steep learning curve, difficult interoperability with other tools (such as external transport models),
 the lack of a freely-available version, and/or limitation to specific operating systems.
 Several `python` interfaces to the open-source `PHREEQC` software exist, including IPhreeqC [@Parkhurst2013],
-`phreeqpython` (https://github.com/Vitens/phreeqpython), and `pyeqion2` [@marcellos2021pyequion]. However, these interfaces
+`phreeqpython` ([https://github.com/Vitens/phreeqpython](https://github.com/Vitens/phreeqpython)), and `pyeqion2` [@marcellos2021pyequion]. However, these interfaces
 are either not object-oriented, poorly documented, and/or only offer access to only a limited subset of the `PHREEQC`
 parameter databases. There are more subtle limitations as well. For example, `phreeqpython` is unable to calculate
 solution conductivity when used in conjunction with the `PHREEQC` `pitzer.dat` database (the most accurate for high
@@ -60,8 +60,8 @@ to piece together outputs from disparate models and literature -- a time-consumi
 `pyEQL` is designed to free researchers from the tedium of identifying and implementing the relevant models
 and compiling the required parameters from literature. It defines a python `Solution` class from which properties
 can be easily retrieved. It implements the Pitzer model [@May2011b] for binary salts, with mixing rules [@Mistry2013]
-for more complex solutions, and decays gracefully to more approximate models (such as the Debye-Huckel activity
-model @Stumm1993) when adequate data is not available. The built-in property database includes Pitzer model
+for more complex solutions, and decays gracefully to more approximate models like the Debye-Huckel activity
+model [@Stumm1993] when adequate data is not available. The built-in property database includes Pitzer model
 parameters [@May2011b] for more than 100 salts, diffusion coefficients [@CRCdiffusion] for more than 100 solutes,
 and an ever-expanding set of additional property data that make the best-available models transparently accessible
 to the end user.
@@ -97,7 +97,7 @@ model parameters are missing from the database.
 ## Interoperate with other scientific codes
 
 `pyEQL` is built to be extensible, customizable, and easy to use in conjunction with widely-used scientific `python`
-libraries. Specifically, it makes use of `pint` (https://github.com/hgrecco/pint) to provide automatic unit conversions and leverages codes in
+libraries. Specifically, it makes use of `pint` ([https://github.com/hgrecco/pint](https://github.com/hgrecco/pint)) to provide automatic unit conversions and leverages codes in
 the Materials Project [@Jain2013] ecosystem -- namely, `pymatgen` [@Ong2013] for chemical informatics (e.g., molecular
 weight, parsing chemical formulae) and `maggma` [@maggma] for accessing the built-in property database.