Interpreting Proton NMR, translated by Robert Herzog

Thanks again to Robert Herzog, a German edition of the Pictorial Guide to Interpreting NMR Spectroscopy is now available.  High-resolution PDF for printing is available here.  Danke Robert!

Edit Thu 19 Jan 3pm: yeesh.  What a mess this is.  I’ve once again been tempted to move to SquareSpace, and once again hesitated with the apparent “soon” release of SqSp 6, and still having another year pre-paid on DreamHost.

The final illustration

There are three distinct steps in this process: a pre-Illustrator step to generate the correct 3D look, a symbol creation step in Illustrator, and the final compositing, which will be discussed in the next post.

Pre-Illustrator

The process begins with a chemical structure.  My software of choice is ChemDoodles for structure (at this writing, I think it’s superior to ChemDraw in everything, and for a fraction the price); you can use anything you want.

Structure editing in ChemDoodles

Save this as a .mol file.  And since you’re reading this workflow, you’re probably a vector user — ChemDoodle has excellent support for vector images.  You can get structures rendered in fancy styles and directly copy-paste into Illustrator; this worked even back at a certain time when OSX’s native vector export was FUBAR’ed.

The wide array of choices ChemDoodle gives you!

We’ll bring this file into Avogadro, an open source molecule viewer. It natively does molecular mechanics structure optimization, and that would work just fine for our simple aniline.  The other options prepare input decks for plugging into commercial (e.g., Gaussian) and open-source (e.g., NWChem) quantum mechanics software.

Molecule optimization in Avogadro

At this point we have a structure on our screen which is of correct geometry.  Select a good angle and export as "vector graphics PDF" (which despite its name, is not a true editable vector in Illustrator!)  Repeat for two different view-points; later on we will use rotate and scale these three views to simulate a chaotic mix of molecules.  I chose the following views.

Views chosen for vector export

Illustrator Symbol Creation

To quickly recap, at this point we have 3 different PDFs for the aniline molecule.  (To keep them organized, I named them "aniline-side", "aniline-in", and "aniline-out" after the position of the functional groups relative to the ring.)  Each of these PDFs contain a black background as well as a separate molecule object (as non-editable construct).  Delete the back black object to get to the molecule (and an axis by default):

The molecule as exported from Avogadro

Note that if your color space is not setup correctly, there will be difference in what is observed in Avogadro and what is seen in Illustrator.  To remove the axis, draw a rectangle over the molecule, select both objects, and use the top as a clipping mask:

I thought it was too dark for what I need, so I reduced the opacity to 75%.  This is entirely optional aesthetics.

Now we come to the ambient music listening time (i.e., mind-numbing bit).  To simulate the drawn appearance, we'll lock (cmd-2) and trace over the figure with the Pen tool (shortcut P).  (I hope you like pulling Bezier curves.)  My recommendation here is to draw closed shapes for each atom, as that allow you the flexibility to control the fills with nice gradients or gradient meshes.  The way I go about generating this is by first drawing overlapping shapes (see left figure), then use the Shapebuilder tool (short-cut shift-M) to trim the edges.  You could use the pathfinder to do these boolean operations if you prefer - but I find the shapebuilder much more speedy for this task.

To generate a true vector, fill these closed shapes with appropriate gradients, using colors sampled from the original.  Adjust the line style and stroke weight to your liking.  Note that you could skip this step and create a hybrid illustration by simply unlocking the original.  In any case, group all the elements together, and you almost have a new symbol!

One note before you're done: unlike normal objects which are knockout by default, Adobe's transparency model place transparent objects as overprints. If you want to avoid partially transparent shapes from overlapping with one another what you could do is to apply a new opaque white fill to the back of the group (see figure for settings).  This takes care of the the overlapping transparent pieces.  (I think there must be a more technically sound way of handling this using blend modes and knockout group options in the transparency panel, but I don't know.)

When you're happy with how your molecule looks, drag it into the symbol tray, and give it a name so you can tell what it is later:

This creates a symbol, which can be introduced in large numbers using the symbol spray tool .

Repeat for other views and molecules as necessary.  The advantage of using symbols, as opposed to straight up manipulation, lies not only in the tools available in Illustrator for spraying and manipulating them, but also in the flexibility of being able to edit any one of the instances and have it propogated to all other instances automatically.

My completed collection of molecules, in charged and uncharged forms, looks like this:

The "charges" are simple circles with a radial gradient and the appropriate parts cut out.  The next part is fun - it leverages all the grunt work to generate pretty pictures.

Illustrator Compositing

With the symbols now created, you can simulate a population of molecules easily using the Symbol Spray tool, then use the Sizer and Spinner to create variations within the sprayed molecules.  For organization's sake, I created three separate layers:

(a) far away layer: scaled down molecules, low opacity (40%).

(b) mid layer: mid-sized molecules, ~75% opacity.

(c) close layer: big molecules, fully opaque.

(If you have tools like Phantasm CS, it might be worthwhile to slightly darken and desaturate the back layers to simulate atmospheric perspective.)

The details of the lower left panel:

I was surprised at how quickly resolving these shapes drag down the computer: while theoretically there are no upper bound to the number of molecules thus sprayed on, my machine was stuttering with more molecules.  This is one reason why I thought there has to be a better way of going about it - there's simply too much grunt work involved for what it is.

The real kicker came after I uploaded it to WikiCommons.  Wikipedia takes SVGs and not PDF/AI files; between a deficient renderer and limited expressiveness of SVG, the files thus generated was displayed incoherently.  (The troubles with transparency reminds me of Illustrator 9's troubles back 10-12 years ago with PostScript and printing.)  At the end I had to rasterize the individual pieces of graphics before packaging it up into a SVG (so that the text remains easily translatable).  While vector shapes are gorgeous (and rational), it's sadly the wrong tool for this particular task.  The better solution will be described in the sister post, How to Draw a Population of Molecules (3D Modeling Route).

Painted from several photo references in Photoshop CS5, took about 90 min end-to-end.  It’s been awhile since I’ve pushed pixels, and the entire process feels quite foreign and unbearably long.  I got pretty impatient with the details that needed to be filled in half way through, so it ended pretty rough and sketchy.  And then, it was not until the end that I realized that, somewhere along the way, I’ve reflected the tower and #$^&% up the light source :(

Edit 8 Jan: This was in part exploring a way of generating generic, modular art, particularly useful for creating cut-scenes in open-source games.  There’s quite a few things that need to be solved, particularly as it relates to perspective, color harmony, usability, and sheer time required to bang out large number of them.  In a way this is like creating high quality, irregular, non-isometric sprites.  One of the explorative attempt was using vector as the source (seeing that it’s my preference), keeping a simpler style.  I didn’t like the output, however, and the search continues.

Looking back at 2011, I was surprised to find that I’ve uploaded >100 illustrations for WikiCommons (even with a thesis to write/defend, and 3 months hiatus in the Andes & Rockies).  Most of these are simple vectorization requests from the Illustration Workshop that’s nothing to write home about, but a few integrates a breadth of information into interesting figures that I ended up being proud of.  Here’s that gallery.

Chloralkali process illustrates how NaOH and chlorine gas is made by electrolysis.  This figure made extensive use of Illustrator’s symbols feature to generate the population of molecules, and the shape-builder in CS5 for simple construction of the irregular container and “solution” from elementary shapes.

Berimbau parts seeks to prepare a vector image with labelled parts.  The benefit of having information in a vector form (particularly, as a SVG) is that the file itself is a plain-text document that can be easily modified with Notepad, making translation possible with little efforts.  For this particular image, the technical challenge was in converting Illustrator’s gradient meshes to something that can be used by Wikipedia’s SVG engine.  The final process involved converting it to a raster image, Live Trace, and manual tweaking.

SSME schematic shows how fuel and oxidizer gets mixed to provide thrust for a space shuttle.  The illustration was mostly constructed with the pen tool, together with the shape-builder.  The outline layer was lifted and duplicated, with the underlayer converted to a live paint group and painted.  The time consuming part was understanding what is actually going on, given the sources were quite confusing and out of my area of expertise.

Thalamus structure illustrates the different parts of the thalamus… which is once again not in my area of expertise and required outside reading.  (Not that I remember very much about it.)  I’m very fond of expansion to show context and multiple scales.  The human image was a high-resolution raster, with pen tool tracing of outlines.

US Air seating chart – This one I’ve written about before.

Sogen Kato home floor-plan describes the discovery of the mummified body of a “100+ yr old” Japanese man, whose family hides the fact that he was dead in order to keep collecting government cheques for over 30 years.  The image was made with Google Sketchup – my first exposure to the tool.  I’d go on to learn about Sketchup in much more detail later in the year.

Wheat nutrition combines information from multiple sources into one coherent image.  The image is not as nice looking as it was executed – wikipedia’s SVG engine does not support either gradient mesh, symbols (as is), or pattern fills.  It also display SVG fonts inconsistently in thumb-nails and final rendering, and between assigning fonts from Inkscape or Illustrator – that’s one that I still haven’t figured out completely.

Overview of Spectrscopy sought to combine the elements involved in spectroscopy in one image, so that the main text can have visuals to refer to.  There’s a few implicit elements in there coming from teaching spectroscopy for a few years – I think experts would get appreciation of this that novices wouldn’t.  The shapes were constructed with the 3D functions in Illustrator, and this prompted the thought of creating the ongoing OpenChemArt project.  This diagram had been translated into several languages.

Reedbed treatment ponds describes different methods for water treatment.  This is one of the first “ambitious” images I tried to make for wikipedia, combining multiple sources of information and representing them in a coherent way.  The single-point perspective was drawn in as is – in CS5 (which I didn’t upgrade to until August) there’s a new perspective tool which might have been useful.

In 2012 I’d like to explore “messier”, painterly vectors.  It’d require some formal trial-and-error to see where the limitations of SVG in Wikipedia arises.  I’d probably also get myself into the raster realm again, since that’s likely where I’ll be when working with the iOS apps I’d like to write on the side.

—–

The to-do list in the upcoming weeks doesn’t involve any other instructional chemistry spreads (sorry).  I’ve a paper that I should really spend time on; there’s several history infographics that I started a few days ago; 3 pages on olfaction; the Pictorial Thesis; the S America trip; history of lipids.  And a narrative comic.  And a poster of balloon-repertoire… argh!  

On the learning side, I’ve been doing lots of anatomy books, and trying to work out a workflow to partially automate the production of a bank of vector graphics.  The goal is to seed a large repository of 10k high-quality vector images pertaining to chemistry which can be easily used by everyone via ChemDraw/ChemDoodles/Powerpoint etc (“Open Chem Art”).  No cigars so far.

Relating to chemistry: I went to my first symposium in Hong Kong today, organized by a govn’ established nanotech company and industry groups.  I marveled at how product-driven the work over this side is (this side of the Pacific, and the academic-industry divide): almost everything (is made to?) seem like it’s about to be incorporated in a commercial product.  The talks themselves flip back and forth between Mandarin, Cantonese, and English, and everyone seems to understand all the languages perfectly.  I enjoyed the symposium thoroughly, and should find out if the local universities open up their seminars to the public.

As usual, clicking on the figure opens a larger version for viewing.  A PDF version for print is available here. I’m not too happy with the final output – I feel as if the spatially-suggested connection between the analogy and the phenomena to be explained is not clear enough.  Suggestions to make this more useful / analogy more clear is appreciated.

Update (17 Dec 2011 Revision R3): The addition of the imagine/now consider label seems to clear up the analogy-phenomena.  I thank Loïc Samuel for the suggestion.

Side-note: a number of people have been looking for this to be published in the last year.  If you’re one of them… I’m sorry for the tardiness!  NMR is the flagship of organic structure identification, revealing much more information than techniques like IR; the corollary is that interpreting NMR is a bigger and more nuanced topic than comparable techniques (on both interpretation and theoretical grounds).  I’ve been going back-and-forth the drawing board to select the right amount of content to include in this overview, and what to reserve for others.  Hopefully this final version is a good compromise.  I encourage and appreciate your comments and questions, and will make updates to the spread with your feedback.

Side-note 2:  Do you speak another language and would like to see this translated?  Contact me and we’ll try to work something out.

Update (R4, 15 Dec 2011):  Updated with corrections.  I thank Shaun MacLean (U of Manitoba) and James Ashenhurst (Mastering Organic Chemistry) for their helpful suggestions.

Click on the image for a larger display, or follow this link for the PDF version.